This is only a preview of the November 1998 issue of Silicon Chip. You can view 28 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. Items relevant to "The Christmas Star":
Items relevant to "Turbo Timer For Your Car":
Items relevant to "Build Your Own Poker Machine":
Articles in this series:
Articles in this series:
Items relevant to "Lab Quality AC Millivoltmeter; Pt.2":
Articles in this series:
Articles in this series:
Articles in this series:
Purchase a printed copy of this issue for $10.00. |
SERVICEMAN'S LOG
Big tellys, PCs & car computers
Why is it that all large-screen TV sets are
located up flights of stairs, making it
impractical to move them? I recently came
across two such monsters, along with a PC
that kept losing its icons. I also got tangled
up with a couple of car computers.
It was with fear and anticipation
that I set off to “look at” a large screen
Sanyo TV, as Mrs Collins had begged
me to. She was unable to tell me the
model number as she could not get to
the back of the set but she repeated
that the fault was no picture. And
because she was recuperating from
an operation on her leg, she really
needed to watch the telly.
When I arrived at the third storey
unit, mentally noting that there was no
elevator, I was surprised to find that
her home was even smaller than I had
imagined it to be. You couldn’t miss
30 Silicon Chip
the 78cm stereo TV as it took pride of
place in the living room. The furniture
was spread out in a semicircle around
the set, as though it was praying to the
craven idol.
How they got the set into this
matchbox must be a story in itself but
one thing was for sure – it was going
to have to be fixed on location, as I
certainly didn’t have the facilities to
take it to the workshop.
When I finally managed to move the
set out a fraction from the wall and
crawl behind it, I established that it
was a Sanyo CPP3310TX-01 (A4-A33
chassis). Because of the restricted
space, it took some effort to remove
the back but this did little to make the
access any easier. Secretly, I prayed
to the Patron Saint of Television for
a merciful and quick conclusion to
this repair.
I had already established that
there was sound but no raster from
any source – be it from the TV tuner
or the AV inputs. Nor was there any
Teletext on Channel 7. I propped up a
large mirror in front of the TV so that
I could monitor any activity from the
rear and then turned up the screen
control on the flyback transformer to
produce a blank clear raster.
Well, the report card on the TV now
showed me that almost all the circuits
were working. However, the luminance and chrominance signals were
disappearing somewhere between the
video detector and the picture tube.
And because there was no picture
from the AV inputs, this meant that
the problem could be shifted a little
further away from the detector – possibly to the analog switching circuitry.
Even so, it was looking more and more
like a problem with the small-signal
decoding stages and for this I needed
a circuit diagram.
Mercifully, Mrs Collins had kept
the one that came with the set and so
I started by checking the nine major
voltage supply rails (B1-B9) from the
switchmode power supply, the flyback
transformer and the remote control receiver auxiliary supply. I also checked
the ancillary rails which are derived
via IC regulators and checked the low
voltage rails to the signal processing
circuits but nothing was amiss.
By now, I was suspecting IC201
(M513086P) – the jungle IC – but I
really couldn’t go any further without
an oscilloscope. I left, having made an
appointment to call back later in the
week. As I drove off to the next job,
one point I had noticed with the TV
was niggling away at me – there was
no on-screen display. I had dismissed
this at the time but now I couldn’t
help feeling that I had been down this
path before.
When I got back to the workshop,
I phoned Sanyo Technical Support
and they also (without prompting)
suggested the jungle IC. They also
advised me to check crystals X421
and X261. To save time, I ordered the
IC just in case this was the problem.
I was back at Mrs Collin’s little flat
a few days later and this time there
was even less room as I had brought
an oscilloscope and a signal generator.
Access to the underside of the main
PC board was impossible and it wasn’t
much better for the top because the AV
board got in the way. Because nothing
could be seen on the screen, I couldn’t
tune the TV into the RF output of the
pattern generator so I injected colour
bars into the video input phono socket
(AV1) instead.
It took me quite some time with the
circuit to finally work out that the AV
module not only selects the various
sources but also splits up the video
into chrominance and luminance
signals. The latter then goes through a
“video unit” which is a noise reducer
circuit, before finally arriving at pin 18
of IC201. The CRO confirmed this and
also that the crystals were oscillating
with the remote control.
Next, I selected a channel number which had good sound before
checking the video output from the
IF pack on test point TP-J with the
oscilloscope. The signal was there
and also at IC201. Reselecting AV1, I
then measured the colour difference
output signals from IC201 on pins
1, 2 and 3 and was surprised to find
them all there. I then followed these
signals through the AV board “Super
Imposer” (IC1307), which switches
through the Teletext, and onto the CRT
socket where the signal died after the
emitter input to the output transistors.
It was then that the bells began
ringing, because I had been here before with a very similar problem in a
Palsonic 5138 TV set (see November
1997). Sure enough, here was the exact
same component, namely a 120kΩ resistor (R692) feeding the base circuit of
the video output transistors from the
+200V rail. And once again, the high
voltage had been too much for this
low-wattage resistor and it had failed,
switching off all three transistors.
If only I could remember all these
faults, although I’m sure I won’t for-
Fig.1: the standby switching circuitry in the Sony KV-C2931S. A break in
the circuit between R634 and D624 meant that Q601 was being turned off
but the fault was intermittent.
get this one the third time around.
Anyway, Mrs Collins was delighted
and I was left with the challenge of
disengaging myself from the tangle
of equipment and getting out of that
tiny flat in one piece.
Another monster
In the meantime, yet another house
call was booked in for a large screen
TV. Why is it that all the biggest sets
are located in the most difficult positions – usually up flights of stairs
with no lifts? Or is it just me being
paranoid?
This time, the set was a 1990 Sony
KV-C2931S, using an AE-1B chassis
made in West Germany. The complaint was intermittent no sound or
picture. Mr and Mrs Mowbray, who
owned the set, were a retired couple
living on the top floor of a duplex
and had enough trouble getting up
and down the stairs as it was, so they
were in no position to assist me with
the TV. Once again, it was a case of
the workshop going to the TV rather
than vice versa.
Because I had serviced so many
of these sets, I thought that the fault
would be easy to fix. It would probably turn out to be a dry joint on the
motherboard and I was quite confident
of an early and successful victory as I
climbed the many stairs.
I arranged with the Mowbrays to get
a blanket and cushions ready and then
gently tilted the set over onto its front
to rest on them (the cushions, not the
Mowbrays). Removing two screws and
a couple of clips gave me access to the
underside of the set and I then spent
over half an hour resoldering many
suspect joints on the motherboard,
especially around resistors R614 and
R653 and around Q608.
Some of the dry joints were inaccessible under the black plastic frame
and the only cheap and quick way to
address these is to cut a small part
of it away. Anyway, I resoldered all
the transformers, transistors, diodes,
plugs and sockets and some of the
power ICs. Finally, satisfied with a
job well done, I reassembled it all and
made sure all was working well before
confidently leaving.
Unfortunately, my triumph was
short-lived – Mr Mowbray phoned
about two weeks later and said it was
doing the same thing again and could
I please call back. Disappointedly, I
returned, and sure enough the picture
and sound would come and go as you
tapped the set. Now I had in fact tested
this thoroughly the last time, when
tapping it or even thumping it made
no difference.
This time I left the set in an upright
position and gently removed the back,
then tapped around the set exploring for more sensitive spots. After a
while, I felt sure that the problem lay
on tuner IF module A, so I reworked
the soldering all over this board, concentrating on the edge connectors, the
tuner and any coils. I then unsoldered
the IF module (VIF901) and reworked
November 1998 31
the soldering inside there, especially
over any coils and ceramic filters
which, I might add, were almost all
dry-jointed.
Once again, full of confidence, I
reassembled everything and tried tapping around the whole set, especially
the tuner IF module. The picture and
sound were rock steady and I left feeling somewhat annoyed that having
soldered so many bad joints the first
time, I had discovered even more the
second time. But wait, there’s more.
This is where the story should have
ended but unfortunately it didn’t.
About four weeks later, a rather tetchy
Mr Mowbray called again and told me
the set still wasn’t fixed. For a while
they could tap it and the picture and
sound would return but now it was
gone completely. The only thing
showing any sign of life was the green
on-screen display.
I returned as soon as possible to the
Mowbrays, armed with the full kit
again – namely the oscilloscope and
pattern generator, which I connected
to the AV1 input SCART socket via an
adaptor. I then began tracing the video
signal from the SCART socket (pin 20)
to the output from the AV board (pin
16, J1-41) and onto the motherboard.
From there, I traced the signal from
the chroma decoder and finally to the
Teletext board (CNV-01 pin 3). But
there was no video going into video
32 Silicon Chip
amplifier stage Q598 (waveform 22)
which supplies the sync separator
(IC501).
It was this puzzling part of the
problem that wasted half an hour of
my time. The circuit clearly shows a
direct path from pin 3 to C592 and
then onto the emitter of Q598 but
in reality, there was a link. This is
designated JW105 but had not been
installed in this set. Shorting across
the link position restored both picture
and sound, so you may be forgiven
for thinking that just refitting the link
would be the answer. However, it does
not explain why simply tapping the
set used to fix the problem.
The answer was that Teletext is an
optional accessory and the link is only
fitted when this option is left out. If
Teletext is included, as it was in this
case, the video goes into the Teletext
unit on pin 3 and comes back out
again on pin 2. This meant that the
problem was now narrowed down to
the Teletext module, or so I thought.
Removing the V board and examining
it revealed yet more dry joints, especially around Q01, the 5V regulator
stage. Unfortunately, re-soldering the
joints didn’t fix the problem – there
was still no 5V rail.
I soon discovered that not only did
I not have 5V but that the 7V rail into
the module was also missing. The 7V
rail comes in on pin 6 of connector
CNV-01 and I traced the line
back to the collector of transistor Q601 (the standby switch)
on the motherboard. There was
8.4V on its emitter (which is
correct) but this same voltage
was also present on the base,
thus completely switching the
PNP transistor off.
Q601’s base is controlled
by transistors Q606 and Q609
via R634, D624 and R618 –
see Fig.1. There should have
been 0V on Q609’s collector
and this proved to be the case.
There was also 0V present on
the cathode of D624 but R634,
which is connected directly to
it, measured 8.4V. This meant
that there had to be a break in
the board track between D624
and R634, although the track
looked perfectly OK.
This track, by the way, is only
about 25mm long and passes
next to a hole drilled through
the board. This might have
been a screw hole but there was
no screw fitted – maybe it was only
used in the course of manufacture and
discarded later.
I scratched away the lacquer covering the track and, using an ohmmeter,
eventually found a hairline fracture
in the copper. Resoldering the crack
fixed the problem and I carefully
reassembled the set. A check some
three months later revealed that
everything was still OK, so I think I’ve
finally fixed the monster.
If I’m wrong, I’m returning with my
best fine-tuning tool – my baseball
bat. John Cleese will have nothing
on me after I’ve given that set a stern
talking to.
Missing icons
In addition to the usual VCRs and
TV sets that come my way, I’m also
often confronted with computer
problems. Some of these can be quite
unusual and I have one such story
this month.
The first was a 1994 ICL ErgoLite
computer, which is actually an Acer
80486 DX-50 in disguise. Its main
problem was that it was unable to save
any setting made within Windows
3.11 on exit, resulting in a blank Program Manager with no icons.
At first, I thought that this would be
a case of simply reloading Windows
over the top of the existing installation
but this didn’t help. I then spent some
time reloading DOS 6.22 and customising its config.sys and autoexec.
bat files to optimise memory but that
didn’t help either. I then ran Scandisk,
a virus checker and defragg
ed the
drive but I wasn’t getting anywhere.
The resources were fine and the computer hardware was fine too.
My next step was to examine sys
tem.ini and win.ini but again I could
find nothing untoward. What finally
put me onto the right track was thinking more carefully about what was
actually happening and especially
why the icons were disappearing (very
Agatha Christie).
This part of the operating system is
controlled by Progman.exe and customised with Progman.ini. Using my
trusty XTree Gold (the one and only), I
examined Progman.ini and compared
it with the Progman.ini from another
computer that worked. Normally, this
text file has two headings, [Settings]
and [Groups], the former giving details
as to how the groups are arranged and
the latter describing what the groups
actually are. However, on the faulty
computer there were two or three lines
of machine code and then the settings
and codes were repeated about half a
dozen times, often with conflicting
information.
At last I was getting somewhere.
Hopefully, all I would have to do is
edit the Progman.ini file but first I
went to the Windows directory and
printed out a list of all the group files
(*.grp) for use as a reference. After
that, I backed up Progman.ini and
then edited the file by removing all
the machine code and then listing
all the Group files under the [Group]
heading. Finally, I edited the lines
under the [Setting] heading, removing
any duplicates in the process, and resaved the file.
This simple measure completely
fixed the problem. The next time I
went back into Windows, all the icons
were there – indeed there were even
some duplicates. A little extra housework made everything shipshape and
the program groups now remained in
place each time I exited and rebooted
Windows.
My guess is that the problem had
been caused by the computer being
turned off (or crashing) while Windows was still run
ning. This had
created lost clusters and chains which
had been repaired as best as possible
by Scandisk. Unfortunately, the Progman.ini file had been corrupted and
could only be manually repaired using
the method just described.
The moral of the story is to always
exit Windows properly before turning
the computer off.
Car computers
My final two stories for this month
are also about comput
ers but this
time, they involve car computers. So
how did I come to get involved in car
computers? Simple – a mate of mine
runs an automotive workshop.
The first story is about a Eunos
that wouldn’t start, although we were
pretty sure that it wasn’t the computer
itself that was at fault because a substi
tute didn’t fix the problem. The car
would crank and there was spark but
no injector pulses. Based on this, we
deduced that the problem probably
lay somewhere between the computer
and the fuel injectors but there were a
few other possibilities. With amazing
luck, we actually managed to obtain a
photocopy of the wiring diagram for
the car but not of the computer itself,
unfortunately.
We began by using a multimeter to
establish that 12V was getting to the
computer and to the injectors – and
this proved to be the case. However,
it was prophetically pointed out at
the time that it isn’t a good idea to
use a multimeter in auto electrics, as
the internal resistance is so high that
it can give very misleading results.
Instead, it is far better to use a 12V
lamp probe but we couldn’t find the
workshop unit at first.
After these initial checks, I decided
to use a CRO to check the input pulses from the two crank sensors (why
two?) and the output pulses from the
computer to the fuel injectors – this
while someone else cranked the engine. Everything seemed to be working
perfectly and so it definitely looked
like a wiring loom problem.
Next, I moved the CRO to the engine
bay and checked the pulses going
directly to the injector connectors.
These were all present, so why wasn’t
it working?
It was at this point that the lamp
probe was found and we checked
the 12V rail to the injectors again.
And whereas the meter had told us
there was a full 12V, the lamp told us
differ
ently. The 125mA globe (24V
3W) hardly lit, which indicated a
high resistance in the wiring loom
and/or the relays between this point
and the battery.
Even though we had the wiring diagram, we still had difficulty following
the leads and identifying the various
plugs and sockets. To begin with, in
order to confirm our diagnosis, we
connected a jumper from the fusebox
to the injectors and cranked the engine. To our delight, it immediately
fired up but the engine stopped as
soon as we removed the jumper
It was while we were tracing the
12V rail with the lamp probe that we
almost had a disaster. The lamp probe
consists of a festoon globe inside a hollow uninsulated metal cylinder with
a sharp point on one end. The other
end has a lead with a crocodile clip
going through an insulator, a spring
and another insulator to the other end
of the globe. Unfortunately, I pulled
too hard on the lead while trying to
reach one of the sockets and broke the
internal insulator, thereby shorting
the lamp completely out.
Immediately, buckets of amps start
ed flowing and the lead started to melt.
Fortunately, I was able to break the
circuit within a second or two of seeing the smoke – but what damage had
November 1998 33
Serviceman’s Log – continued
I caused? Well, as it happened, none
because when we cranked the engine,
it started immediately. The high current had “cleaned out the cobwebs”
and improved the bad connection we
were trying so hard to find.
This was maddening because this
was now a classic Clayton’s fix – ie,
the repair you have when you are
not having a repair! Luckily for us,
we had already narrowed it down to
a plug and socket near the fusebox
marked X07 (white and red wire)
and actually wiggling it with a x1
ohmmeter across it saw the needle
tremble slightly.
So rather than leave it with a “she’ll
be right, mate” attitude, we soldered
a jumper lead across the plug and
socket. The car was then soak tested(!)
but it gave us no more grief.
Airing a Subaru
The final story in this computer
trilogy of bravely going where no-one
else will go concerned a 1990 Subaru,
which had an electronically-controlled suspension height system.
This consisted of four air-bags in lieu
of coil springs that were individually
pumped up and down with a compressor and controlled by solenoids.
These were in turn controlled by a
34 Silicon Chip
microprocessor, either automatically
or when commanded by the driver.
Of course, the problem was that
nothing was happening when the
buttons were pushed – the suspension
remained in its lowered position.
Various other people had had a go at
trying to fix the problem and the bone
was being pointed at the computer as
the likely source.
Once again, we had a wiring diagram for the vehicle, which was a
relief, but not for the computer. Initially, I was given the computer box
only and told “go fix”. I must confess
that I wasn’t all that optimistic at first
but what did we have to lose?
The computer consisted of a large
microprocessor chip and a lot of
peripherals, plus seven 2SD1392
transistors whose collec
tors went
directly to the connection sockets.
Initially, I spent some time checking
for all the obvious things (ie, dry solder joints, dry electros and short/open
circuits) before turning my attention
to the transistors. These all measured
good except for one which was badly
cooked. You beauty!
I reported back that I had found a
fault in the unit but pointed out that
some external fault must have caused
the damage. There was nothing for it
but to check it out in circuit with a
lamp probe.
First, we established that there was
12V going into the computer on pins
2 & 3 and that there was also 12V on
the collectors of all the transistors.
We then identified which transistor
did what. Four controlled the airbags
at each wheel, one the compressor
relay and the other two the charge
and discharge valves.
Our next step was to short each
of the collector connections on the
socket to ground in turn (with the
computer unplugged). Each solenoid
operated correctly except for the dis
charge valve which is controlled via
pin 1. And guess which transis
tor
controlled pin 1.
Temporarily, we took the discharge
valve out of circuit and refitted the
computer. This confirmed that it was
actually doing its job because we
could hear the relays working. So why
wasn’t the suspension being raised
and lowered?
The answer was that the discharge
valve was not only short circuit but
was also stuck open. Unfortunately,
it is mounted on the compressor and
this can only be accessed by removing
lots of other parts and putting the car
up on the hoist (the compressor and
valves sit just above and in front of
the passenger’s front wheel).
At first we thought that we would
be scuppered by the hoist because the
wheels would drop and the computer
would think that the system was fully
discharged so we couldn’t test it. However, we managed to confirm that the
air-switch solenoids were operating
but that there was no air to pump the
airbags. We checked the reservoir and
pipes for leaks but found none, which
meant the problem was definitely the
discharge valve. Unfortunately, this
is an integral part of the compressor
assembly and is a totally unserviceable unit.
It wasn’t easy removing and replacing the compressor but a new one at
– wait for it, almost $2000 – finally
fixed it. After all that, give me a telly
SC
any time.
|