» TANDON: THE HEAD DRIVE
BEHIND COMPUTER WORLD
» COMPUTER CAUSES A SEACHANGE/
IN CUSTOM HOUSE /

» CENTRONICS BUFFER /

| VHFAJHF WIDE BAND AMPLIFIERS /

> RECORDING PLAYBACK AMPLIFIER /

» BATTERY LOW' INDICATOR

1 4.05

Front cover

The ghostly figure,
known as the Head
and Torso Simulator
(HATS), seen here be-
ing fitted with a tele-
phone handset by
technician Helen
Christian, is used in
tests at British
Telecom's Research
laboratories to
measure sound
pressure.

The tests take place in
an anechoic chamber.
An artificial ear on the
simulator holds a
miniature microphone
to monitor the
loudness and fre-
quency response of
the telephone.

BUDGET INDICATES
NEGATIVE THINKING

Electronics industry in the country has little to cheer about In the latest budgetary
proposals of the Union government. The ethos of modernisation and an all out
support to computerisation, coupled with liberal policies which marked the scene a
couple of years ago has now disappeared.

Contrary to the otf-proclaimed goal of bringing down the prices of television sets, an
increase on excise duty on the picture tubes adds to the cost. That people having
too much of an ■entertainment’ would not mind paying a little more is at best a cruel
joke.

What if the 5 per cent hike in excise duty on computers had not been proposed, one
may ask. Can't the government find that couple of crores of rupees from elsewhere,
ask another. What if the duties were reduced further to promote the market, queries
a third person. The answer appears to be that the government's thinking has
undergone a change. The electronics industry in general and the computer industry
In particular do not seem to find the pride of place as they used to be in the past.
Industry sources rightly observe this trend as "negative" in nature.

However, it is reassuring to hear from the industry that despite these new imposts and
consequent marginal hike In prices of end products the market for television sets
and computerss may not fall.

Added to the fiscal problems is the lack of a definite direction In which the Industry
Is to go. The lukeworm attitude of the government towards matters electronics has
left the industry in chaotic condition with no clear set goals. It Is high time the
government machinery woke up to the need for strengthening itself to evolve
suitable policies and perspectives.

POWER LINE MODEM

The NE5050 from Philips Components has been designed for
sending and receiving data over the AC mains network, coaxial
cables or twisted-pair cables. The modem described here is a
mains-based application of the NE5050. It works in conjunction
with an error-correcting computer program for exchanging data
or remote control of equipment.

by J. Bareford

A modem (acronym for MOdu-
lator/DEModulator) is almost in-
variably used where the distance between
computers, or a computer and
peripheral equipment, exceeds the
capabilities of the well-known RS-232
interface with associated cables. In prac-
tice, this means that some sort of
modem is necessary when the data rate
and distance exceed 1200 baud and
about 30 metres respectively. In most
cases, the modem is located physically
close to the computer or peripheral
(sometimes it is internal to it). Modems
generally use frequency-shift-keying
(FSK) of a carrier to convert the logic
levels received from the computer's RS-
232 outlet into tones that can be carried
over, say, the telephone network. In re-

ceive mode, the tones from the modem
at the other end of the line are
demodulated and converted to RS-232
levels for sending to the computer.

The present modem does not use FSK,
but ASK (amplitude shift keying) for
reasons discussed below. Similar to cer-
tain types of intercom, the NE5050-
based modem is connected to the remote
station via the mains network.
Background to amplitude shift
keying

The mains network is by no means ideal
for data communication. Impulse noise,
voltage dips, line impedance modulation
and high-frequency signals are but a few
of the sources of interference to be taken
into account. Improperly decoupled

fluorescent tubes, dimmers, refrigerators
and washing machines are notorious for
the high levels of ‘mains pollution’ they
cause.

Clearly, the design of a practical mains
modem should anticipate high levels of
interference and possible corruption of
data owing to the above appliances.

In radio technology, it has been known
for almost 100 years that CW (con-
tinuous wave or modulation type Al), or
simply switching the transmitter on and
off, is the simplest,, yet most
interference-resistent, modulation
method available. Figure 1 shows how
CW is used by the present modem — a
120 kHz carrier is generated and digital
input data determines when the carrier is
to be superimposed on to the mains
lines. Collision, or more precisely summ-
ing of data, however, occurs when two
modems connected to the network

transmit simultaneously. Thanks to the
use of ASK, this only leads to distortion
of data, not to overloading of the
modem input. By setting up an error-
detecting data exchange protocol in the
computer, messages between modems
can be repeated until they are correctly
received. The use of a communications
program on the computer for combat-
ting data collision and interference
simplifies the modem hardware con-
siderably, and at the same time makes it
virtually computer-independent.

An integrated modem

Apart from the electrical connection and
the component values, the circuit
diagram of Fig. 2 shows the internal
structure of the central part, the NE5050
in position ICi.

The transmitter in the modem chip is
composed of a carrier oscillator, a TTL
buffer/input amplifier, and a line driver
that also functions as the amplitude-
modulator. External components Cs,
Cio and L3 tune the oscillator to
120 kHz. Capacitor Ch does not form

part of the tuned circuit, but serves to
decouple the internally generated supply
voltage of '/iUb which is used for bias-
ing the oscillator. The generated carrier
is applied to the line driver in which am-
plitude modulation takes place. The car-
rier is modulated by the data signal ap-
plied to pin 19 of the chip. Together with
Ti, T’, R7, R« and R«, the driver forms a
class-AB output stage that gives the ASK
signal enough power to be superimposed
on to the mains lines. For reasons of
safety, this is done with the aid of a
double-insulated line transformer with a
turns ratio Lia:Lib:Lic= 1:4:1. A number
of components with specific functions
are arranged around this transformer.
C12 and Rio ensure a sufficiently high
termination impedance for the line
driver. Ci suppresses the mains fre-
quency (50 or 60 Hz), and Di and D2
have the double function of transient
suppressor and limiter for the received
120 kHz signal. Under no conditions
should the indicated diodes be replaced
by common zener diodes which these are
far too slow in this application, and,

therefore, unable to protect the mains
modem chip from damage by voltage
surges.

The input of the modem, pin 20, nor-
mally receives not only the signals from

Fig. 3. Simple extension of the modem in-
terface to enable connection to an RS-232

Fig. 4 . Prinled-circtiil board for (he mains modem.

other modems, but also its own trans-
mitted signal. In the present application,
the receiver is, however, disabled while
the modem is in transmit mode. This is
achieved by having the transmit input
drive Ti. When this is turned on, it
pulls the comparator output, pin 10,
low, so that the bistable can not change
state. When the data input line is low, no
carrier is transmitted.

The received signal is first applied to an
amplifier provided with a band-pass
characteristic. The high-frequency roll-
off point is internally set to 300 kHz.
Dimensioning C-i allows defining the
lower roll-off point in accordance wit the
carrier frequency used. To ensure selec-

tivity at the carrier frequency, a band-
filter, L2-C5, is inserted between the in-
put amplifier and the detector. C+ and

Resistors ( ± 5%l:

Ri;R2 = 5K6

R3;Ri3=1M0

Ri = 10M

Rb = 220K

Re=10K

R7;R9=1R0

Ra;Ri2 = 22K

Rio=10R

RH-47K

Ru-IKO

Capacitors:

Ci =470n; 630 V
C2 = 6n8

C3;C7;Cn ■ lOOn
C4;CB;Ca;Cl0-4n7
Cs = 10n
C9*27p
Cl2=1p0

Ci3= lOOOp: 40 V; radial
Ct4= IpO; 16 V: radial

Semiconductors:

Di;D2 = BZT03C15 + (Philips Components)
D3= 5V6; 400 mW zener diode
D4...D7 incl. = 1N4001
Da= red LED
T 1 = BDX77 +

T2 = BDX78*

T3;T4 = BC547B

ICi = NE5050 + (Philips Components)

IC2 = 7812

* Listed by Universal Semiconductor Devices

Li = VAC 2KB 490/255 (VAC
Vacuumschmelze GmbH e Werk Hanau •
Gruner Weg 37 e 6450 Hanau 1 a West-
Germany. Tel. +49 6181 362-1; telex
4184863; fax +49 6181 362645).

L2;L3=390pH

St = double-pole on/off switch.

Fi = 250 mA delayed action fuse with PCB-
mount holder.

Tri= PCB mount transformer; 3 VA; 2x7.5 V
8200 mA.

Kt = 3-way PCB-mount terminal block.

Heat sink for IC2.

Moulded ABS enclosure, e.g. Bopla E440, or
OKW A9030065 .

PCB Type 880189 i

components internal to the detector
create a low-pass filter for shaping and
cleaning the digital pulses. This filter not
only suppresses high-frequency signals,
but also sets the maximum data rate —
in this case, to 1 Kbit/s. Background
signals at the mains frequency are re-
jected by the AM-suppressor. This works
by storage of the average direct voltage
level in C-. When no input signal is
available for more than 4 s, the voltage
on C- would rise slowly to a value that
results in a logic high level at the output.
This is prevented by R3, R4 and Rs.

The comparator, in combination with
Cs, cleans the detected pulses, whose
edges are straightened again by the inter-

Fig. 5. Receiver amplifier gain vs frequency for different values of the high-pass capacitor.

standing considering the cost. PRO-
COMM is set to the Kermit mode with
the following line settings (ALT-P; op-
tion 7): 8 data bits; 1 stop bit, no parity;
300 baud; half-duplex and a time-out of
999 ms. In Kermit mode, PROCOMM
allows the user to define the packet size.
Initially, go to the Kermit setup menu,
and select a small packet size to keep
resending time low.

The Kermit protocol works basically as
follows. The first packet sent by the
computer is accompanied by a CRC byte
(CRC = cyclic redundancy check). The
CRC byte generally provides better
results than a checksum by virtue of a
different method of calculation: the
checksum is obtained by addition, the
CRC by division. After reception of the
data in the remote computer, the CRC is
checked, and a message is returned to in-
dicate whether or not the packet has to
be resent. This process is repeated, if
necessary, until correct data has been re-
ceived.

nal bistable. This has an open-collector
output that drives a simple computer in-
terface set up around Tj. Table 1 lists
the possibilities of configuring this inter-
face in accordance with three interfacing
standards. Since an RS-232 interface
works with positive and negative voltage
levels, the interface should be extended
as shown in Fig. 3. Ris and Ris simply
raise the ground potential of the inter-
face to half the supply voltage of the
modem. This results in the circuit driv-
ing the RS-232 interface in the computer
with a voltage swing of ±6 V, which is
adequate for correct operation in most
cases. Ground of the circuit is, therefore,
not ground of the RS-232 interface. One
additional resistor, R17, is needed to
protect the data input of the modem
against the voltage levels of up to ±12 V
supplied by the computer’s RS-232
driver.

Noise suppression by the modem can be
improved by increasing the value of Cs
and Cs to 10 nF and 100 nF respect-
ively. This measure effectively results in
a lower bit-rate of 300 per second, but
speed up communication between
modems since less information needs to
be sent back and forth on account of
corrupted data. Finally, some ex-
perimenting may be required with the
value of C4 — a lower value results in a
narrower bandwidth of the input ampli-
fier. Possible capacitor values lie be-
tween 470 pF and 1 nF.

Construction: safety first

For your own safety, the power line
modem must never be constructed on a
printed circuit board other than the one
shown in Fig. 4.

Completion of the board with reference
to the parts list is not expected to cause

difficulty. The unit is fitted in an ABS
enclosure provide with a grommet and a
strain-relief clamp for the mains cord.
The connector or socket for the bidirec-
tional serial link to the computer should
be located as close as possible to the
rfelevant connections on the printed cir-
cuit board, so that the wires can be kept
as short as possible.

One adjustment

To begin with, the data input of the
modem should be held at about +5 V.
This is easiest done by connecting a
27 kS2 resistor between the input and the
+ 12 V line in the modem. Never apply
power until a thorough check of the
completed board, and the way it is con-
nected to the mains, has been made.
Power up and use an oscilloscope to in-
spect the waveform at pin 20 of ICi.
Adjust the core in Li for maximum am-
plitude of the carrier. When an oscillo-
scope is not available, an analogue
voltmeter may be used instead, but only
if this is known to be able to work at
120 kHz in the alternating voltage range.

Sending and resending
packets: enter Kermit

As already hinted at, reliable data com-
munication with the modem can only be
achieved when the computers at both
ends run a communications program
capable of error detection and correc-
tion. Owners of the Commodore C64
computer are advised to use General
Electric’s excellent program

HOMENET.

The prototype of the power line modem
was tested under control of the PC com-
munications package PROCOMM ver-
sion 2.4.2, whose capabilities are out-

Once the maximum feasible parameters
for data communication with the aid of
the Kermit protocol are known with both
modem stations, the chat mode in PRO-
COMM can be selected for on-line com-
munication between connected PC
stations. H

HOMENET is a registered trademark
of the General Electric Corporation.
The HOMENET communications
package for C64 computers may be
obtained by contacting The Industry
Standards Staff, General Electric Cor-
poration, Fairfield CT 06431, U.S.A.
Reference: Philips Components
AN1951.

Procomm is a registered trademark of
Datastorm Technologies Inc., P.O. box
1471, Columbia MO 65205, U.S.A.

The latest version of Procomm is
stated to cost US $35.00 including
disk. Datastorm’s auto-answer BBS
service can be contacted 24 hours a
day and 7 days a week on telephone
number USA 314 449-9401.

Note: in Philips Components' Appli-
cation Note ANI9SI on the NE5050, a
line transformer identified as TOKO
AMERICA #707VX-T1002N is rec-
ommended for 110 V mains networks.

.4.29

HYBRID VHF/UHF WIDEBAND
AMPLIFIERS

Recently, Philips Components have added a number of new
devices to their well-established OM3xx and OM9xx ranges of
hybrid wideband amplifiers made in thick-film technology. The five
new integrated circuits provide a wide range of gains, and should
be of particular interest for the design of VHF/UHF wideband
boosters, since they require remarkably few additional compo-
nents. A fully worked out application of the new chips in such a
booster is included in this article.

The new devices in Philips Components’
series of integrated wideband amplifiers
include a single-stage type, the OM2045
with a gain of 12 dB, a two-stage type,
the OM2050 with a gain of 18 dB; and
two three-stage types, the OM2060 and
OM2061, with gains of 23 dB and 28 dB,
respectively. All of these can be used as
RF gain blocks with an input and output
impedance of 75 Q, in the frequency
range between 40 and 860 MHz.

Since virtually all that is necessary for
building a reliable wideband RF ampli-
fier with good specifications is contain-
ed in a single chip, many applications are
feasible. The amplifiers are, for instance,
ideal for use in the domestic cable net-
work for radio and TV, in which ad-
ditional gain is often required to over-
come cable losses. Radio amateurs, too,
will find the amplifiers useful for
general-coverage reception experiments,
as the 6-m band, 2-m band and 70-cm
arc covered in one go. One further appli-
cation is the use in 480 MHz or
612 MHz intermediate-frequency (IF)
amplifiers of indoor units for satellite
TV reception which incorporate a
surface-acoustic wave (SAW) filter with
high insertion loss.

A practical design

The circuit diagram of Fig. 1
demonstrates the simplicity of a
VHF/UHF wideband amplifier set up
around one of the new OM20xx types.
Apart from a supply and, of course, the
hybrid chip, all that is needed to obtain
a complete RF amplifier are two
capacitors and a small choke if a two- or
three-stage amplifier chip is used.
Thanks to the simplicity of the circuit, it
can be housed in a compact enclosure.
The supply voltage for all amplifier
chips is 12 V ±10 | 7o at a maximum cur-
rent drain of 110 mA (OM2070), allow-
ing the use of a simple power supply
composed of a small 15 V mains trans-
former, a 500 mA bridge rectifier, a
4.30 elektor India april 1989

B1 = B40C500

Philips Components’ latest

Circuit diagram of the wideband aerial boo:
i a series of hybrid amplifiers.

Semiconductors:

Bi -B40C600 bridge
1C 1 = 7812
IC2 * see text

mounting.

socket.

220 p F smoothing capacitor*and a 7812
integrated voltage regulator with the two
usual decoupling capacitors.

Construction of the RF
amplifier

The printed-circuit board shown in
Fig. 2 was designed to make construc-
tion of the wideband amplifier as simple
as possible, while still allowing the con-

Component mounting plan of the double-sided printed circui

supply H
output

jpply (+)
utput/supply H

1 common
1 output/supplyH

structor to choose and use any of the
five new amplifier chips. Since the pinn-
ing of these is, unfortunately, not consis-
tent (see Fig. 3), short wires are used in-
stead of PCB tracks to connect input,
output and supply terminals. In view of
the relatively high frequencies involved,
it is imperative that these wires, notably
the earth connections, are not longer
than 1 to 2 mm. In all cases, reference
should be made to Fig. 3 to ascertain the
pinning of the selected chip.

The power rating of the 15 V trans-
former on the PCB should be in accord-
ance with the RF amplifier chip used —
see Table 1 for the main specifications of
these. When the OM2045 is used, a
1.2 VA transformer should do. The use
of the OM2070, however, calls for a type
rated at not less than 3.3 VA. It should
be noted that some transformers require
two short pieces of wire between the sec-
ondary terminals and the tracks leading
to the AC connections of the bridge rec-
tifier.

The PCB is cut in two along the dashed
lines. The part with the round, etched,
holes is drilled to accept the input and
output sockets, and the grommet for the
mains cable. After drilling, this part of
the PCB is soldered vertically on to the
main amplifier board as shown in the
photographs. Small pieces of tin-plate
are bent to shape and soldered round the
input and output sockets for additional
screening.

The main board may now be populated,
with the exception of the amplifier chip,
C t, Cs and C<.. The centre pin of voltage
regulator ICi is soldered at both sides
of the board.

Ten non-connected solder spots are
reserved for IC 2 , whose pins are connec-
ted with the aid of wires as outlined
above. It is recommended to fit these
connections at the reverse side of the
board. The input marked supply (+) is
connected to point P. Three ICs, the
OM2060, OM2061 and OM2070, require
an additional connection between the
supply and the chip output. This con-
nection is made in the form of a 5.6 pH
choke between the output and point P,
as shown in Fig. 4.

Coupling capacitor Cj takes the RF in-
put signal direct from socket Kj to the
input of IC 2 . The amplified RF output
signal is coupled out to K: via Cs. To
prevent stray inductance and possible os-
cillation, the wires of Cj and Cs should
be kept as short as possible. Capacitor
C<. (2p2) may be added for extra sup-
pression of interference. M

91 1

i _

V * «*■ 2

v — — — *

1 - ** ',1

p* s cjyp

Fig. 4. Series-connected supply choke L2 is fitted at the reverse side of the board.

4.32

Fig. 5. Completed prototype of the wideband RF amplifier.

CENTRONICS-COMPATIBLE
PRINTER BUFFER

from an idea by R. Degen

Today’s computers and the programs that run on them are
capable of generating massive amounts of data that is, in some
way, to be put on paper. Users of computer-assisted design and
engineering (CAD/CAE) and desk-top publishing (DTP) programs
need not be told that the printer or plotter is almost invariably a
slowing-down factor in the system. At printing time, therefore, the
user is often forced to sit with his arms crossed, or go out to have
a cup of tea, because the computer has insufficient memory left
to store the whole of the printable file. Intermediate storage of
data on disk and so-called spooler programs only partly resolve
this annoying problem.

The versatile printer buffer described here is a state-of-the-art
design that eliminates printer wait times. Just look at the main
specifications below to convince yourself that this is your next
home-made computer peripheral.

Printer wait times arise when the
amount of data to be fed to the printer
exceeds the free memory capacity of the
computer. Today’s wordprocessors,
CAD/CAM/CAE and DTP programs
are so large that, believe it or not, very
little memory is left for the work file, be
it a text, drawing or graphics image.
Often, no more than a few tens of
kilobytes are left of the 640 or so in-
stalled in the PC. The programs then in-
variably use a disk drive to temporarily
store the excess data, which is ‘spooled’
to the printer output via the small, inter-
nal buffer and a background program.
Meanwhile, however, the user can not ex-
it the program, and further text or
graphics editing may be slowed down
considerably because of the spooling
process.

Documentation and other text files are
becoming ever larger, too. Many so-
called Public Domain programs and
PC utilities are accompanied by a

compressed documentation file, which,
when de-compressed ( unpacked or
uncrunched) by the user, results in a
printable .DOC or .MAN file of

Printer buffer

• Centronics compatible

• Memory: user-configurable from
32 KByte to 1 MByte, or from
128 KByte to 4 Mbyte in six steps.
Option to use either 32 KByte or
128 Kbyte static RAM chips (xx256
or xxl024).

• Low current consumption (40 mA)
enables powering from printer. Op-
tion to use mains adapter with DC
output

• Compact unit thanks to surface-
mounted components

• No microprocessor

• Repeat mode; buffer does not load
data while feeding printer.

100 kilobyte or so, which takes 15 to 30
minutes to dump on most matrix
printers. Most modern matrix and ink-
jet printers can be fitted with extra
buffer memory, but the cost of such an
extension is often quite high relative to
the price of the basic printer. The most
expensive of add-on buffers often pro-
vide ‘only’ 64 KByte, which is no great
help when very large files are handled.

A non-used character?

The present circuit is based on the fact
that no printer prints ASCII character
00. In practice, the operation of the
printer buffer is as follows: the computer
writes data into the memory of the
printer buffer. When the data flow to the
printer buffer ceases, a user-defined
delay is introduced before the data lines
are made logic low, so that the remaining
memory is loaded with zeros (00). The
printable file is thus held in the printer
elektor India april 1909 4.33

Fig. 1. Basic internal structure of the printer buffer.

buffer, and its size is known. When this
process is completed, the file(s) held in
the printer buffer arc ready for sending
to the printer. The total content of the
buffer memory, including the zeros, is
then fed to the printer (the zeros, of
course, do not appear on paper!). The
computer is called upon only when the
size of the file to be loaded into the
printer buffer exceeds the available
storage capacity (this depends on the
memory configuration selected by the
user, and will be reverted to). Provided
the computer has not produced a time-
out error in the mean time, the re-
mainder of the file is loaded after the
printer has completed printing the mem-
ory content of the buffer.

Obviously, to avoid printable files being
loaded in two or more passes, the
buffer’s memory should have capacity
at least equal to the size of the largest
anticipated printable file. Few problems
are expected here, however, considering
that 128 KByte RAMs can be Fitted in

the circuit. A few examples: the text file
for this very article is 29,287 bytes large
(WordPerfect 4.2), while the circuit
diagram, originally drawn with the aid
of OrCad-SDT3, takes up 360 KByte
(size A3 sheet). The Postscript DTP file
used for composing galley-proofs of this
article with the aid of Ventura Publisher
1.2 occupies 422 KBytes. CAD pro-
grams, such as PCB design and
schematic drawing packages, invariably
switch the matrix printer to its graphics
mode, and little needs to be said of the
‘printing speed’ then achieved. . .

Functional description of the
printer buffer

The printer buffer is a relatively complex
circuit and it is, therefore, useful to first
get aeqainted with its general structure,
shown in the block diagram of Fig. 1.
The function of the keys on the printer

buffer is as follows:

WAIT

When several files are to be printed, the
printer buffer can be switched to wait
mode so that it can load all printable
files in succession.

REPEAT

This key enables the buffer to print the
same file more than once (copy func-
tion).

RESET

The buffer can be reset and re-initialized
by pressing RESET. Internal bistables
and the memory size counter are reset to
zero. It should be noted that reset over-
rides the repeat function, so that re-
printable characters in the buffer may
corrupted. The RESET key should not,
therefore, be actuated before the buffer
has completed feeding out all of the
copies selected with the repeat function.

With reference to the block diagram in
Fig. 1, the central part is the
buffer’s memory with its associated ad-
dress decoding and address counting cir-
cuits. The address counter is clocked
during the loading as well as feeding out
of data. Loading is clocked by the strobe
pulses supplied by the computer, and
feeding out by an oscillator. At the com-
puter side, an input buffer is provided
for the databits, and a clean-up circuit
that shapes the strobe pulses and
prevents double clocking. A third block
takes care of the BUSY and ACK
(acknowledge) handshaking with the
computer.

The buffer loads and stores data as long
as strobe pulses are applied by the com-
puter. The strobe detect block monitors
the reception of strobe pulses. When
these fail, the oscillator clock is enabled,
either to fill the remainder of the inter-
nal memory with zeros, or, when the
memory is full, to start feeding the

The block marked WAIT FOR INPUT is
controlled by WAIT switch Sj which
allows the strobe detect signal to be over-
ridden, thus forcing the buffer to load
further data (but only if free memory is
still available).

The functions of blocks REPEAT,
IN/OUT SELECT and RESET are ob-
vious. IN/OUT SELECT determines the
data transfer direction: from computer
to buffer (IN), or from buffer to printer
(OUT). The databus buffers are required
to ensure stable signal levels even when
the maximum number of RAMs, 32, is
installed.

The BUSY and STROBE signals derived
from the previously mentioned oscillator
control the data flow between the buffer
and the Centronics input of the printer.

The circuit in detail

The above functional blocks are found
back fairly easily in the circuit diagram
of Fig. 2.

The input handshaking circuit of the
printer buffer is composed of IC12 and
D-type bistables FFj and FFj. Circuits
IC7 and ICs form the address counter,
and IC10-IC11 the address decoder. The
memory of the printer buffer is formed
by static CMOS RAMs in positions ICis
through ICj 2. Bistable FF: and timer
IC? function as strobe pulse detector
that determines when the file(s) has
(have) been loaded completely. Direction
switching (IN/OUT) as outlined above is
effected by bistable FFi and Schmitt
trigger gates Nn, Nis and N19.

The central oscillator is an R-C type
built around NAND Schmitt-trigger
gate N ifi. Inverters Nn and N12, finally,
supply the strobe signal for the printer.
The memory extension circuit is shown
in Fig. 2b. Each extension card holds
8 RAM chips, which are either 32 or
128 KByte types. The extension(s) is/are

essentially connected in parallel to the
basic memory on the main board.

Timing is essential

The letters shown at a number of essen-
tial points in the circuit diagram refer to
the timing diagram of Fig. 3.

Bistable FFj slightly lengthens the
strobe signal. A, which is supplied by the
computer’s Centronics port, so that a
well-defined rectangular signal, B, is ob-
tained for driving gate N20. This sup-
plies the computer with handshaking
signa ls BUSY (C) and, via Ns and FFj,
ACK (D). Note that some com-
puters use B USY a s the handshaking
signal, others ACK, and still others
both. The printer buffer is compatible
with all of these.

The strobe detection circuit uses a Lin-
CMOS ( Linear Complementary Meta I
Oxide on Silicon) timer Type TLC555
(IC9) from Texas Instruments. The first
strobe pulse from the computer triggers
the TLC555, which drives its Q output
logic high. This causes the output of in-
verter Ns to go low (signal G), so that
the data reception indicator, LED Ds,
lights. When the strobe pulses cease,
timing capacitor Cs is charged via R?
and preset Pi, which allows setting a
delay between 5 and about 30 s. When
this delay has lapsed, the voltage on Cs
resets the TLC555. As long as strobe
pulses are being received, however, Cs is
discharged by Ti, so that IC9 can not be

The rising edge of signal G clocks
bistable FF2. Since the D (data-) input
of FF2 is logic high, output Q goes low.
This results in the output of N19 going
logic high (signal H). The event marks
the switching over from IN (computer to
buffer) to OUT (buffer to printer), and
at the same time causes the BUSY line to
the computer to be actuated.

After a short delay introduced by R12-
C?, the oscillator around Nia is started.
The memory space available after
loading the file(s) is then filled with
zeros by disabling the data input latch,
IC12, and pulling the datalines to the
RAMs to ground with the aid of 8-way
resistor network Ri«.

When the address line selected with
RAM-configuration switch block S2
goes logic high, the outputs of FFi
toggle. Functionally, this means that the
RAM is switched over from read to write
(WE, signal N, is actuated). Via Ni
and Nu, the clocking of FFi also causes
the address counter to be reset in
preparation for the feed-out operation.

Signal P controls gate Nis, and so
enables the communication with the
printer to be established. Gates Nis and
inverters N11-N12 conver t the o scillator
pulses to strobe pulses (PSTB; signal

R) for the printer, which responds to
them by actuating output line BUSY
(signal Q). This stops the oscillator while
a character is printed. When BUSY is de-
actuated, a new strobe pulse is
generated.

Components Ris and C9 delay the
strobe signal briefly with respect to the
selection signal, F, for the address
decoding circuit. This is done to ensure
that the datalines are stable when the
strobe line goes low.

The next clock pulse applied to FFi
causes this to revert to the start state, and
FF2 to be reset via C2 and Ra. This
brings the circuit back in the initial state.

The second part of the timing diagram
illustrates what happens when the
printer buffer is fully loaded. Bistable
FFi takes control of the data-buffers,
and switches the circuit to the OUT
mode (buffer to printer). Components
R>2 and Ci ensure correct timing of this
operation, preventing loss or corruption
of printable data. After printing, D9
rapidly discharges Ci, and so prevents
the oscillator from running on, which
would result in the last character being
printed a number of times. During this
operation, the computer is set to wait
when insufficient memory is available. It
will be clear that fitting enough memory
in the buffer is the best way to avoid this
situation.

More details . . .

The time before the buffer starts feeding
data to the printer can be adjusted with
Pi (max. 30 s). The delay can be set in
accordance with the type of data sent to
the printer. Graphics data, for instance,
generally gives rise to a fairly heavy cal-
culation load, so that quite some time
may lapse. When the available maximum
delay of 30 s is too short, or when a
number of separately loaded Files are to
be printed in rapid succession, the buffer
may be set to WAIT mode with the cor-
responding key. When WAIT is de-
actuated, the set delay is introduced
again, and printing may recommence.

The RESET key rc-initializes the buffer
as at power-on. Printing may, of course,
also be interrupted at any time by turn-
ing the printer off-line (SELECT or ON-
LINE key).

Extra copies of the printout may be ob-
tained by pressing the REPEAT key. One
proviso here, however, is that the
previous run was not interrupted by a
reset. This is because the reset circuit
works asynchronously and may,
therefore, modify the memory content.

Building the printer buffer

The printer buffer is built partly with
surface-mount assembly (SMA) parts.

eleltor India apriM989 4.35

The basic circuit is composed of two
printed-circuit boards: the main board
(Fig. 4a), which holds the digital control
and memory circuits, and the keyboard
(Fig. 4b), which holds the 3 control keys,
and 4 LEDs.

Memory configuration: look before you
leap

It was already noted that the user deter-
mines the memory size of the printer
buffer. A third PCB is, therefore, pro-
vided as a memory extension unit (see
Fig. 5) that accepts two types of static
RAM: 32 KByte (43256 or 84256) or
128 KByte (841024). This third board is
only required to increase the memory
size of the basic printer buffer beyond |
256 Kbyte (32 Kbyte chips used) or I
1 MByte (128 KByte chips used.). Atten-
tion: it is not possible to mix 32 KByte
and 128 KByte RAM chips. When the
memory is to be upgraded, either the I
first choice must adhered to, or all chips |
must be removed and replaced by other
types. On the PCB, there can be no
doubt about the location of the chips,
since 32 Kbyte and 128-KByte types are
supplied in a 28-pin and 32-pin package
respectively.

Five short wire links at the copper side
of the PCB define the maximum size of |
the memory (either 1 MByte using 32
32 KByte RAMs, or 4 MByte using 32
Fig. 2b. Circuit diagram of the optional memory extension board which, depending on the 128 KByte RAMs) hence, the wire
RAM chips fitted, provides 256 KByte or 1 MByte of additional buffer capacity. links are marked 1M and 4M. Fit the

Resistors (±6%);
Ria;R 2 i = 8-way SIL re
Pi — 1M0 preset H

Capacitors:

Ci-lpO; 16 V
Cs-lOp; 16 V
Cl2=100p; 26 V

ICi2;ICl3« 74HCT373
ICl4»7806

1C 1 6 . . . IC 22 Incl. =43266 INECI or 62266 or
84256 (Fujitsu) (32Kx8)l or 841024 (Fujitsu)

Miscellaneous:

Ki;K2;K3= 10-way pin header.

K4= 34-way pin header.

K5;Ke= 20-way pin header.

Si = push-to-make button with large black cap
(ITW 61-10204001* or Digitast.

ly DIL-sv

:h block.

S3= push-to-make button with large red cap
(ITW 61-10200001* or Digitast.

S4= locking button with large black cap (ITW
61-2020400)*.

2 off 20-way IDC sockets.

1 off 36-way Centronics (female) connector fc
panel mounting.

1 off 25-way female D-connector for panel
mounting.

Flat-ribbon cable as required.

Enclosure: e.g. BICC-Vero Type 4775-1410.

PCB Type 890007-1 '

PCB Type 890007-2

* ITW Switches • Division of ITW limited •
Norway Road, Hillsea e PORTSMOUTH
P03 6HT, Tel: (0706) 694971. Telex: 86374.
Fax: (07051 666352.

SURFACE-MOUNT ASSEMBLY PARTS:

Resistors:

Ri;Rio = 100R

R2;R3;R7;Rn;Ri3:Ri4;Rie;Ri0;R22. . . R25
incl. -1 OK
R4;R17 = 1M0
Rs;Ra;Re;Ris = 330R
Ra;Ri2»100K
R 20 -IKO

Capacitors: C2,C7;Ca= InO
C3;Ce;Cie;C27;C28« lOOn
C4;Ca;Cio= lOOp
Cn = 470p

Semiconductors:

Dl;D2,'De;De;De=BAS32 (SMA equivalent of
1N4148)

ICi;IC2- 74HCT14

IC3;IC4=74HCT132

ICs;IC6»74HCT74

IC7;ICa = 74HCT4040

ICa = TLC556 (Texas Instruments)

ICio;ICii = 74HCT154

links marked 1M when xx256 R AMs are
used, and the links marked 4M when
xxl024 RAMs are used.

A dual-in-line (DIL) switch block. Si, is
used for setting the actual memory size.
Only one switch may be closed at a time:
switch 1 selects 32 KByte, switch 2
64 KByte, switch 3 128 KByte, and so
on, to Ss which selects 4 MByte. It is
seen that each switch doubles the
amount of memory, and extending the
memory means, therefore, doubling its
size (it is not possible to add, say,
32 KByte when 128 KByte is already
available: the next step is 256 KByte).

With the cost of the RAM chips in mind,
the possibilities for future extensions
should always be studied beforehand.
For instance, for a 256 KByte configura-
tion, there is a choice between eight
32 KByte and two 128 KByte RAMs.
The latter option may currently be the
more expensive, but has the advantage of
allowing a future upgrade to 1 MByte
(on the main board) or 4 MByte (with 3
off 1 MByte extehsion boards).

Fitting the SMA parts

The SMA parts are the first to be
mounted at both sides of the main PCB,
which is double-sided and through-
plated.

There is no mystery about fitting SMA
parts if a few basic precautions are ob-
served:

• SMA components generally do not
have a printed type or value indi-
cation: therefore do not remove them
from their labelled package before they
are due for mounting;

• use a low-power, temperature-
controlled, soldering iron with a fine

tip, and clean this after every soldering

• use thin (<1 mm dia.) soldering wire
to avoid short-circuits between adjac-

• solder as quickly as possible to pre-
vent overheating the component;

SMA integrated circuits should be
placed and aligned carefully. Then
solder two corner pins and once more

Fig. 5. Component mounting plan of the memory extension board. Depending on the mem-
ory configuration, 10-way header K» is connected to Ki, Kz or K< on the main board.

verify whether all pins line up correctly
with the relevant solder islands.

For passive SMA parts, it is best to first
pre-tin one of the tracks with a tiny
amount of solder. Position the part, and
heat the connection on the pre-tinned
track. Then solder the other part con-
nection. Again, avoid overheating and
excess amounts of solder tin.

When all SMA parts have been fitted, a
magnifying glass is used to inspect their
connections to the tracks on the board.
Also check all solder joints for possible
short-circuits.

The standard parts

The first non-SMA part to be fitted is 8-
way DIL switch Sz. This is mounted as
an SMA integrated circuit, slightly above
the PCB surface so that its pins are ac-
cessible for soldering. If changes in the
memory configuration are not foreseen,
Sz may be omitted: the connection that
selects the relevant RAM size is then
made by a wire link. Proceed with fitting
the sockets for the integrated circuits,
and the memory extension connector(s).
The rest of the construction is entirely
straightforward.

The memory extension board is not
through-plated. With the exception of a
number of capacitor leads, the points
where through-contacting is effected are,
fortunately, located well away from com-
ponents. Start the construction of this
board with fitting the through-
contacting wires, as these are difficult to
reach once the IC sockets have been

mounted. A simple method of through-
contacting the PCB is to temporarily in-
sert four M3 screws with nuts in the cor-
ners of the board, so that this is a few
millimetres above the working surface.
Then insert the through-contacting wires
vertically until they rest on the work sur-
face. Cut off the excess wire, and solder
at the top side. Once all the wires have
been fitted, the board may be reversed
and the screws removed. The free side of
each through-contacting wire is then
(quickly) soldered to the relevant spot.

Power supply

The printer buffer may be powered
either by the printer or by an internal
power supply. Consult the manual sup-
plied with your printer to check whether
this supplies +5 V at pin 18 of its Ccn-

PRINTER BUFFFER: MEMORY EXTENSION
BOARD

Capacitors:

C29. . . C36 incl, = 100n
Semiconductors:

IC23. . IC30 incl. = 43256 or 84256 or 62256
I32K x 81 or 841024 (256Kx8)

Miscellaneous:

<7= 34-way angled pin header.

K8= 10-way angled pin header.

Ka;Kro= 34-way IDC socket.

Kri;Ki2» 10-way IDC socket.

Flat-ribbon cable as required.

PCB Type 890007-3

tronics input connector. If this is not so,
wire link Y is fitted, and the 5 V regu-
lator circuit on the main PCB is powered
from a mains adapter with 8 to 12 VDC
output. Wire link Y is omitted, and wire
link X is installed, when the buffer is
powered from the printer. When the ex-
ternal power supply option is used, it is
recommended to connect the mains
adapter via a small, 2-way DC-input
socket as used on portable cassette
players and some older types of pocket
calculator.

Cables and connections

The main board has two 20-way pin
headers for connecting the input and
output cables. The pin headers mate
with 20-way IDC sockets secured on to
short lengths of flat ribbon cable. The
input cable is fitted with a 36- way Cen-
tronics (‘blue-ribbon’) connector, the
output cable with a 25-way D-connector.
This arrangement allows the printer
buffer to be connected with the aid of a
pair of standard, inexpensive, printer
cables.

Fig. 7. Pinning of ihe inpul/outpul connectors. Ks and K 6 . on the main buffer, and their wiring to a 36-way Centronics input connector (in-
put) and a 25-way female D-type (output).

Figure 7 shows the wiring diagram of the
input and output cables. The pinning of
the input and output connectors is ident-
ical. On these, the interconections arc
made as indicated. When a 25-way D-
conncctor is use d at the output of the
buffer, pin 15 (ERROR) may be
used to carry the +5 V supply voltage
taken from pin 18 of the Centronics con-
nector at the printer side.

The memory extensions are bused and
connected to Kj via a 34-way flat-
ribbon cable. Each memory extension
board has a 10-way pin header which is
then connected to headers Ki to Kj on
the main board, observing the logic
order of the extension boards: the first is
connected to Ki, the second to K:, and
the third to K.t.

The control panel, of which a suggested
lay-out is given in Fig. 8, is connected to
the main board via individual wires.
Switch S-i is a 2-position locking type
from 1TW.

The size of the control panel is such that
it is easily installed vertically behind the

front panel in an ABS enclosure Type
4775-1410 from BICC-Vero. The main
board and the control board are
mounted on to an aluminium base plate.

A drilling template for this support plate
is given in Fig. 9.

Test time. . .

The power LED on the printer buffer
should light at full intensity when the
unit is switched on. If it does not, the
power supply in the printer is not
capable of delivering the required cur-
rent, and a separate power supply should
be used as discussed earlier. It should be
noted that the printer buffer draws a
small cur rent from the computer via the
STROBE connection. This current
causes the power LED to light dimly.

Once the presence of the correct supply
voltage has been ascertained, the WAIT
key is pressed. The associated LED
should light. Release WAIT.

Send a file to the printer buffer, which
shows reception of data by lighting the
input LED. After a delay determined by
the size of the file, the input LED goes
out. The following delay depends on the
memory size, and is about 15 s with
256 KByte installed (remember that
some time lapses before the non-used
part of memory has been filled with
zeros). The output LED will now light.

Printing commences, and the output
LED goes out when the print job is
finished. The repeat function may now
be tested. When the relevant key is
pressed, the associated LED lights, and
the buffer should feed out a copy of the Fig. 9. Drilling template for a support bracket that holds the keyboard.

previously loaded file. When a number
of files are to be loaded for printing in
one go, the WAIT key is actuated before
the first file is sent to the buffer. Pressing
this key remains possible until the file
has actually been loaded. The WAIT key
is pressed again when the last file in the
batch has been sent to the printer buffer.

The memory configuration switch, S2,
may be replaced by a single wire link if
frequent changes in the RAM size are
not anticipated. Other options for this
switch include an 8-way rotary type, or
mounting it on to the rear panel of the
printer buffer and connecting it to the
main board via a length of flat-ribbon
cable and a 16-way DIL header. The
rotary switch is a particularly useful ar-
rangement because it allows memory
size to be reduced quickly when a rela-
tively small file is to be loaded (because
less memory is available, less time is
needed to fill the non-used part of it
with zeros).

N

DESIGN IDEAS

COUNTER WITHOUT COUNTER

Under this paradoxical title we present a design idea for a
versatile counter concept that uses an EPROM instead of the
expected counter chip.

by N. Korber

The circuit described here can be con-
figured as an up- or down-counter from
0 to 99 in BCD or 8-bit binary mode,
with reset, preset and enable inputs
available. All control inputs are digital
compatible, allowing the user to define
his own control hierarchy. Moreover, the
control inputs may be active high or ac-
tive low.

The counter is actuated by the positive
transitions in the clock signal, and
handles input frequencies well into the
MHz-range. All inputs and outputs are
TTL-compatible. The counter is so
remarkable because its features and ver-
satility are achieved with only a handful
of commonly available components.

The Moore system

From a point of view of information
technology, the present counter is similar
to a so-called synchronous transforming
Moore circuit. The indication synchron-
ous has to do with the clock signal and
the way in which the inputs are driven. A
Moore circuit is a logic unit that pro-
cesses input parameters x and internal
conditions z to produce output states y.
Each condition is associated with only
one output state. As a result of input
parameters and internal conditions, the
Moore circuit steps through a number of
states. The system not only uses current-
ly available information, but also infor-
mation acquired from past operations,
whose system conditions have been
recorded. A clock signal is required to
switch the system to the next state.

In practice

A real Moore system is composed of a
switching network and a memory. In the
case of the present counter, the input
parameters are the data applied to the

Fig. 1. Basic design of an EPROM-based counter.

1 4.43

control inputs. These are connected to
an EPROM, which combines these data
with the current conditions, to generate
a system state. This state is copied into a
latch (IC2 in the circuit diagram), and
so becomes the current state. In prin-
ciple, output state y would then have to
be generated with the aid of a further
switching element. This procedure is not
needed here, however, by virtue of the
EPROM, which, if properly pro-
grammed, ensures that the output value
(i.e., the counter state) is exactly the
value that corresponds to the current
state.

The operating principle of the circuit
discussed here is fairly complex, but may
be explained with the aid of a
hypothetical, circuit, based on an
EPROM Type 2764 addressed between

0000 and 1FFF, that serves to count
seconds pulses applied to the clock input
by an external circuit.

First consider the basic timing of the
events that take place in the circuit.
Because switch ENA (enable) is connec-
ted to address line A12 of the EPROM,
it divides the available memory capacity
in two equal halves. Each of these is, in
turn, subdivided in two by line All
(switch PRESet), and again in two by the
RESet switch connected to A10. For the
actuation of the RESet switch to cause
the displays to indicate 00, it is necessary
that all memory sub-partitions ad-
dressed with A10=l (for example, 0000
to 03FF, or OBFF to FFFF) contain data
00.

With this in mind, and returning to the
seconds counter, it is clear that 00 should
be displayed when RESet is actuated. To
achieve this, data available at outputs Q0
to Q7 of octal bistable IC? must cause
the display drivers Type 9368 to drive
those display segments that together

A positive pulse transition at the clock
input of the 74LS374 causes the logic
state applied to each data inputs, Dn, of
the chip to be copied to the correspond-
ing output, Qn. Assuming that the cir-
cuit is to function as a down-counter,
EPROM address line A9 is made logic
high by closing switch DOWN. In the

1 KByte memory area adressed,
512 bytes may be programmed such that
the displayed value is decremented. Care
should be taken to ensure a correct pro-
grammed sequence, since 59 must be
displayed once the counter has been
started.

The circuit diagram of Fig. 1 clearly
shows the sub-units of the counter: the
external controls in the form of switches,
the EPROM, the latch circuit, and the
display drivers for two 7-segment LED
displays. The control parameters of the
counter, ENAable, RESet, PREset and
DOWN, and the current state, are ap-
plied in parallel to the address inputs of
the EPROM. The EPROM uses the ad-
dress so obtained to produce infor-
4.44 elektor India april 1989

mation on the next state. This infor-
mation is made available on data out-
puts DO to D7, and, a little later, on out-
puts Q0 to Q7 of the latch, from where
it is sent to the displays. This process
starts with each rising edge of the clock
signal.

Programming the EPROM

The actual contents of the EPROM de-
pend on the application of the counter
circuit, and must, therefore, be provided
by yourself.

The 8 Kbyte EPROM is perhaps best
thought of as 8 blocks of 256 bytes.
Each of these blocks is addressed by ap-
plying a particular (model-) dataword to
the address inputs. When the control
parameters remain umchanged, only the
information applied to inputs A0 to A7

determine the output state. The changes
from one state to another are a function
of the preprogrammed contents of the
relevant block. If, for example, RESet is
pressed, the change to the next state
results in the displays reading 00.

As a practical programming example,
Table 1 lists the EPROM contents for a
down-counter and an up-counter with
60, cyclic, states — the seconds counter
discussed above. For a similar counter
with, say, 100 states, the memory lo-
cations up to $1198 have to be loaded,
with $1198 and 1199 reading 99 and 00
respectively (UP function). Similarly,
the down-counter starts at $1200 with
data 99, and 98 at $1299. Table 2 shows
a further example of how the EPROM
may be programmed: in this case, a two-
digit decimal up/down counter is ob-
tained. B4

THE DIGITAL MODEL
TRAIN — PART 2

by T. Wigmore

The second part in the series describes a locomotive decoder
that is constructed in surface-mount technology. In conjunction
with the associated digital control system, it enables up to 80
trains to be controlled independently. The associated control
system will be described in a future article, but in the mean time
the present decoder may be used with the Marklin digital system
or any two-rail model track.

The use of surface-mount techniques
(SMT) makes it possible to construct a
locomotive decoder from standard com-
ponents that is compact enough for fit-
ting into a locomotive. These techniques
are undoubtedly new to many readers,
but this article will show that there is no
real mystique about them.

The present decoder enables both a.c.
and d.c. locomotives to be controlled in-
dependently of one another. In its
simplest form, it is suitable for use on
tracks with a centre rail (Marklin or Trix,
e.g.) or with an overhead power line. The
addition of the two-rail adaptor dis-
cussed later in the article enables the
decoder to be used with other model
railway systems. It should be noted,
however, that although the decoder is
compact, it can not be fitted in
locomotives smaller than HO.

In Marklin stock, the space for the
decoder board is ensured, because the
change-over relay may be removed. This
is possible, even desirable, since the
decoder enables a change of direction
(and the consequent switching of the
head and tail lights) to be effected elec-
tronically.

As already stated, the decoder may be
used with a.c. as well as d.c. systems. It
is thus possible to convert a d.c.
locomotive for use on an a.c. track by
providing it with a slip contact and the
present decoder.

In principle, it is also possible to use a
Marklin locomotive on a track of differ-
ent manufacture. If, however, that track
is a two-rail system, the wheels of the
locomotive must be electrically
separated and provided with appropriate
contacts. Moreover, if a two-rail track is
used, the adaptor described later is also
required. Although possible, this conver-
sion is, therefore, in general not prac-

The two-rail adaptor is also constructed
in surface-mount technology and may be
mounted on to the decoder. The
resulting sandwich is only 2.5 mm

• independent control of up to 80

• may be driven by Marklin HO system
or Elektor Electronics Digital Model
Train System

• suitable for a.c. and d.c. locomotives

• suitable for three-rail system or, with
optional adaptor, for two-rail systems

• motor current max. 1 A (peak 1.5 A)

• protected against thermal overioad

• speed controlled in 1 6 steps

• automatic change-over of
independently lit head and tail lights

• lamp voltage 10 V or 20 V as required

• optional: memory from external buffer
capacitor

• compact dimensions through surface-
mount technology:

35 x 24 x 7.5 mm (decoder only)

adaptor)

Table 2. Technical dala of the locomotive
decoder.

thicker than the decoder board by itself.
The use of the two-rail adaptor makes
the locomotive decoder independent of
the polarity of the supply and data con-
nections. As an aside, this also solves the
eternal problem of reversing loops.

Compatibility

Apart from the possibility of its use in a
large variety of locomotives, the decoder
may be operated with a number of con-
trol systems. In this context, it is perhaps
of interest to know that it was designed
originally and solely as part of the
Elektor Electronics Digital Multi-train
System which will be described in this
series of articles. However, with a few
simple changes (such as the baud rate),
it also proved usable with the Marklin
digital HO system. It is, of course, im-
portant to differentiate between the
Marklin decoders and the present one.
In view of the required compactness of
the decoder, the spare function offered
by Marklin has had to be sacrificed in
the present decoder (see Fig. 14).
Marklin uses the lowest speed step
(binary 1000) for reversing, assuming
that this step will not be used in practice,
since the motor does not operate
smoothly at this (average) low voltage.
To use this voltage for reversing the
direction of travel, it has to be decoded
and then used as the clock signal for a
bistable. In the present decoder, this

would have required iwo additional ICs,
which would have made the board too
large. Fortunately, in most cases the
spare function is not required anyway.
Where it is needed, there is no alterna-
tive to using a Marklin decoder.

Note that if the present decoder is driven
via Control 80 of the Marklin system,
the spare function is used for reversing.
A more important difference between
the Marklin decoder and ours manifests
itself if locomotives converted for.use in
a digital system are run on conventional
(non-digital) tracks. The Marklin
decoder may be used with conventional
control systems, i.e., the speed may be
varied according to the amplitude of the
(alternating) supply voltage and the
direction of travel may be changed by an
over-voltage pulse of not less than 24 V.
The present decoderdoes not offer these
facilities: in fact, a 24-V pulse (in prac-
tice, this value is normally considerably
higher) would probably put paid to the
power stage. It is therefore strongly rec-
ommended to use the decoder only with
digital tracks.

Furthermore, the present decoder does
not support Fleischmann’s digital FMZ
system nor that from Trix. However,
locomotives in those systems (and most
others) may be converted with the pres-
ent decoder to make them suitable for
use in multi-train set-ups. It will then de-
pend on the rail system whether, apart
from the locomotive decoder, the rail
adaptor is also required.

Start at the beginning: the
rails

The most important difference between
a digital model railway and a conven-
tional one is that in the former, just as in
real railways, the rails carry a constant
voltage. To prevent that in these cir-
cumstances all locomotives on the track
4.46 aleHor India .priM989

travel along at full speed, they are all fit-
ted with a decoder and a speed con-
troller. In other words, as in real life, the
speed of the locomotive is varied on
board, just as if it had an engine driver.
The instructions to the “engine driver”
emanate from a central control, which in
turn is controlled by a number of in-
dependent drive units or a computer.

Fig. 13. The track voltage in a digital model
railway systems is switched between -20 V
and +20 V to ensure that the control instruc-
tions reach the locomotives via the rails in a
serial data format. The locomotive power is
obtained by rectifying this alternating

The instructions from the central control
are transmitted to the locomotives by
switching the supply voltage between
+20 V and -20 V. The voltage on the
rails is thus an alternating one, but it has
a d.c. component whose value depends
on the transmitted data.

Each period contains 18 “marker”
pulses: each of the nine pairs of pulses
defines a bit with three possible states:
00=logic 0; 10=logic indeterminate;
ll=logic 1. In this way, 9-bit words are
formed: the first four are interpreted by
the decoder as address and the other five
as data. The logic indeterminate state is

used only for forming addresses; the five
data bits use only logic 0 and logic 1.
Not only locomotives, but also turnouts
(points) and signals may be controlled
via the rails. Normally, locomotives are
addressed constantly so that they exhibit
real-time behaviour. The response time
of the system becomes gradually longer
as more locomotives are taken into use.
An added advantage of the constant
voltage on the rails is that permanent
lighting of the train is possible without
any difficulty. This applies, of course,
also to the head and tail lights, even
when the train is at standstill.

Fig. 14. The data bytes for the locomotives
consist of nine bits. The first four bits form
a trinary locomotive address. The other five
are data that enable direction, speed and —
in the Marklin system — a spare function to
be controlled. One data byte is 3.8 ms long.

Block diagram

The operation of the locomotive decoder
may be seen from the block diagram in
Fig. 15. The rail voltage is full-wave rec-
tified to create a supply for the power
stages. Since the rail voltage is a square
wave, the resulting d.c. is very pure, i.e.,
it has virtually no ripple. A lower direct
voltage for the logic circuits is derived
from the supply for the power stages.
The serial data are translated by a special
decoder. These data are transmitted
direct by the ‘red’ power line in three-rail
systems or processed by the two-rail
adaptor in two-rail systems. The trinary
address part (first four bits) is compared
with the address set on the decoder. If
the two match, the next five data bits are
accepted immediately. When the same
five data bits have been received twice in
succession, they are accepted as true and
placed in the output latch.

Four of the five available data bits are
used for operating the 16-step speed con-
trol; the fifth determines the direction of
travel.

The speed is set with the aid of a digital
pulsewidth modulator (PWM). The
modulator consists of a four-bit counter
with oscillator and a four-bit compara-
tor. The counter (port A of the compara-
tor) runs constantly. The four bits of the
speed control are present at port B. De-
pending on the number at port B, the
duty factor of the output signal varies
between 0 and 15/16 at a frequency that

Fig. IS. The block diagram of Ihe locomotive decoder.

is 1/16 of the counter frequency.

The PWM signal drives the output stage.
Between this stage and the motor there is
a fullwave rectifier. The polarity of the
resulting direct voltage depends on the
direction bit. This bit also serves to
change over the head and tail lights that
are driven by two half-wave circuits.
The lighting voltage may, if desired, be
halved with the aid of a square-wave
signal.

An undervoltage detector completes the
set-up. If the supply voltage drops below

powered length of track, all signals are
disconnected from the power stage and
the logic circuits are set to the low-power
state. In this state, the logic circuits are
able to store the last received data for a
short time, thanks to an optional exter-
nal buffer capacitor. When the supply
voltage recovers, the locomotive travels
on at the last set speed.

Circuit description

mc145029 of the same family as the
mc145027 used in the turnout and signal
decoder described in Part 1 of this series.
The significant difference between these
two circuits is that in the former bit 5 is
a data bit, while in the latter it is an ad-
dress bit.

With the aid of wire links, a locomotive
address may be set at address inputs Ai-
A»: more about this under ‘construc-

Network Ri-Ci sets the baud rate of the
decoder as required for the locomotives,
while R2-C2 serves to detect the intervals
between the data bytes.

Four of the data bits are fed to the four-
bit comparator; the fifth, which, owing
to Ni, is also available in inverted form,
is used to change over the direction of
travel and the lights via Nj, Nj and the
power stage. The counter with integral
oscillator, IC2, is designed so that the
frequency of the msb (Q<. in this case) is
about 140 Hz, which is also the fre-
quency of the pwm signal (output of
ICj). This frequency was chosen
because it will not cause undue problems
owing to the self-inductance of the
motor: higher frequencies may limit the
motor current.

The power stage, ICs, is a Type L293
from SGS. This chip contains four half-
wave circuits, of which two may be com-
bined into a full-wave bridge for bipolar
motor control. The other two are used
for the head and tail lights; these may
therefore be switched relative to either
earth or the positive supply line.

a certain value, for instance, because the in the circuit diagram of Fig. 16, the Gate N2, Re and R- form the under-
locomotive is travelling over an un- data decoding is carried out by ICi, an voltage detector. If the supply voltage

Fig. 16. The circuit diagram of the locomotive decoder.

piil 1989 4.47

drops below 8 V the potential at junc-
tion Rn-R? is interpreted by Na as a logic
0. This causes the reset input of the
counter to become active, and this
results in the PWM signal going low and
the internal oscillator being stopped,
which limits the current consumption.
At the same time, the remaining inputs
of ICs are made logic 0 via N.i and Ni.
This is necessary to protect the chip
(since its power supply is cut off) and to
limit the current provided by the logic
circuits.

The supply for the logic circuits in ICs
is derived direct from the main power
supply, because these circuits draw a
fairly large current. When the main
supply is present, the potential at junc-
tion R»-R? is limited by the clamping
diodes on board N2 to a value that is
slightly higher than that of the supply
voltage for the logic circuits. If required,
the E2 input of ICs may be connected
direct to this junction. In that case, vir-
tually the full supply voltage is available
for the lights. Since that voltage is fairly
high (20 V), the E2 input is fed with a
280 Hz square wave, which causes the ef-
fective voltage for the lights to be re-
duced to 10 V.

The main supply is obtained by full-wave
rectification in D?-Dio of the a.c. on the
rails. Capacitor Ci ensures continuity of
the supply during the zero crossings of
the rail voltage and blocks the counter
emf generated by the self inductance of
the locomotive motor when this is
switched off (remember, the motor is
pulse driven). A bonus of free-wheeling
diodes Di-Dr. that act in conjunction
with Cj is the virtual elimination of
wheel sparking. This in turn means less
interference in the electronic circuits and
less contamination of wheels and rails.
The lower supply voltage for the logic
circuits is derived via D2, R», Di, and
Cs. This voltage must lie between 3V
and 6.3 V (which is the maximum per-
missible input voltage of ICs). The
value chosen in the present circuit is
5.5 V because that is the rating of the
possibly required external buffer capaci-
tor. If the supply fails, and Cs is re-
tained as the only buffer, the circuit
stores the last received data for about
5 — 10 seconds. This period is determined
primarily by the current drawn by ICi
(25-50 pA) and the leakage current
through Di. The use of an external
buffer capacitor lengthens the period
and this may be essential where
locomotives are used in conjunction
with the Marklin digital system and a
track with conventional block protec-

Two-rail adaptor

Since Marklin uses a three-rail system, it
is always clear which of the connections
in the locomotive are the brown lines

(outer rails) of the system and which the
red line (centre rail). This is not so in
two-rail systems, where the polarity of
the supply lines can be reversed by
reversing the locomotive. This is im-
material with regard to the main power
supply, but it causes complications as far
as the data are concerned.

In the Marklin system, data are present
on all three lines, but those on the red
line are inverted with respect to those on
the brown lines. The two-rail adaptor
determines which is the red line and
which are the brown lines and inverts the
data where necessary.

The circuit diagram of the adaptor is
shown in Fig. 17. Multivibrator MMVi
detects the intervals between the data
bytes. If its input is logic high during the
interval, the connected supply line is
brown, the data on which must be in-
verted. This is done by retriggering
MMV2 at the start of the next data byte
which causes Ni to (continue to) work
as an inverter.

If the supply polarity changes, the red
line is connected to the input of MMVi,
MMV2 is no longer retriggered, and Nj
ceases to invert the data.

If the supply polarity changes at pre-
cisely the time a byte is transmitted, the
data comparator in the locomotive
decoder prevents the acceptance of a
partially inverted byte.

The supply for the two-rail adaptor is
derived from that of the locomotive
decoder.

The construction of the locomotive
decoder and two-rail adaptor will be de-
scribed in next month’s instalment.

Corrections

In the parts list of Part 1 Ti, a

Type BC547 is omitted.

In Fig. 8, bit 9 at the top is jogic 1 and
not logic 0 as indicated.

PRACTICAL FILTER DESIGN (3)

by H. Baggott

A practical filter may be designed in a number of ways. It may
be a passive type, constructed from resistors, capacitors, and
inductors, or it may contain active components that take the
place of inductors. Both these types are considered in this third
part in the series.

The design of a practical filter depends
on the requirements, the application and
the available components. Simple filters
arc normally designed as passive types.
None the less, complex filters may very
well be of the passive type also, although
the size of the necessary inductors is
often a severely limiting factor. Since the
value of inductors for low-frequency
filters is often quite high, the modern
tendency is to use active filters for low-
frequency applications. However, cross-
over filters for use in loudspeaker
systems are often still of the passive type.
In this article a number of filter designs
complete with formulas for their practi-
cal realization will be described. These
considerations will be confined to low-
pass sections, since these form the basis
of all other types. High-pass filters are a
direct derivative of low-pass sections,
while band-pass networks with a fairly
wide response are constructed from a
mix of low-pass and high-pass sections.
Band-pass filters with a narrow response
and all-pass filters will be considered
later in the series.

• One whose input and output ter-
minating impedances are equal (primar-
ily used in h.f. applications). This type
of filter is normally constructed in a n-
or T-shape as shown in Fig. 12. A num-
ber of sections may be simply cascaded
to form a C-L-C-L-C or an L-C-L-C-L
network. Note that Ri is the internal
source resistance.

• One that is connected to a signal
source with negligible internal resistance
and terminated into an impedance Rl.

Fig. 13. Passive filter connected to a source
of negligible internal resistance and ter-
minated in Re.

Passive low-pass sections

Two versions of passive low-pass section
will be considered:

This type, normally constructed in a T-
shape (see Fig. 13), is used primarily in
low-frequency applications. Several sec-
tions may be cascaded to form an I^C-
L-C-L network.

For clarity’s sake, the sections are shown
with an odd number of capacitors and
inductors, although an even number is
perfectly permissible (for instance, one
capacitor and one inductor).

Other combinations of input and output
impedance are, in principle, possible, but
the two versions described here will suf-
fice for the vast majority of passive filter
applications.

The tables, given later in the series, show
the component values for each of the
two versions at a frequency of 1 Hz. The
value of the inductor, L ', at the required
cut-off frequency is calculated from:

L' =LRi/f [7]

Fig. 12. Passive filters with equal inpul and and that of the capacitor, C\ from:
output impedances (Ri = Ri.): (a) n-type; (b)

T-Type C' = C//Rl

Active low-pass sections

Configuration with voltage follower.

The simplest form of active low-pass sec-
tion, shown in Fig. 14, uses an opamp
connected as voltage follower. Amplifi-
cation in the pass-band is unity. This
type of filter should be driven from a
signal source with very low internal im-
pedance. The output impedance of the
Filter is also very low. Fig. 14a shows a
two-pole version (one pair of conjugate
poles), whereas Fig. 14b illustrates a
three-pole type (one pair of conjugate
poles and one real pole). The three-pole
version can be used only in odd-order
layouts in view of its single real pole.

Depending on the required function, a
number of these sections may be cas-
caded. For instance, for a sixth-order
filter three two-pole sections need to be
connected in series; for a fifth-order net-
work, a two-pole section is connected in
series with a three-pole version. A func-
tion requiring an odd number of poles
may also be realized with a number of
two-pole types followed by a passive RC

[ 8 ]

network as shown in Fig. 15a. If the in-
put impedance of the circuit connected
to the filter output is so high that it may
be ignored, a buffer terminating the RC
network is not needed. In other cases the
circuit of Fig. 15b may be used, in which
the amplification of the opamp may be
set with the aid of resistances Ra and
Rn (amplification A = \+Ra/Ru).

Fig. 15. A real pole may be obtained from a
simple RC network (a). The addition of an
opamp (b) enables buffering and amplifi-

The transfer function of the two-pole
filter in Fig. 14a is:

7(/co)=l/[CiC2(/co) 2 +2C2(/aj)+l] [9]

in which all resistors have been given a
value of 1. The value of the two
capacitors as a function of the real and
imaginary part of the complex pair of
poles may be computed from:

Ci = l/2tta [101

Ci = a/2n(a ! + p 1 ) [11]

The transfer function of the three-pole
network in Fig. 14b is:

r(/co)=l/[CiC2C3(/co)’ +

+ 2Cj(Ci+C 2 )(/<d 1 )+
+(C2+3C3)(/co)+11 [12]

In this equation, the values of the
capacitors can not be given simply as a
function of a and p. Their computation
really needs to be done with the aid of a
computer.

In a network with one real pole, the
value of the capacitor is given by:

C = l/2ti« [13]

The values of the capacitors in two- and
three-pole filters with a voltage follower
are calculated from the tables (/"= 1 Hz!)
4.50 elektor India april 1989

at the required cut-off frequencies. This
is done by choosing a value for R and
determining the required cut-off fre-
quency and then calculating C' from

C'=C/fR [14]

When two or more sections are cas-
caded, the value of R need not be the
same for each section, but that of the
frequency must, of course, remain the
same throughout.

If a number of two-pole sections is to be
combined with a section with one real
pole (Fig. 15a or Fig. 15b) to obtain an
odd-order filter, bear in mind that the
tables give capacitor values for the last
two-pole section and the passive section
that arc different from those given for
the three-pole filter.

Fig. 16. A two-pole filter with adjustable
amplification.

Two-pole filter with amplification. The

two-pole section in Fig. 16 offers vari-
able (preset) amplification. The various
components are calculated from:

C 1 =(A+l)(l+/)Va J ) [15]

C 2 =l

Ri = o/2ti4(a J +/) J ) [16]

Rt=<ARi)/<A +1) [17]

R3=ARa [18]

The computation is usually started by
giving an arbitrary (standard) value to
Ci and then calculating the other com-
ponents from the given formulas.

This type of filter, if desired, may be
combined with the other filters de-
scribed earlier. It is, for instance, poss-
ible to create a sixth-order network from
two sections as shown in Fig. 14a and
one as illustrated in Fig. 16.
State-variable filter. In some appli-
cations, the state-variable filter offers
definite advantages over the filters de-
scribed so far. The poles and zeros of
this type of filter can be arranged fairly
accurately, which in other types is next
to impossible owing to the effect the
components have on one another.
Moreover, the bandwidth and amplifi-
cation of the opamp have little effect on
the filter characteristics. This type of

Fig. 17. The state-variable filter is used
mainly in applications where the poles and
zeros must be arranged fairly accurately.

filter can have some degree of amplifi-
cation.

The various components are calculated
from the following formulas:

R i = l/[2ti4Cy(a 2 + p 1 )] [19]

R 2 =l/4noC [20]

7? 3 = 1 /2nCy(a 2 + p 1 ) [21]

R*=R} [22]

If the amplification is unity, i?i=i?3.
Here again, the calculation is started by
giving C an arbitrary (standard) value,
after which the other components can be
computed

To finalize the design, the resonance fre-
quency, fo, and the Q-factor of the filter
are calculated for a real cut-off fre-
quency, /, from:

fo=f]/(.a ! + p ! ) [23]

Q = fo/2fa [24]

The value of R 3 is adjusted to give maxi-
mum voltage at the band-pass output of
the filter (output of Ai) when a signal of
frequency fo is applied to the input.
Also at the output of Ai, the bandwidth
is measured and R 1 adjusted until this
corresponds with that taken for the cal-
culation of the Q-factor ( H=fo/Q ).

It is clear that in the case of a state-
variable filter it is advisable to make R 1
and R 3 a series combination of a fixed
and a multi-turn preset potentiometer.
Next month’s instalment will deal with
high-pass sections and their compu-
tations.

EPROM-CONTROLLED TIME
SWITCH

by J. Vinckier

Time switches can take many shapes. Where apparatus has to be
switched on and off at regular intervals, a simple electronic switch
will suffice. This article describes such a device that is controlled
by a suitably programmed EPROM. It has battery back-up to
ensure continued operation in the case of mains failure.

The circuit consists essentially of three
parts: a crystal-controlled time base, 1C1
and IC2; an address counter, IC4; and
an EPROM, 1C3.

The time basis is controlled by a
32.768 MHz crystal, XI, which is of a
type frequently used in quartz watches
and clocks. The 14-stage counter on
board IC1 divides the oscillator signal to
2 Hz. That signal is fed to a second
counter, IC2, whose Q6 output is con-
nected to address counter IC4. The fre-
quency of the signal at Q6 is 1/64 Hz:
the address counter therefore receives a
pulse every 64 seconds, upon which it in-
creases its content by 1.

The Q0 — Q10 outputs of the address
counter are connected to address lines
A0— A10 of the Type 2732 EPROM.
After 1,350 clock pulses, that is, 24
hours, the address counter must be reset
to 0. For that purpose, data output D7
of IC3 is connected to the reset inputs of
IC1, IC2 and IC3. At address 1350 in the
EPROM, a logic 1 is programmed at bit
position 7, so that when this counter
position is reached, all counters are reset
to 0.

Data lines D0— D3 of the EPROM are
used as control outputs, so that up to
four different apparatuses may be con-
trolled, each at the same or different
times. Depending on the data at memory
positions 0 — 1349 in IC3, these outputs
are logic 0 or logic 1.

To enable manual control of the ap-
paratuses, each output is provided with a
three-position switch, SI— S4. The
switches make it possible for an ap-
paratus to be switched on, switched off,
or to be connected to the time switch.
The on and off switching may also be
done with the aid of relays or electronic
switches (optocoupler with triac). In the
latter case, take great care to ensure good
electrical isolation and a safe construc-

The supply for the switch may be ob-
tained from a mains power supply with
possibly a 5-V regulator. In case of
mains failure, there is a 4.5-V back-up
battery, Bl. In that condition, only the
four ICs remain powered to ensure the

continued running of the clock. How- of IC3 to go low, so that the data are
ever, T1 disconnects the outputs of the present at the EPROM outputs. During
EPROM to limit the total current drawn mains failure, T1 is switched off and the
to a minimum. During normal mains OE input is high,
operation, T1 is switched on via the Spring-loaded switch S5 is provided to
+ 5 V line, which causes the OE input reset the time switch manually at any

elektor india april 19894.51

given time: this has to be borne in mind
during the programming. It is thus, for
instance, possible to arrange the pro-
gramming to start at 8 o’clock in the

morning. Reset switch S5 is then pressed
at 8 o’clock next morning and from then
on the programming will ensure that all
instructions are carried out correctly.

Construction and
programming

The circuit may be constructed on a
piece of prototyping or vero board. Even
the batteries and switches may be
mounted on this board. It is, however,
strongly advised to fit the mains-
carrying parts, including any relays
(mechanical or electronic) on a separate
board. Note, however, that a prototyping
board is not safe for this purpose. The
low-voltage and mains-carrying parts
may be mounted on the same board if
they are electrically well separated by the
removal of some tracks and solder pads
between the two sections.

Programming

It is possible to calculate the switching
times by hand and a simple calculator.
For example,

• start time (address 0) - 0800 h;

• output 1: on 0900 h, off 2100 h;

• output 2: on 2000 h, off 2200 h;

• output 3: on 1830 h, off 2030 h;

• output 4: off 1830 h, on 1930 h.
First tabulate the relative times, that is,
the periods between the start time and
the switching times. The EPROM ad-
dress is then computed by converting the
relative time (RT in the table) into
seconds and divide the result by 64. The
examples above are worked out in the
table. Note that the memory positions
between two addresses must be filled in
with the last stated data, for instance,
the addresses 1 to 55 must be given the
same data as those for address 0. It
should be borne in mind that all ad-
dresses from 1350 onwards must have a
1 at data bit 7 to ensure that the reset
function operates correctly.

NEW PRODUCTS

Accelerometer

THE VII accelerometers from Shinken
Co. of Japan are piezoelectric transduc-
ers which convert mechanical move-
ments into electrical signals proportional
to the acceleration. The accelerometers
feature rigid construction wide fre-
quency range and high linearity and are
used in conjunction with any digital vib-
ration meter and charge amplifier. Nine
models are available in this series. The
charge sensitivity is from 45 pc/G to 26
pc/G. Voltage sensitivity is from 50 mV/
G to 20 mV/G. The maximum accelera-
tion these units can withstand is from 100
G to 1000 G. The temperature ranges
from 70°C to 250°C.

M/s. Murugappa Electronics Ltd. •
Agency Division • 29, Kamraj Avenue
Second Street • Adayar • Madras-600
020 • Phone 41 33 87.'

Digital PH Meter

The Naina solid-state digital pH meter
features electrode socket and in-built
voltage stabiliser. The continuous range
extends from 0 to 14 pH and 0 to ± 1999
mV with relative accuracy of ± 0.01 pH

and ± mV. Options include : recorder
output, automatic temperature compen-
sation, temperature indication and slope
correction.

for further information write to:

M/s. Naina Electronics Pvt. Ltd. • 181/6,
Industrial Area • Chandigarh-160 002 •

DEALING WITH ELECTROMAGNETIC
INTERFERENCE

by Alan Baker, BSc(Eng), ACGI, CEng, FIMechE

Electromagnetic interference (EMI),
almost ignored about ten years ago, has
now become important, primarily
because of the proliferation of electronic
equipment in aerospace, motor vehicles
and other industries.

A feature of such equipment is its sensi-
tivity to electromagnetic emanations in
the environment, and it cannot be com-
pletely shielded from them. Conse-
quently there is widespread interest in
assessing the levels of EMI likely to be
encountered in various circumstances
and in developing valid methods of
testing the susceptibility of different
types of electronic systems.

Laboratories devoted to these activities
have therefore sprung up worldwide,
some for particular industries and others
in university engineering departments.
The latest, for the automotive industry,
is the electromagnetic compatibility
(EMC) laboratory officially opened last
year for the Motor Industry Research
Association (MIRA) at Nuneaton.

This is especially timely in view of the
rapid growth of vehicular electronic ap-
plications such as ignition systems,
engine and transmission management,
adaptive and active suspensions, an-
tilock brakes and four-wheel steering.
The effect of any malfunction caused by
EMI on these items can range from
simple annoyance to a catastrophe.

Europe’s largest laboratory

MIRA’s EMC department is not yet six
years old but its early growth rate was so
rapid that the need for expansion was
already obvious to its then director, Dr
Cedric Ashley, before the end of 1984.
He subsequently gained financial sup-
port from Britain’s Department of Trade
and Industry, the Department of Trans-
port, the Metropolitan Police in Lon-
don, and seven companies (Eaton, Ford,
Jaguar, Lotus, Lucas, SPA and Saab
Scania).

With MIRA’s £500 000 contribution
there was enough money to do the job
properly, resulting in Europe’s largest
laboratory devoted to this particular sec-
tor of automotive technology. The EMC
laboratory now has contracts from firms
in continental Europe, the United States
and the Far East.

Pride of place in the laboratory goes to

a large anechoic chamber made of steel
and measuring 22 m long x 10 m wide
x 7 m high. This is big enough to take
the maximum weight of 38 tonnes for ar-
ticulated trucks, buses or coaches as well
as cars. Most vehicles for testing are of
course cars but the sizes are useful for
them too as accuracy of measurement is
impaired if the walls are too close to the
source of emissions.

In the interest of repeatability over a
broad range of radio frequencies, the
chamber walls and ceiling are lined with
energy absorbing pyramids which are
1.8 m long and made of polyurethane
foam impregnated with carbon.

Regenerative braking system

On the floor of the chamber are two sets
of vehicle-driven aluminium rolls and a
turntable. The rolls drive a large two-
axle electric dynamometer installation
with a continuous absorption capacity
of 150 kW per axle — an input that

demands a highly efficient cooling sys-

The diameter of the rolls is unusually
large at 1.5 m to keep tyre temperature
down to acceptable levels, and their
maximum peripheral speed is 160 km/h.
One of the axles is fixed while the other
can be moved to give any desired
wheelbase between 2 m and 6.5 m.

The dynamometer operating modes in-
clude road load, wheel slip and antilock
braking — the latter with decelerations
as high as 2 G. A lot of energy has to be
dissipated on continuous cycling tests,
despite the relatively low mass of the
rolls, so the dynamometer incorporates a
regenerative braking system.

Braking can be applied to both axles but
it is not normally used on the driving
one because of possible tyre-rating prob-

The dynamometer, made by Brush Elec-
trical Machinery of Loughborough,
enables a wide selection of vehicles to be
driven as though on-the road while being

subjected to a full spectrum of elec-
tromagnetic radiations.

The turntable, also supplied by Brush
Electrical Machinery, has a diameter of
6 m and a load capacity of 10 tonnes
and is used for static investigations, par-
ticularly of vehicle field coupling which
can be fully measured. Bulk current ab-
sorption spectra obtained on this equip-
ment can be used for — among other
purposes — predicting worst case anten-
na positions for susceptibility tests on
the dynamometer, thus saving quite a lot
of valuable time.

Other facilities

Alongside the main chamber are the
control rooms equipped for both manual
and computer controls. MIRA’s
specialists reckon that this combination
of equipment and control systems gives
them full competence to evaluate the
EMC performance of most common
types of vehicle in repeatable and

ELECTRONICS NEWS

Plan Outlay

The Hanning Commission has ap-
proved Rs. 2,725 crores for the tele-
com sector for 1989-90. The com-
mission has been soft on telecom for it
effected only a marginal cut on the
projected demand of Rs. 2,775
crores.

The DOT is expected to meet 75 per
cent of its expenditure through inter-
nal resources. The budgetary support
is only about Rs. 225 crores. the de-
partment has sought to raise an addi-
tional Rs. 300 crores by issuing
bonds and the Planning Commission
has supported the DOTs plea.

The 1989-90 plan envisages commis-
sioning of five lakh new phone con-
nections against the four lakhs for
1988-89. Against 69 cities connected
by STD in 1988-89, next year 113
cities will come under the STD net-

Indo-Soviet PC

The Soviet Union is seeking Indian
technology for the manufacture of
LBM compatible personal comput-
ers. It has agreed to set up a joint ven-
ture with India for the manufacture of
these computers.

The Soviet Union may buy back
100,000 computers annually from
the proposed unit There is a large de-
mand for PCs in the Soviet Union

4.54 elektor India april 1989

realistic operating conditions over a fre-
quency range from 10 kHz to 1 GHz and
with field strengths of up to 200 V/m at
1 m distance.

Outside the laboratory building is a site
for measuring whole vehicle radio fre-
quency emissions in actual field con-
ditions to meet statutory requirements.
There are also indoor facilities for
assessing the susceptibility of electronic
sub-systems.

One of these facilities is for bulk current
injection, carried out by clamping on
electrodes and thereby inducing high fre-
quency currents in the wiring loom, on
or off the vehicle, so that their effects on
the electrical equipment can be studied.
A highly adaptable routine has been
developed whereby the testing can take
in a single conductor, a group of them or
the entire harness.

The susceptibility of electronic equip-
ment to transverse electromagnetic mode
(TEM) radiation can be measured on a
standard stripline over a frequency range

which are manufactured on a big
scale. However, the Soviet Union is
keen on switching over to the IBM
compatible because of their universal
acceptability.

The decision to set up a joint venture
unit for the small computers was
taken at a meeting of the Indo-Soviet
working group on electronics and
computers. The working group also
envisaged collaboration in other fields
including manufacture of electronics
components and software. The Soviet
Union is keen on exporting ICs to
India It has agreed to undertake an
agressive marketing drive popularise
the Soviet made ICs. In the next few'
months, the Soviet Union will hold a
series of seminars in India relating to
the ICs.

VCR-VCP Technology

India is exploring the possibility of ob-
taining the technology for VCR-VCP
manufacture in the country through
small Japanese companies.

Japanese multinationals have been
advised not to transfer the “head
drum” technology to any country and
even the sale of VCRs in CKD condi-
tions is being stopped. However,
small component manufacturers,
who are the captive suppliers of big
Japanese firms, are believed to be
exempt from this restriction on
technology transfer.

In India, the Electronics Trade and
Technology Development Corpora-
tion (ET & T) has been allowed to
have a collaboration with the Tohei of

from 10 kHz to 400 MHz at field
strengths over 20 V/m. For small objects
this apparatus is complemented by a
TEM cell with a lower maximum fre-
quency but the capability for con-
siderably higher field strengths of up to
1000 V/m.

Last of the laboratory’s facilities is a
relatively small non-anechoic screened
chamber measuring 5 m long x 3 m
wide x 2.3 m high. It is used for giving
radiated susceptibility and emission tests
to components and sub-systems by the
special techniques defined in various
European and United States military
standards.

Generally, therefore, the MIRA
laboratory does not only offer a
capability test but provides the world’s
motor industry with a problem-solving
service, ranging from the evaluation of
circuit design and the rectification of
faults to the complete design of equip-
ment and its final electromagnetic com-
patibility testing.

Japan. Tohei is a manufacturer of
VCR components, including the tape
deck mechanism. In the 1970s, the
company supplied the decks for over
three million VCRs manufactured by
Hitachi. The company will manufac-
ture the tape deck mechanisms for
India with a promise of technology
transfer, it is reported.

ET & T proposal envisages 25 per
cent indigenisation from the very be-
ginning while the other private man-
fuacturers have agreed to reach 70 per
cent indigenisation in five years.

The large Japanese companies are
just assemblers of components made
by small, captive component man-
ufacturers and these small units pos-
sess the technology. Tohei, being a
small sector company, can transfer
the technology without much diffi-
culty, it is believed. ET & T hopes to
reach the international prices ofVCRs
and VCPs in two or three years.

Tohei is one of the six companies ear-
lier short listed for collaboration. The
company has already been haring a tie
up with the Calcutta firm, Sonodyne-
Indian Fine Bank, for making cutting
dies and material processing machin-
ery. Sonodyne-IFB will be nominated
suppliers for key parts of VCRs and
VCPs produced by the ET & T and
ECIL, joint venture. The company
would be allowed to assemble the
VCRs manufactured by ET & T.

RECORDING/PLAYRACK

AMPLIFIER

This amplifier enables accompanying slide presentations with
stereo commentary or music whilst using a separate track on the
cassette recorder for controlling the slide changer.

This simple to build amplifier
makes use of a 4-track re-
cording/playback head fitted
in lieu of the 2-track head in an
ordinary cassette recorder
used for audio-visual presen-
tations. Four-track heads can be
picked up occasionally from
surplus outlets, or purchased as
a servicing part for auto-reverse
cassette players. The proposed
signal assignment on the
4-track head is shown in Fig. 1.
The non-used track in between
those assigned to the stereo
programme and the control
signal ensures acceptable
levels of crosstalk. The pro-
gramme and the slide control
signal can be recorded separ-
ately. If required, the control
pulses can be erased by revers-
ing the cassette and recording
silence.

Circuit description

The circuit diagram in Fig. 2
shows that the recording/
playback amplifier is set up
around the Type TDA1002A in-
tegrated amplifier from
Mullard. Although this compo-
nent is not aimed at the high
quality market, its technical
qualities are still adequate for
use with most types of (semi-
portable) stereo cassette
recorders or decks commonly
available, The TDA1002A re-
quires relatively few external
components to make a versatile
recording/playback amplifier.
As shown in Jhe circuit
diagram, the chip comprises a
preamplifier and a recording
amplifier with automatic level
control (ALC). The control
range of the ALC circuit is
stated as SO dB +2 dB.

When 4-pole toggle switch
Si is set to record as shown
in Fig. 2a, the slide control
pulses are applied to pin 1 of

the TDA1002A via switch sec-
tion c and terminal C. Also, the
record LED lights (section a),
and one side of the symmetrical
output of the record /playback
head is grounded. Since ter-
minal D is connected to E (sec-
tion d), the preamplifier works
with feedback circuit C4-R3
connected in series between
output pin 4 and input pin 2.
The amplified signal is available
at the output of the circuit, but it
is also fed to the input of the re-
cording amplifier (pin 8) via
Rs-Ci, and the input of the ALC
circuit (pin 6) via R«. The input
of the recording amplifier is
held at a fixed potential with the
aid of voltage divider R11-R10-R9.
The negative feedback network
between output pin 9 and input
pin 7 is composed of R12-R1S
incl. and Cu-Cm incl. The ALC
circuit acts on the output of the
recording amplifier via
R17-R11-C9. The limiting time
(10 ms typ.) and the recovery
time (35 s typ.) are fixed with C is
and C16-R19 respectively. The
amplified and limited re-
cording signal is fed to the head
via Cio-Rie, terminal A, switch
section b and Ri.

When Si is set to the play mode,

terminal A is grounded via
switch contact b, LED Di is ex-
tinguished, and the playback
signal from the head is fed to
the input of the TDA1002A.
Switch section d selects
negative feedback network
Rs-Cs-Rs between the output
and the input of the preampli-
fier, which is so dimensioned to
give the appropriate gain. Note
that the ALC and the recording
amplifier are ineffective in the
playback mode. The output
signal can be fed to the avail-
able slide change circuit.

The circuit is fed from a
regulated 10 V supply set up
with IC2. The unregulated input
voltage should not exceed
about 15 V. It may be necessary
to redimension R20 in accord-
ance with the regulated or
unregulated voltage available in
the cassette recorder.

Construction and
setting up

The layout for the printed cir-
cuit board for building the re-
cording/playback amplifier is
shown in Fig. 3. Note that the
board does not hold the parts
shown in Fig. 2a. It is rec-

ommended to fit the completed
PCB at a suitable location in the
cassette recorder. Most types of
low cost, modern cassette deck
have plenty of space inside to
house a small additional PCB,
and generally do not present
problems as regards the supply
current for the proposed ampli-
fier. The 4-pole toggle switch.
Si, and the LED, Di , are fitted on
the front panel.

Carefully remove the existing
stereo head, and make sure to
have connection data of the
4-track head available before
this is fitted. In some cases, the
new head requires a few
modifications to be carried out
on the existing head mounting
assembly, and possibly to the
azimuth adjustments also. Refer
to Fig. 1 and connect the two re-
cording/playback channels of
the head to the existing cir-
cuitry in the recorder. Connect
the head section for the slide
control signal to R1-R2 (these are
fitted direct onto Sib) via a short
length of shielded, symmetrical
wire, grounded centrally at ter-
minal B as shown in Fig, 2a. It
should be noted that the equal-
ization characteristics of the
TDA1002A are dimensioned for
Fe cassettes only. The total gain
of the device is about 40' dB at
an average distortion of 0.5%.
The maximum input and output
voltage are 20 rnVrms and 2 Vims
respectively. The input im-
pedance is about 16 kS. De-
pending on the sensitivity of
the head used, the stated value
of Ri6 may have to be slightly
altered to ensure the amplitude
of the recording signal.
Resistors R3 and R* determine
the gain of the playback ampli-
fier. Low frequency oscillation
may occur when the amplifier is
dimensioned for a relatively
low gain.

Test the newly installed 4-track

head by first playing a pre-
recorded stereo music cassette,
or, if available, a test cassette.
Adjust the head until the quality
of the playback signals is ac-
ceptable. Rewind the tape to
the start, and replay it while
using the recording/playback
amplifier for recording the
slide control signal on the third
track. Stop the tape, rewind it,
and check whether all signals
are played back at acceptable
levels of distortion and
crosstalk. Slide change signals
of a wide amplitude range can
be recorded without causing
tape saturation, thanks to the
ALC circuit in the recording/
playback amplifier. R

Parts list
Resistors (±5%):

Rr = 100R
Rb;Rii = 33K
R. = 100K
Rh- 68K
R.t=2K7
Ru = 180R
Ri«;R,s.27K
R..-2K2
Rtr;Ri. = 10K
R«=1M0
R» = 1K5
Capacitors:

Ci:C<;Ci;Ci;Cu>= lOp; 16 V
Ci;Ci< =3n9
Cj- 100p;16 V; axial
C»:Cir = 10n
Cr - lOOn

Ci = 47p; 16 V; axial
Cn 9 220p

Cu=22p; 16 V: axial
C.i = 15n

Cii=4p7: 16 V: axial
Ci«=220|i; 16 V; axial
Cia = 330n
Ci9 = 820p
Semiconductors:

Di= red LEO

ICi=TDA1002A - *

ICt=78L10'

Miscellaneous:

Si = miniature 4-pole toggl

4-track recording/playback
for cassette recorder.

tg/playback amplifier.

an infuriating

electronic
nuisance

Practical jokers will want to hide the
circuit in such a way that it will take
some time to find it. For this reason, it
must be small; furthermore, it will have
to be battery-powered — a mains cable
would be a dead give-away. The circuit
described here fulfils both requirements;
it fits or. a small p.c. board and is
powered by a small 9 V battery.

The light sensor is an LDR. In the dark,
its resistance is quite high; preset
potentiometer PI is adjusted so that the
inputs of the CMOS gate N1 are just at
logic zero under these conditions. The
calibration procedure will be described
later.

The two CMOS gates. N1 and N2, are
connected as a 'trigger' circuit. When
the voltage at the inputs of N1 falls
below the trigger threshold, the output
of N2 switches to logic zero. Transistor
T1 is turned off. and Cl can now charge
up through R5.

The voltage across Cl rises so slowly
that it takes a few minutes for it to
reach the upper trigger threshold of the
second trigger circuit, N3 and N4. At
that point, the output of N4 swings up
to logic one - i.e. practically the full
supply voltage. This takes the reset
input of the 555 timer (IC2) high,
enabling this 1C. The 555 is used in an
oscillator circuit, driving a loudspeaker,
so that an irritating tone is produced.

Have you ever been kept awake by a cricket? You switch off the light
and snuggle down, and just as you're drifting off to sleep the insect
starts to make an irritating noise. As soon as you switch on the light to
look for it, it stops again. Tracking down this type of noisy nocturnal
nuisance can be infuriatingly time-consuming. The same result can be
obtained electronically. What's the point? Well, just for the fun of it.

victim turns on the light to
for the source of the noise, the
resistance of the LDR decreases sharply.
The trigger circuit (N1/N2) changes
state, turning on T1. Cl discharges
rapidly through R4, the output of the
second trigger circuit goes 'low' and the
oscillator is turned off.

When the light is switched off again, the
circuit again waits a few minutes before
making a noise. Very infuriating . . .

Calibration

Preset potentiometer PI must be
adjusted so that the inputs of N1 are at
logic zero when the circuit is in the
dark. The easiest way to do this is to
connect a voltmeter to the output of
N2. First. PI is adjusted so that this
output swings up to nearly full supply

1 4.57

voltage; then PI is turned back until the
output switches to the 'low' level
(practically 0 V) - with the LDR in the
dark, of course. This completes the
calibration.

The time delay, from the moment the
light is turned off to the first squeak
from the oscillator, can be modified
according to personal taste by altering
the value of Cl. In the same way, a
different frequency can be obtained by
selecting a different value for C2. The
ratio of resistor R9 to RIO determines
the type of sound obtained.

Finally, the sound level depends on R8.
Note, however, that this resistor should
not be less than 100 f2. Any loudspeaker
impedance from 4 n up can be used;
the higher the impedance, the louder
the output. M

SMALL 8 BIT A-TO-D CONVERTER

The TLC648 and TLCS49 from Texas In-
struments are each complete data ac-
quisition systems on a single chip. Each
contains an internal system clock,
sample-and-hold, 8-bit A/D converter,
data register, and control logic circuitry.
For flexibility and access speed, there
are two control inputs: I/O Clock and
Chip Select (CS). These control inputs
and a TT-fr-compatible three-state output
facilitate serial communications with a
microprocessor. A conversion can be
completed in 17 jjs or less, while com-
plete input-conversion-output cycles
can be repeated in 22 ^s for the TLC548
and in 25 ps for the TLC549.

The internal system clock and I/O clock
are used independently and do not re-
quire any special speed or phase rela-
tionships between them. This in-
dependence simplifies the hardware
and software control tasks for the
device. Due to this independence and
the internal generation of the system
clock, the control hardware and soft-
ware need only be concerned with
reading the previous conversion result
and starting the conversion by using the
I/O clock. In this manner, the internal
system ' clock drives the 'conversion
crunching’ circuitry so that the control
hardware and software need not be con-
cerned with this task.

When CS is high, the data output pin is
in a high-impedance condition and the
I/O clock pin is disabled. This CS con-
trol function allows the I/O Clock pin to
share the same control logic point with
its counterpart pin when additional
TLC548 and TLC549 devices are used.
This also serves to minimize the re-
quired control logic pins when using
multiple TLC548 and TLC549 devices.

The control sequence has been de-
signed to minimize the time and effort
required to initiate conversion and ob-
tain the conversion result. A normal con-
trol sequence is:

1. CS is brought low. To minimize errors
caused by noise at the CS input, the in-
ternal circuitry waits for two rising
edges and the a falling edge of the inter-
nal system clock after a CS1 before the
transition is recognized. However, upon
a CS rising edge, DATA OUT will go to
a high-impedance state within the tdis
specification even though the rest of the
IC’s circuitry will not recognize the tran-
sition until the tsuicsi specification has
lapsed. This technique is used to pro-
tect the device against noise when used
in a noisy environment. The most signifi-
cant bit (MSB) of the previous conver-
sion result will initially appear on the
DATA OUT pin when CS goes low.

2. The falling edges of the first four I/O
clock cycles shift out the 2nd, 3rd, 4th

and 5th most significant bits of the
previous conversion result. The on-chip
sample-and-hold begins sampling the
analog input after the 4th high-to-low
transition of the I/O Clock. The sam-
pling operation basically involves the
charging of internal capacitors to the
level of the analog input voltage.

3. Three more I/O clock cycles are the
applied to the I/O pin and the 6th, 7th

and 8th conversion bits are shifted out
on the falling edges of these clock
cycles.

4. The final, (the 8th), clock cycle is ap-
plied to the I/O clock pin. The on-chip

sample-and-hold begins the hold func-
tion upon the high-to-low transition of
this clock cycle. The hold function will
continue for the next four internal clock
cycles, after which the holding function

terminates and the conversion is per-
formed during the next 32 system clock
cycles, giving a total of 36 cycles, After
the 8th I/O clock cycle, CS must go high
or the I/O clock must remain low for at
least 36 internal system clock cycles to
allow for the completionof the hold and
conversion functions. CS can be kept
A new conversion may be started and
the ongoing conversion simultaneously
aborted by performing steps 1 through
4 before the 36 internal system clock
cycles'occur. Such action will yield the
conversion result of the previous con-
version and not the ongoing conversion.
For certain applications, it is necessary
to start conversion at a specific time.
This is achieved by stopping the I/O
clock after the leading edge of the 8th
pulse. Conversion is started by making
the I/O clock low at the desired time.
During the period when the I/O clock is
high, the sample-and-hold .will continue
to sample the analogue input.

.4.59

10 REM COMPUB0ARO TEST PROGRAM FOR 8 BIT AO-CONUERTER TLCS4B/549
20 REM

30 PORT’ = 0FDH: REM CS = I , clock -

40 REM HA I HL OOP *

50 DO: REM a ° Forever

G0 G0SU6 1000
70 PRINT -AD-value • ' .

80 PRINT US1NGOII * , VALUE, CR .

000 REM READ CONVERTER
010 PORT I = 0F9H:

020 VALUE ‘ 0

030 FOR BITCNT - 8 TO 7

040 VALUE - VALUE *2 > . OR . t PORT I . AND. 01

050 PORT’ - CFBH: PORT I • 0F9H:

050 NE>.T BITCNT
070 FORT I - 0FDH:

080 RETURN

low during periods of multiple conver-
sion. When keeping CS low .during
periods of multiple conversion, special
care must be exercised to prevent noise
glitches on the I/O Clock line. If glit-
ches occur on the I/O Clock line, the
I/O sequence between the micropro-
cessor/controller and the device will
lose synchronisation. If CS is taken high,
it must remain high until the end of con-
version. Otherwise, a valid high-to-low
transition of CS will cause a reset con-
dition, which will terminate the conver-
sion in progress.

The circuit shows how two converters
may be connected to a Type 8051 or
Type 8048 controller. It is, however, also
possible to connect fewer or more than
two converters.

The program shown is a test aid for con-
verters connected to a Type 8051. If the
ports of the 8051 shown are already oc-
cupied, or if something else has been
connected to the other lines of port 1,
the program must be suitably altered
(the highest bits of PI must be logic
high).

The supply voltage is used as the refer-
ence voltage, which obviates the use of
external components. The differential
reference input does, however, offer full
freedom in choosing a different refer-
ence voltage.

The total conversion error introdu
by the converter is 4^0.5 LSB w
U:cl=5 V.

The current drawn by the IC is 3

CALSOD, A LOUDSPEAKER
DESIGN PACKAGE

Until recently, computer-aided loudspeaker design and optimizing
could only be implemented on mainframes. Fortunately that has
changed, and a new, comprehensive, design package, CALSOD,
is now available for PCs as well. This article reviews CALSOD, and
reports on its use in a practical test.

Designing a good-quality loudspeaker
box invariably requires solid background
knowledge, a lot of time, and reliable
test equipment. If any one of these three
ingredients is lacking, the final design
will almost certainly fail to give satisfac-
tory results. Serious designers will no
doubt have the relevant equipment and
background knowledge, but often lack
time to go through the stages of testing
and redesigning the box. The design of a
multi-way loudspeaker system invariably
commences with setting up a theoretical
model on the basis of available data on
the drive units to be used. Next, a proto-
type is built to clear the way for practical
tests. Measurement results generally de-
viate widely from those expected on the
basis of the calculations. This is so
because it is very hard, if not impossible,
to include each and every parameter in
the calculations. Filter response, im-
pedance, frequency and phase
characteristics of the drive units are all
fairly simple to determine on their own,
but complex calculations in simulation
programs are required to predict their
combined effect, leading up to the total
response of the filter and drive units.

Unfortunately, the development and use
of such simulation programs is the ex-
clusive domain of leading loudspeaker
manufacturers. Not only the software in-
vestment, but also the mainframes used
for running these programs are well out
of reach of individual box designers and
small companies. The arrival of
CALSOD has changed this radically.
Other software packages for loudspeaker
design and optimizing, offering a
price/performancc trade-off at least
equal to that of CALSOD, are, to the
best of our knowledge, not available.

Computer-aided design

CALSOD stands for Computer-Aided
LoudSpeaker Optimizing and Design.
Although ‘Design’ would normally
precede ‘Optimizing’, the acronym
covers the function of the package very
well. A series of extensive tests with
CALSOD has spurred our enthusiasm
about the program. The redesign feature
of the program was tested on existing
loudspeaker systems. Remarkably,
CALSOD’s computed response was

found to correspond exactly with the
measured response.

CALSOD is actually a set of -sub-
programs that together offer the
possibilty to calculate everything a
designer needs to know to achieve opti-
mum results from the available drive
units, whose technical characteristics are
first entered in the program (impedance,
frequency and phase response;
Thiele/Small parameters, if available).
Obviously, accuracy of the computed
results is determined to a large extent by
the accuracy of the input data. A num-
ber of fairly simple program modules
then allow converting the measured
curves into a kind of equation used for
processing by the program. Examples of
available filter modules include one
capable of generating a second to fifth-
order Butterworth characteristic, one
representing the response of a drive unit
in a closed box, a bass reflex box, a
passive radiator, and so on. Small ir-
regularities in a response curve can be
simulated accurately with the aid of so-
called ‘minimum phase equalizers’,
which are essentially tuned circuits
whose resonance frequency, Q (quality-)

factor, and amplification or attenuation
can be specified by the user.

After the curves have been simulated
with the aid of modules, these can be
‘fitted with’ the appropriate filters. All
data is put into a text file that looks
similar to a netlist for SPICE. The inte-
grated word processor is then used for
making a file for each loudspeaker. The
file contains the component values, and
the way components are connected to
nodes in the network. Global values can
be entered for filter specifications, e.g.,
representing an ideal filter terminated in
a pure resistance.

Next, the target response curve is speci-
fied, e.g., that of a fourth-order
Linkwitz filter dimensioned for a cut-off
frequency of 5 kHz. The file with all
data is then read into the program, after
which network analysis is performed.
The user is then in a position to study all
the relevant parameters: frequency and
impedance characteristics of the box,
output voltage of the filter, input im-
pedance of the loudspeaker(s) plus filter,
and the acoustic output signal of the box
plus filter. The target response curve can
be projected over the measured response,
so that deviations can be assessed before
the optimizing process commences.
CALSOD changes component values in
the filter until the acoustic output signal
is a reasonable approximation of the
target specification. The user is in a pos-
ition to state beforehand which compo-
nents may be redimensioned by the
program. All loudspeaker sections arc
processed in this way to obtain a larger
file that contains optimized data for all
sections.

The complete system is then ready for
analyzing. Individual curves can be
displayed separately, as well as the sum
signal produced by the loudspeakers,
measured at a predefined distance from
the box. CALSOD even offers the possi-
bility to indicate vertical and horizontal
position of the loudspeakers on the front

panel of the box, as well as inter-
loudspeaker distance relative to the
listening position. This facility allows
studying the effect of, say, a 3 cm
displacement of the tweeter, or a 10 cm
displacement of the listener. Finally,
CALSOD is a capable of optimizing the
complete system, working effectively
towards the realization of the target
response.

Practical and with plenty of
options

CALSOD is a well-designed and
remarkably practical program that will
prove invaluable to the designer who
knows what he is doing. Evidently, the
program is and remains but a tool that
works on the basis of the user’s ex-
perience gathered from previous loud-
speaker designs. None the less, this tool
greatly simplifies formerly often tedious
and time-consuming work. The optimiz-

ing procedure can provide really good
results, and the options for analysing a
complete system are unique. On a less
positive note, the program is fairly
cumbersome to work with. As in SPICE,
changing a single value in the input file
is basically simple, but time-consuming.
Before a new analysis can be performed,
the user must return to the word pro-
cessor, change the text where appro-
priate, load the modified file into
CALSOD, and restart the analysis.
Remarkable in view of the fairly heavy
calculation load, the review package of
CALSOD did not support the use of a
maths co-processor in the PC.

1 4.61

“BATTERY LOW” INDICATOR

by J. Ruffell

Today there are innumerable pieces of equipment that are
powered by batteries, both dry and rechargeable. In many cases,
it is difficult to determine whether those batteries are still fresh or
fully charged or if they need replacing or recharging. Here is a
small circuit that monitors the battery voltage and gives an
audible warning when that voltage becomes too low.

The indicator described here is small
enough to enable it being Fitted inside
the battery-operated equipment, such as
a portable shaver or receiver. It draws a
current of not more than 1 mA, so that
it does not noticeably increase the load
on the battery.

Circuit description

The circuit is based on two opamps that
are housed in Type TLC272 chip.
Opamp Al, connected as a comparator,
compares the battery voltage, applied to
the inverting input via potential divider
R1-(R3+P1), with a reference voltage of
about 4.7 V that is applied to the non-
inverting input. Owing to the low zener
current, the reference voltage is not
always exactly 4.7 V. However, when the
battery voltage drops, the potential at
the inverting input decreases much more
rapidly than that at the non-inverting
one, so that the comparator always
toggles at the same battery voltage. That
voltage may be set very accurately by PI.
When the battery voltage is at a normal
level, the potential at the inverting input
of Al exceeds the zener voltage. The out-
put of the comparator is then virtually
nought. When the zener voltage exceeds
the voltage across R3 + P1, the compara-
tor toggles, which causes the level at its
output to rise to that of the battery
voltage. Capacitor C2 is then charged
slowly via R5. The potential across the
capacitor (at the inverting input of com-
parator A2) is compared by opamp A2
with the voltage at its non-inverting in-
put. Because of the feedback via R7,
that voltage does not have a fixed value,
but that does not matter in this circuit.
When the potential across C2 has at-
tained a value that is higher than that of
the voltage at the non-inverting input of
A2, the output of this opamp goes low.
Darlington Tl, and consequently buzzer
Bzl, is then switched on. The buzzer is a
d.c. type with built-in oscillator. In this
condition, the potential at the non-
inverting input of A2 is pulled down a
few volts via R7; in other words, there is
a degree of hysteresis. Because of that,
4.62 elektor india aptll 1989

the buzzer will continue to draw current
from C2 until the potential across the ca-
pacitor (and thus at the inverting input
of A2) has decreased by a few volts. The
comparator then toggles so that its out-
put goes high, which renders the buzzer
inactive. From then on C2 charges again
and the process repeats itself until the
quipment is switched off and the battery
is replaced or recharged.

The indicator is suitable for use with
battery voltages between 4.5 V and 15 V.
When a Type TLC272 IC is used, the cir-
cuit draws just under 1 mA. Use of a
Type TLC27L2 reduces this to 250 nA at
9 V.

Construction

Since the whole circuit consists of only
15 components, it is easily constructed
on a small piece of prototype or vero

board. The shape of this should be
adapted to the space available in
whatever equipment the indicator is to
be used.

The circuit is preset as follows. Assum-
ing that the battery voltage is 9 V, the
buzzer should start operating at about
7 V. Connect a regulated, variable power
supply to the circuit and set its output to
precisely 7 V. Turn PI to maximum re-
sistance. With a multimeter, measure the
voltage at the output of Al (test point
A): this should be virtually nought.
Slowly turn PI until the output voltage
of Al suddenly rises to 7 V: this is the
correct setting of PI. Within a few
seconds, the buzzer should sound.

The indicator can then be fitted into the
relevant equipment. Its battery connec-
tions should be soldered to suitable take-
off points behind the on-off switch.

To put your mind at rest: the title does not imply that the circuit
described here enables a computer to see. But if you want to use
your computer for controlling external equipment without connecting
this direct to the computer, the proposed circuit will 'keep an eye' on
certain output signals of the computer and on that basis switch the
equipment on and off. In other words, it provides an optical coupling
between the computer and the equipment to be controlled. This does
imply, of course, that a monitor screen is available and that the
computer has some graphics facilities. Otherwise there would not be
much to see for the eye!

computer eye

control by
monitor screen

The circuit is based on an opto-electronic
comparator as shown in figure 1. The 'eye'
proper is formed by two light-dependent
resistors — LDRs — R1 and R2. The
voltage level at their junction is applied to
the inverting input of the comparator, IC1,
via R4. The non-inverting input of 101 is
held at a fixed reference voltage. The
comparator toggles when the level at its
pin 2 is lower than the reference voltage.
Transistor T1 is then on, and the relay is
actuated. At the same time, T2 conducts,
so that the LED, Dl, lights to indicate the
state of the circuit visually.

When the level at the inverting input of
the comparator is higher than the refer-
ence voltage, the relay is not energized,
and Dl is out.

The idea is that the control program
includes instructions which cause two

light areas to appear on the monitor
screen as required. The intensity of one of
these areas should be constant, while that
of the second should be either low or
high (dark or light). The preferred mode
of operation is for the second area to be
dark when the external equipment should
be switched on, and bright when it is to
be switched off.

The LDRs should be attached to the moni-
tor screen where the two light areas
appear. The voltage (about 2 Vpp) at the
junction of these resistors is a measure of
the difference in brightness between the
two light areas on the screen. Super-
imposed on this voltage is, of course, the
sawtooth voltage produced by the SO Hz
line scan oscillator. Resistor R4 and
capacitor Cl, and to some extent PI,
ensure that this sawtooth voltage does not

1

affect the correct operation of the com-
parator.

Construction of the circuit is not critical:
all components, except the LDRs, are fit-
ted on a small prototyping board. The
LDRs are connected to this board by suffi-
ciently long pieces of stranded equipment
wire. It is recommended to fit them in
suitable shrink sleeves or swathe them in
insulating tape in such a way that only the
light of the two areas on the screen falls
onto them (see photograph). They can be
attached to the screen with some self-
adhesive tape. If the equipment to be con-
trolled is switched on when it should be
switched off, and vice versa, simply inter-
change the LDRs.

Presetting of the comparator is not critical
as long as the change-over frequency of
the two light areas is of the order of 1 Hz.
In that case, PI is simply set so that the
relay is actuated and de-energized in
rhythm with the change-over frequency.
When that frequency is higher, eg. when
the circuit is used for data transfer, the
presetting of PI becomes more critical.

The maximum allowable change-over fre-
quency depends on the cut-off frequency
of the low-pass filter, R4/C1, which here is
less than 10 Hz. Optimum setting of PI is
then best achieved by applying a square-
wave voltage at a frequency of about 8 Hz
to the comparator input. Measure the out-
put at pin 6 with an analogue voltmeter
(10 V d.c. range) and adjust PI so that this
level is half the value of the supply
voltage. Although the pointer of the
voltmeter quivers somewhat, the setting
can be carried out without any trouble. If
you have an oscilloscope, it is, of course,
preferable to use that for the presetting.
Note that the current through the relay
coil should not be too high: when a
BC 547 is used for Tl, it should not exceed

100 mA. That means that the resistance of
the coil should be not less than 50 S for a
supply voltage of 5 V, and not less than
90 Q at 9 V. The rating of the relay contacts
depends on the equipment to be con-
trolled.

Current consumption of the circuit
amounts to only a few mA plus the current
drawn by the relay coil. For data transfer
operation only, the relay is not required:
the signals are then taken direct from the
collector of Tl. K

Did you know. . .?

Robot has come to mean an intelligent
and obedient but impersonal machine;
it is derived from the Czech robota —
forced labour. The word robot was first
used in Karel Capek's play Rossum's
Universal Robots (1920).

(OED)

Gain is a ratio, normally expressed in dE
For an amplifier it is the ratio of output
power to input power; for an aerial, it is
the ratio of the voltage produced by a
signal entering along the path of greatest
sensitivity to that produced by the same
signal entering an omnidirectional aerial.
Although often used as such, it is not
synonymous with amplification, which is a
number indicating by how many times an
electronic device increases an electrical
signal. Gain is, therefore, 10 or 20 times
the logarithm of the amplification, depen-
ding on whether that refers to a power or
a voltage increase. H

NEW PRODUCTS

Coil Winding Machine

Arsun Engineer offer hand operated coil
winding machine for winding a variety of
coils, using 0.913 to 0.06 mm (20 to 46
SWG) enamelled copper wires. Any
type of bobbin square, round or rectan-
gular can be would on the machine. The
machines sturdy in construction having
mechanical cast iron, brass plate pinion
and bevels gears of steel. The shafts are
mounted on bearings for smooth running
of the machine. The wire from the real is
guided by a pulley on reel carrier and two
pulleys fitted on the carriage, which is
guided on two inverted V ways for high
accuracy. The traverse of whole carriage
assembly is by means of lead screw and
friction mechanism and its direction is
reversed manually at the end of each
layer by a movement of lever, while the
machine is in running condition.

A quick reset type six digit revolution
counter is provided for counting the
number of turns. This machine is suita-
ble for regular productions as well as
small quantity prototype developments,
repair shops, technical Institutions and
maintenance departments.

M/s. Arsun Engineers • S6/1, Vithalwadi
Industrial Estate • Bhavnagar-364 001
(Gujarat).

Digital Vibration Meter

The digital vibration meter V-l 103 from
Shiken Co of Japan has a charge
amplifier input circuit. It offers shock ac-
celeration measurements using Peak
Hold performance and random vibration
measurements using true RMS calcula-
tions, adding to general vibration mea-
surements of acceleration, velocity and
displacement.

Provisions are made for low cut and high
cut filters, and AC DC and digital out-
puts. As it employs a two-way power
supply system of AC and car battery sup-
ply, the instrument is suitable for
laboratory or field use.

M/s. Murugappa Electronics Ltd. •
Agency Division • 29, llnd street •
Kamaraj Avenue • Adyar • Madras 600
020 • Phone: 41 33 87 •

Electronically Temperature
Controlled Soldering Station.

Reliance Electronics offer an electroni-
cally controlled soldering station that
controls and monitors the temperature
of the tip of the soldering iron to the pre-
determined set value in the neighbour-
hood of ± 5%. The instrument utilises
thyristor power control and the tempera-
ture range can be adjusted from 170±C
to 450°C. Protection of the tip from vol-
tage spikes and magnetic fields is en-

Three-degit LED is provided to know
the tip temperature. Compact size,
flame-proof skeeving of the solderiing
iron, and lightweight soldering iron with
iron-coated tips are the features.
Operating input voltage can be 230 VAC
± 20% , and output voltage power can be

M/s. MRK & Brothers Engineers. • 310
A, Commerce House • Nagindas Master
Road • Fort • Bombay-400 023 •

up Based Digital Printer

Hoshakun have developed a microp-
rocessor based digital printer in two
types, one to accept only one input and
the other up to 16 inputs. A normal 18
column numeric printer is used to print
the date received from outside paramet-
ers like temperature pressure, pH, etc. It
accepts the signal in terms of D.C. mV/
mA current, any thermocouple or any
PRT bulb input. A built-in 5— digit red
LED indicates either the measured
parameter like °C Dc or the real time
clock. The software is user-friendly i.e.
after the instrument is switches on, a
series of questions will appear on the dis-
play. the user is supposed to enter the
data, in response to the questions asked
by the instrument. For the thermocouple
input the linearisation is a standard fea-
ture. Printing interval is programmable
from 01 minute to 99 minutes. Battery
back-up can be supplied optionally. In
case of multi-point, the user can state the
type of input and the range for each of
the input and this can be programmed.

M/s. Hoshakun • Vivek Appartments •
Plot No. 15 • Tulsibagwale colony •
Sahakarnagar No. 2 • PUNE-41 1 009 •

NEW PRODUCTS

Humidity Dry Bulb Temperature
Recorder

Minicorder-870 combines analogue and
digital instrumentation techniques in an
instrument to provide both digital dis-
play of % RH and dry bulb temperature
and strip chart record of the values at 12"
hour. The recorder employs latest solid-
state sensor. It is housed in an ultra com-
pact cabinet of 144 x 144 mm and record-
ing mechanism consists of only 3 sub-as-
semblies. The recorder can be com-
manded to record continuously cither re-
lative humidity or dry bulb temperature
or alternatively both.

Two displays provide continuous read-
ings of the parameters. Even 15 days of
continuous recording is possible. A
choice of two types of sensors is availa-
ble.

Instrument Research Associates Pvt.
Ltd. • P B No. 2304 • No. 228 • Magadi
Road • Bangalore-560 023 • 350830
355836 •

Programming Module

Professional Electronic Products offors
advance Bipolar Programming Module
PGM 8530, which can be used with
PEP's Universal Prom Programmer PP-
85. PGM 8530 is capable of program-
ming Bipolar PROM’s of Signetics, Na-
tional Semiconductor and Texas Instru-
ments. A Comprehensive user manual is
supplied alongwith the system for opera-
tion of PGM 8530. Some proms of diffe-
rent make are listed which can be prog-
rammed through PGM 8530. Signetics
825-23, 123, 126, 129, 130, 131 137, 135,
147, 195, 115, 140, 141, 180, 181, 270,

183, 191, 321 etc. National Semiconduc-
tors 74S/54S-188 288, 287, 387, 570, 571 ,
572, 573, 184, 185, 472, 473, 474. 475,
180, 181, 190, 191, etc. and I. I’s 24S-10,

M/s. Professional Electronics Products •
Opp. Old Octroi Post • Delhi Road •
P.B. 316 • MEERUT-250 002 (U.P.) •
Phone: 20460, 20159 •

Breakdown (Flash) Tester

Arun offer a high voltage breakdown
(flash) tester for output capacity of 0 to 3
KV, 0 to 5 KV, and 0 to 10 KV with leak-
age current of 3 to 200 mA in different
models. Voltage and current are indi-
cated on two seperate square moving coil
type panel meters. Accuracy of measure-
ment is +5%. A safety measure is the
audio-visual alarm devision the front
panel when HT is on.

The HV tester is useful for checking
breakdown (flash) voltage of insulation
material, electrical and electronics com-
ponents, motors, transformers,
switches, chocks, coil, oil, varnishes, re-
lays, etc.

M/s. Arun Electronics Pvt. Ltd. • 2 E,
Court chambers • 35 New Marine Lines
• Bombay-400 020 • Tel: 259207/252160

Proportioning & Dispensing
System

The Model I&J 100 is a proportioning
and dispensing system for spoxy , polyes-
ter, polyurethane, silicone and other two
component resin formulations. Its, fea-
tures includes: accurate and automatic
proportioning and dispensing flowable
liquids and pastes; positive displacement
piston metering; ratio adjustment from
1:1 to 100:1; pressure feed reservoirs
from 6 ounces capacity to 5-gallon; au-
tomatic pushbutton dispense switch; and
3-way recharge/dispense valves with air
operator actuator.

I&J Fisnar Inc., 2-07 Banta place • Fair
Lawn • N.J. 07410 • U.S.A. •

Temperature

Indicator/Controller

Hoshakum analogue temperature indi-
cator/controller Type: ATICB-003 is av-
ailable in range of from 200 to 1600°C
with suitable sensors like Cr/Al.Fe/
Const, thermocouple and Prt bulb also.
Automatic ambient temperature com-
pensation and broken sensor indication
are incorporated. The instrument is 230
VAC mains operated and conforms to
DIN standard panel cutout.

M/s. Hoshakun • Vivek Appartments •
Plot No. 15 • Tulshibagwale colony •
Sahakarnagar No.2 • PUNE-411 009 •

NEW PRODUCTS

Metallurgical Microscope

Vaiseshika offer the metallurgical mic-
roscope Type 7001 to conduct critical
examination of metallurgical samples.
Optical detection of surface ir-
regularities/inclusions and faults in met-
als can also be observed with the micros-
cope. The instrument reveals significant
specimen details with relief in black and
white or brilliant colours. It has ac-
hromatic objectives of 10 x and 45 x and
wide field eye pieces of 10 x and 15x.
These superior optic combinations pro-
vide a wide magnification range of from
25 x to 2500 x. Inaddition, coarse and
fine focusing arrangements enable verti-
cal displacement with 5 micron resolu-
tion. Perfect observations are facilitated
through optically aligned illumination
system. Spring loaded optics offer
safeguards against pressure and errone-
ous operations. Special provision is
made for rapid sliding of specimen plat-
form to accomodate bigger specimens.
The instrument, is housed in a wooden
cabinet; comes with an illustrated in-
struction manual for complete installa-
tion of the equipment with built-in trans-
former for illumination.

M/s. Vaiseshika Klectron Devices • Post
Box No. 57 • Near Allahabad Bank •
Ambala Cantt-133 001 (India) •

Signal Generator

Equilab Model TP-707 is a solid state
AF-RF signal generator covering the
range of 420 kHz to 24 MHz in four
ranges. It works on 3 V battery. Its sim-
ple operation mechanical stoutness and

clear scale reading make it suitable, to
measure comparative gain and sensitiv-
ity in broadcast receiver.

M/s. Electronic Instrument Laboratories
• B-69/004, Anand Nagar • Chhatrapati
Shivaji Road • Dahisar (East) •
Bombay-400 068.

Electronics Maintenance
Cleaner

Accra Pac (India), in collaboration with
Accra Pac Inc., USA, have introduced
the Safeguard Contact Cleaner for elec-
tronic maintenance. The cleaner is a spe-
cially formulated solvent which restores
electrical continuity of all types of con-
tacts and controls. Expceptionally pure
solvents under high pressure quickly
penetrate surface pores removing
grease, dirt, oil and oxidation, and
evaporates quickly leaving behind abso-
lutely clean contacts. It is said to have ex-
cellent dielectric properties and improve
performance and reliability of all elec-
tronic equipment.

The non-flammable, non-toxic.
Safegard Contact Cleaner is useful for
silver and precious metal contacts, TV
turners, miniature controls, solenoids,
circuit breakers, potentiameters, selec-
tor switches, volumes and tone controls,
relay contacts, thermostat control, dis-
tribution panels and all other electronic/
electrical contacts.

Accra Pac (India) Private Ltd. • 917,
Raheja Chambers • Nariman Point •
Bombay-400 021 •

Coil Winding Machine

The 152NC and 154NC vertical deflec-
tion yoke coil winding machine are CNC
machines that give flexibility and
facilities without compromising on the
throughput. The winding specifications
can be fed into the NC device manually
through keyboard retained, and recal-
led. The rotational axis and axial feed
are perfectly synchronised. Feeding and
making programmes are simple.

The model 152 NC can wind 2 half coils
at a time while 154 NC winds 4 coils at a
time. The machine comes with standard
accessories like the core chucks and bob-
bin support devices. The speed of wind-
ing can be 1200 RPM. The 152 NC are
particulars suited for handling vertical
coils of various specifications.

M/s. Muragappa Electronics Limited •
Agency Division • 29 Ilnd Street • Kam-
raj Avenue • Adyar • Madras-600 020 •
Phone: 41 33 87 •

LEARN-BUILD-
PROGRAM

The Junior Computer book
is for anyone wishing to
become familiar with
microcomputers, this book gives
the opportunity to build and
program a personal computer at
a very reasonable cost.

The Indian reprint comes to you from

elekt©?

Send full payment by
M.O./I.P.O./D.D. No Cheque Please.
Packing & Postage free

to: EfckTOR ElECTRONICS pVT It(J.
52-C. Proctor Road, Grant Road |E).
Bombay-400 007.