This is only a preview of the October 1994 issue of Silicon Chip. You can view 33 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Items relevant to "Beginner's Dual Rail Variable Power Supply":
Items relevant to "Build A Talking Headlight Reminder":
Items relevant to "Electronic Ballast For Fluorescent Lights":
Items relevant to "Computer Bits":
Articles in this series:
|
SERVICEMAN'S LOG
Two symptoms – one fault or two?
I’ve got a really weird one this month – two
quite different visual symptoms & two faulty
parts mixed up in a crazy tug-o’-war. Less
traumatic was the set that went green; it fooled
the customer more than me.
The weird story is about an AWA
colour TV set, model 4303, using one
of the “Q” series chassis. It would be
about 10 years old and is one of several
in a local motel. As with many other
AWA chassis types, the “Q” series are
actually made by Mitsubishi.
I first heard of the problem when the
motel proprietor rang me, identified
the set, and explained that one of his
guests had reported that the set had
lost colour. When he later checked the
report it was quite correct. There was
no colour but, as he added, there was
also a black line or strip about 50mm
wide at the top of the picture.
That should have alerted me – well,
alerted me more than it did. But I did
speculate as to whether I had two
separate faults – which seemed most
likely – or whether it was a single fault
with a funny origin; and I didn’t mean
funny ha-ha.
As it transpired, there weren’t any
laughs anywhere in the episode. I
don’t know what the record is for
frustration factor but, on a scale of
1-10, this must have been nudging
the nine mark.
Having thus set the scene, let’s get
down to details.
No ordinary fault
The customer delivered the set
to the workshop and I turned it on
while he was there. And yes, his
description was fairly accurate; there
was no colour and there was a black
band about 50mm wide at the top of
the picture.
But there was more to it than that
and I quickly realised that this was
no ordinary fault. For a start, it was
obvious that the black band was not
simply a result of reduced vertical
scan, involving either a compressed
or non-linear image. What image was
there was normal and the black band
was, as it were, overlaid on the image.
In other words, the scan was normal,
but there was some kind of spurious
blanking problem.
The other thing I noticed immediately was that the junction between
the picture and the black band was
not a straight line, as one would have
expected. Rather, it was a “wavey”
line, perhaps best described a rough,
shallow sinewave of about 12 cycles. I
also discovered that I was able to brute
force the set into momentary bursts
of colour by carefully fine tuning it,
although there was no setting that
would hold it.
But having noted all this, I was no
wiser as to whether it was one fault or
two, although I tended to favour the
two fault theory. In any case, I could
only tackle one set of symptoms at a
time, so I decided to tackle the blanking problem, mainly because I felt
more confident about where to start.
Unfortunately, there was one other
trap waiting for me. I didn’t have a
“Q” chassis circuit for this particular
model, which uses a 90 degree picture
tube. The closest I had was one for a
110 degree tube but, as far as I knew,
Fig.1: the faulty section in
the AWA 4303. Part of IC201
is shown on the left, with pin
9 in its bottom right hand
corner. Diode D204 is below
it & to the left, while diode
D203 is to the right near
transistor Q213. The burst
gate transistor (Q601) is at
extreme right.
56 Silicon Chip
the sets were iden
tical in all other
respects.
And I fell right into the trap. I spent
a great deal of time trying to find my
way around the chassis from this
circuit and found that I was getting
nowhere. I eventually realised that
it was almost the same but not quite,
Finally, a colleague came to the rescue
with the correct circuit.
Having cleared that hurdle, I started
all over again. I was concentrating
on the circuitry around IC201 and,
in particular, around pin 9, which
apparently feeds blanking pulses into
the blanking section – see Fig.1. There
are two diodes connected to this pin
– D204 and D404. D204 is adjacent to
this pin on the circuit, while D404 is
some distance away down near the
scan coil assembly, being associated
with a small resistor network (R417,
R418, R419 & R420).
Also under suspicion were some
electrolytic capacitors, including
C414, C409 and C408, in the adjacent vertical output stage. I replaced
these first, without any result, and
also checked various resistors in this
part of the circuit. They all measured
spot on.
That left the diodes still under suspicion. But first I decided it would
be a good idea to do a voltage check
around IC201. Fortunately, all the pin
voltages are shown on the circuit and,
with one exception, they all measured
well within tolerance.
As you may have guessed, the exception was pin 9. It is marked 6.1V
but I measured only about 2V. So did
we have a faulty IC? I had a spare on
hand and it was not a big job to fit it.
It made no difference but at least I had
cleared it of suspicion, a point about
which I was thankful later on.
That left the diodes as the next
prime suspects. I went to D204 first.
The simplest and most reliable way
to check it was to pull it out and fit a
new one, which I did.
Now, at the risk of seeming to state
the obvious, there would appear to be
only one of two possible results from
such a move: either the fault would
be cured and that would be the end
of the exercise, or (2) it would make
no difference and the diode would be
cleared of suspicion. Surely, those are
the rules?
At least, that’s what I thought until
I replaced diode D204. But no; this
circuit had its own ideas. We now
had complete picture cutoff; in other
words, the situation was worse than
before. My first reaction was to suspect
that the replacement diode was either
faulty or unsuitable. I didn’t have a
direct replacement but had used a
1N914 small signal diode, which I felt
was adequate.
I tried another 1N914, then several
other types, but always with the same
result; total picture cutoff. I checked
the original diode for leakage and although the indication was only slight,
I felt sure it was leaky. Yet when I
refitted it, I could at least get the original picture.
To say that I was confused would
be putting it mildly. It really threw
me; what on earth was going on? Although I eventually decided that the
original diode was faulty and that the
replacement was OK, I was no closer
to an explanation. About the only thing
that was clear was that there was another fault somewhere which still had
to be found. Apparently, the original
“two-faults” concept was valid but not
in the manner I had envisaged.
A real clue
But speculation didn’t help in a
practical sense and I was at something of a loss as to what to do next.
In desperation, I went back to pin 9.
And this provided the first real clue;
the voltage here had now jumped
from a too-low value of around 2V to
a too-high value of about 8.4V. Well,
I suppose that made sense in a way;
excessive voltage into the blanking
circuit would do just that – it would
blank the picture.
But where was this excessive voltage coming from? In order to follow
the next steps, it is necessary to study
this part of the circuit carefully. First,
there are three resistors in ser
ies,
R209, R223 and R224 (in that order),
from pin 9 to the 12V rail. Their job
is to establish the 6.1V at pin 9 shown
on the circuit.
Also connected to pin 9 is R210,
C217 and diode D204. And somewhere
via that network a spurious voltage
October 1994 57
was being introduced. All I had to do
was find out how.
The first thing I did was to disconnect R224 at the 12V rail, which should
have removed all voltage from pin 9.
But it didn’t; we still had the 8.4V.
Next I disconnected R210. Well that
achieved something; the pin 9 voltage
dropped to zero. Thus inspired, I abandoned the circuit and began tracing the
copper pattern from R210, checking
with the meter probe as I went.
Of course the pattern was far more
complex in reality than it appears on
the circuit and I ran up a lot of garden
paths and encountered a lot of brick
walls over the next 15 minutes or so.
But suddenly I struck oil; a diode
marked D203 (near Q231), the other
side of which connected to the 12V
rail. More importantly, its polarity
was such that it was opposing the 12V.
I had no idea what its real function
was – and still haven’t – but it was
obvious that, if it was leaky, it could
be the culprit.
So out it came. And was it leaky?
A sieve is the only comparison I can
offer. So in went a new one and all our
troubles were over.
With the benefit of hindsight, it
58 Silicon Chip
appears that I might have been better advised to stick with the circuit,
because the offending diode is right
alongside IC201, connecting to the
12V rail where this emerges from the
12V regulator transistor (Q231). But of
course it was a lot further away on the
board than it appears on the circuit.
So that was the solution. But it had
been a most frustrating exercise. It is
bad enough to have two components
fail at the same time but they usually
produce distinctive symptoms. In this
case, not only were both failures in the
same part of the circuit but, worse than
that, they were actually opposing one
another in the effect they had.
Thus, while the leak in D203 was
attempting the raise the voltage on
pin 9, the leak in D204 was pulling it
down – and succeeding rather too effectively. But this was the only reason
the set was producing any image at all;
as I found out when I replaced D204
and made matters worse.
Which is all delightfully simple to
explain when looking backwards; it
only we could look forward as easily.
And what about the colour failure?
How can that be explained? Again,
once the fault was tracked down and
corrected, the connection became
obvious (no pun intended). If we go
back to the junction of R210 and C217
and follow this circuit to the right, we
come first to the anode connection
of diode D203, which caused all the
bother.
From here the circuit continues to
the right, connects to the cathode of
D205, and then to resistor network
R586, R584 & R585 (this network
connects to the horizontal output
transformer, from which it picks up
horizontal pulses). The circuit then
leaves the main board, via connector
pin 10, and goes to connector pin 10
on the chroma board, then via R624
to the base of Q601. And Q601 is the
burst gate transistor.
Most importantly, this is a DC circuit
all the way; whatever spurious voltage
appeared on this line from D203 would
appear on the base of Q601, modified
only by the divider network of R624
& R625. And, incidentally, R625 is
incorrectly shown as 22kΩ; it is, in
fact, only 2.2kΩ. Even so, there would
be a lot more voltage at this point than
normal, effectively upsetting the burst
gate function.
Of course it was a happy ending for
the customer but only partially so for
Yours Truly. I was glad to have solved
the problem but I wish I’d done it a
little quicker.
Little green pictures
And now for something a little
more straightforward, although it did
have its period of confusion. Among
other things, it demonstrated how a
customer’s description of a fault, no
matter how well intentioned, can set
one thinking in the wrong direction.
It involved an HMV colour set, model 12641. The same chassis is used in
the model 12642 and in the JVC model
7765AU. The owner first brought the
set in several months ago, with the
complaint that,”... the picture goes
green – but only sometimes”.
Well, the “only sometimes” didn’t
exactly cheer me up but otherwise
I assumed it would be a fault in the
picture tube drive system; either the
green gun being turned hard on, or the
red or blue gun (or perhaps even both)
being turned down in some way.
I turned the set on while he was
there and, sure enough, it was producing a normal picture. I suggested
he leave it with me for a few days and
so the set sat in a corner of the bench
and ran all day and every day for the
next week or so. And it never missed
a beat; there was not even a suggestion
of a green cast.
Finally, I suggested that as we
weren’t getting anywhere, it might be
better if he took the set back home until
the fault became more predictable.
And that was the last I heard about it
for the next three months or so. In fact,
I had almost forgotten about it when
the owner suddenly turned up with
it, saying, “It’s real crook now – goes
green every day.”
And so it was back into the corner
of the bench. But he was right this
time. It had been running for less than
half an hour when the fault suddenly
appeared. But as soon as it did, I re
alised that I had been thinking along
completely wrong lines.
It wasn’t a green cast; instead, it
was green faces, with all other colours
similarly incorrect. Well that put a
different complexion on things (oops,
sorry about that) and that meant a
completely different approach. It was
in no sense a picture tube drive problem; it was phase fault which meant an
inversion, shift, or upset of some kind.
But the interesting aside here is that
it was only the green flesh tones that
attracted attention. That’s not surprising in one way, I suppose, since these
are normally the centre of attention.
At a more practical level, the most
likely cause of such a problem would
be failure somewhere in the half-line
frequency (7.8kHz) chain, starting at
the phase discriminator, where this
frequency is generated in the process
of pulling the crystal oscillator into
phase with the burst frequency.
The 7.8kHz frequency is used to
operate the reversing switch which
changes the colour phase on each alternate line in synchronism with the
transmitter. And when it misbehaves,
which it can in variety of ways, it can
do dreadful things to the colour.
(In order for the receiver to perform
this switching in correct phase with
the transmitter, the PAL system employs a swinging burst signal. This
4.43MHz reference burst is shifted 45
degrees, plus or minus, on alternate
lines and the receiver uses this shift as
a code to identify each line. The high Q
of the crystal oscillator averages these
two shifts, while the phase discriminator, which controls the crystal phase,
also provides the half-line frequency
for the reversing switch).
Fig.2: the 7.8kHz oscillator circuitry in the HMV 12641. This oscillator
consists of transistors X304 & X305, with X303 designated as a 7.8kHz
killer. The 7.8kHz signal is fed to the demodulator IC (IC302) at top right.
In this set, one of the easiest points
of access to the 7.8kHz chain is at
transistors X304 and X305, described
jointly as the 7.8kHz oscillator – see
Fig.2. This feeds a 7.8kHz signal into
the demodulator IC (IC302). There
is also X303, which is described as
a 7.8kHz oscillator killer. However,
X304 and X305 were the most likely
suspects.
But ease of access was not the only
reason I selected this point. The transistor type used here, 2SC458, is one
that I regard as a mite unreliable, being
prone to intermittent behaviour.
Finally, a brief check with the CRO
confirmed that this was where the
frequency was running into trouble.
So it really boiled down to which
of the two transistors was the most
likely culprit. Well, it was a 50-50
chance and I took a punt on X305.
And for once I picked it in one; I fitted
a replacement and all the faces were
back to normal.
I ran the set on the bench for several
days with no sign of the problem and,
although I remembered it had done
this before, decided to pass it back to
the customer with the least possible
delay. But I warned him to contact me
immediately at any sign of the trouble.
Subsequent checks have confirmed
that there has been no recurrence of it.
So that was it. It wasn’t a highly
scientific exercise but was more a
result of previous experience, plus a
certain amount of luck. And it does
happen that way sometimes. But the
customer’s description did throw me
initially.
Back to wireless
To finish off this month, I’m breaking right away from the usual to indulge in a little nostalgia. In the hurly
burly of modern high-tech electronics
– and the high-tech service equipment
which it demands – we sometimes
forget, or perhaps never knew, about
electronics in its infancy.
It wasn’t known as electronics then
of course – it was radio or, before that,
wireless. Which was fair enough,
because the wireless set was virtually
the only manifestation of what was to
become electronics.
But regardless of what it was called,
or the state of the art, the equipment
of the day needed servicing. For the
wireless enthusiast of the day, or the
local garage mechanic who doubled as
a wireless expert, this was more often
than not undertaken on the “by guess
and by God” basis.
As for service equipment – well this
was often limited to a few basic tools
– pliers, screwdrivers and a soldering
October 1994 59
iron. And diagnosis was on the basis of
visible faults: loose terminals, broken
leads, unlit valve filaments, or obviously defective controls.
Which was OK up to a point. But
wireless sets used batter
ies – and
that, as they say, was the ‘ard part.
Meters were few and far between,
quite crude, and very expensive.
They didn’t even approach the simple
1000Ω/volt multimeter which was
later to become the mainstay of radio
servicing.
The accompanying photograph is
of one of the very early attempts at
a meter for use with wireless sets. It
was passed over to me by a colleague,
who acquired from a non-technical
friend who found it in some junk in his
workshop. Apart from that, its origin
and history remain a mystery.
It was a highly specialised piece of
equipment. With a 0-50V scale and
with prominent markings at 22.5V and
45V, it could have had only one role in
life: to test B batteries. For the benefit
of younger readers, the B battery – or
high tension battery – came as a 45V
unit, tapped at 22.5V.
A small set (eg, one valve with
60 Silicon Chip
earphones) would use one light duty
version, while larger sets would use
at least two heavy duty types to give
90V, or three to give 135V. And they
were horribly expensive.
Advertisements from wireless magazines of that era suggest that the keenest price for a
45V heavy-duty battery would be
£1/5/0 ($2.50), or £3/15/0 ($7.50) for
Fig.3: this pocket meter was the latest
thing in test equipment in the 1920s.
All it could do was test the B battery.
a set of three. But the basic wage was
then only around £3/12/6 ($7.25).
Most people earned a little more than
that, say around £4/0/0, but a set of
batteries would still make a mess of a
week’s wages (work that out in modern
terms)!
With average use, but without wastage, a set would provide 6-9 months
of use. And that, by any standards,
made a wireless set an expensive
thing to run.
So nobody discarded them until
they were convinced that they really were exhausted – and that the
deteriorating perfor
mance was not
due to some other cause. Hence the
popularity of the little meter portrayed
here. For the repair man, it provided
the proof needed to sell another set
of batteries. And the enthusiast who
owned one was the envy of his peers;
his popularity – and a regular invitation to dinner – was assured.
The meter itself is almost certainly
a moving iron type, renowned for its
simplicity rather than sensitivity, and
recognised by non-linearity at the low
end of the scale. The colleague who
passed it over to me checked it against
his “u-beaut” digital meter and, to
the accuracy with which he could
read the simple scale, pronounced it
“spot on”.
He also checked its sensitivity, and
found that at 45V it drew about 10mA.
This was probably more by accident
than by design but it would not have
been an unreasonable load with which
to test these batteries. Typical current
drains would have been 10-15mA.
How old is it? I’ve passed it around
to several old timers but no-one’s game
to admit to ever having seen one in
actual use, for fear of revealing their
age. However, history suggests that it
would have been popular in the early
1920s, or about 70-plus years ago.
Having said all that, the thing that
stands out most in my mind is the
point I made at the beginning; the very
narrow application for the device. It
could do only one job – test individual B batteries. As a general purpose
meter for use in the wireless set itself,
it was virtually useless. Unless the set
used only one 45V battery, it could
not be used even to confirm that the
HT voltage was present anywhere in
the set itself. A meter to do that was
several years down the track.
But that’s how things were in the
SC
good old days.
|