This is only a preview of the June 1995 issue of Silicon Chip. You can view 27 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 "Build A Satellite TV Receiver; Pt.2":
Articles in this series:
Items relevant to "A Train Detector For Model Railways":
Items relevant to "A 1-Watt Audio Amplifier Trainer":
Articles in this series:
|
SERVICEMAN'S LOG
Faults that don’t obey the rules
Frustration is the theme of this month’s notes.
It’s nice to restore a device to full working
order but still very frustrating when it is not
clear why it failed, or why it behaved as it
did when it failed.
The first frustrating story concerns
an NEC colour TV set, model N2092.
It belongs to a local motel – a new
customer – and it turned out to be one
of those frustrating jobs which, while
satisfactorily concluded at customer
level, leaves a legacy of doubts and
queries as to just why it behaved as
it did.
It started with a 9 o’clock phone call
from the motel proprietor, asking me
to come and have a look at a TV set
which, to use his own words, “wasn’t
going”. That expression prompted
me to ask whether it was completely
dead – an unfortunate phrase perhaps
– to which he replied that, yes, it was
completely dead.
And he wanted me to service the
set in the motel, because the set was
bolted to a shelf in the motel room. But
I had to explain that I did not make
house calls, that service was seldom
practical away from the workshop,
and that, in any case, the set would
have to be unbolted before I could
work on it.
For once, I had struck a customer
who was quite reasonable about such
matters. He appreciated the problems
and agreed to bring the set to the shop.
However, he did stress that he would
like it back that day, if that would be
possible.
Naturally, I couldn’t make such a
promise. But I did say I would look
Fig.1: the horizontal output transformer circuitry in the NEC N2092. Pin 2
is at bottom right and feeds diode D503 via fusible resistor R522. Note the
waveform at pin 2.
40 Silicon Chip
at the set immediately and do what I
could, depending on the fault.
All of which doesn’t have much to
do with technicalities, but is simply an
example of the various matters which
have to be sorted out before a set is
even sighted.
Anyway, the customer turned up
a little later with the set in the back
of a ute. I set it up on the bench and
turned it on while he was there. And
that was the first setback. Far from
being completely dead, the set was
very much alive with a full raster on
the screen and a healthy hiss from the
speaker. Granted, there was no picture
or sound and I suppose, to the customer, it might just as well have been
completely dead.
Oh well, my fault for not being
more specific. But it did change the
situation somewhat. On the positive
side it appeared that the rear end was
working, particularly the horizontal
de
flection circuit and all that goes
with it. And that, in turn, suggested a
front-end fault.
Unfortunately, it also ruled out the
chances of a clear-cut fault, as in a
completely dead set. And that, in turn,
meant that there was less chance of a
quick fix and I advised the customer
accordingly. Again he was quite understanding and so he left me to it.
When I turned the set on again
some time later, it came up as before.
Then the phone rang and I turned the
sound down to remove the hiss while I
dealt with the call. When I eventually
returned to the set, it was displaying
a first class picture. What’s more, the
sound had also returned to normal, as
I quickly found when I advanced the
volume control.
So the fault was intermittent; the last
thing I needed when the customer was
hoping for a same-day job.
What I did need was a service manual – or at least a circuit. But I had
neither. The best I could dig up was a
circuit for a similar model, which I felt
might be sufficient for the job. In fact it
served very well, its main shortcoming
being that it lacked any waveforms.
Later – much later – I found a colleague
with the correct version, complete
with waveforms.
The fault returns
The set had been turned off while I
was searching for the circuit and, when
I turned it on again, it came up in the
fault condition. In fact, this was to be
the pattern; switch it on from cold and
the fault would appear. Then, after
anything from a few minutes to half an
hour, it would come good. Similarly,
once up and running, it would need
to be turned off for up to half an hour
for the fault to reappear.
This was something of a mixed
blessing. It was helpful to be able to
create the fault, almost at will, but the
half-hour wait each time was highly
inconvenient and time-consuming.
I left the set for half an hour or so,
while I attended to another, more routine job, then switched it on again. It
came up faulty and I quickly switched
it off. I then took the back off, pulled
the works out, and began finding my
way around the boards with the aid
of the circuit. And, since it appeared
to be a front-end fault, I concentrated
on the tuner and IF sections.
Next, I switched the set on again
and made some quick voltage checks
before it came good. And I hit it almost
in one; both the tuner and the IF section are fed from a 12V rail – which
didn’t have 12V on it. So that was it
–all I had to do was find out why there
was no 12V. And I was silly enough
to imagine that this would be quite
straightforward.
It was no problem to trace out the
12V rail on the circuit. It was a conventional arrangement, derived from
a tapping (pin 2) on the horizontal
output transformer (T502). From this
point, there was a 2.2Ω fusible resistor (R522); a diode (D503); a 4700pF
capacitor (C523) in parallel with the
diode; and the 2200µF main filter capacitor (C524) 2200µF. In short, it was
perfectly conventional and it looked
like a snack.
I checked the 2.2Ω resistor and the
diode but could find nothing wrong
with these parts. But I did suspect that
there might be a dry joint to one diode
lead, so I resoldered these and those
of the 2.2Ω resistor.
By this time, the set should
have cooled into its fault condition but, when I switched
it on, it came good immediately. This seemed like a
good omen but I have been
caught before in this situation. I turned it off for another
half hour to let it really cool
down. And, incidentally,
these half hour periods were
adding up; the day was slipping away and there wasn’t
much time left if this didn’t
fix it.
Unfortunately, it didn’t.
The set came up faulty as
before. So what next? The diode seemed the best bet and,
to save time, I simply tacked
another diode in parallel with
it on the copper side of the
board, crossed my fingers and
switched on.
The result was completely
unexpected. The set really
was completely dead now;
no raster, nor sound hiss, no
sign of life at all. After the
first shock, I did some probing
with the meter and eventually
realised that there now virtually a dead short on the 12V rail, with
only a couple of ohms to chassis.
And so began the laborious task of
tracing the 12V rail and isolating various sections in an effort to pinpoint
this fault. Naturally, as readers can
imagine, tracing this rail on the circuit
is one thing; tracing it in reality is
something quite different. It weaved
and wandered all over the place and
was almost impossible to follow in
places.
The only good point was that it
used a number of links and these
proved valuable in isolating various
sections. I think I lifted about five links
altogether and, including inevitable
interruptions, spent about two hours
tracking it down.
The faulty parts
The faulty components were associated with pin 38 of the jungle chip,
uPC1420CA. This pin is fed from the
12V rail via isolating diode D504. Also
connected to it is zener diode ZD501
and resistor R514 (12kΩ, 2W). The
other end of this resistor connects to
the 120V rail. This is the kick-start
network, which is needed to start the
June 1995 41
SERVICEMAN’S LOG – CTD
horizontal oscillator at switch on.
Both D504 and ZD501 were shot
(dead short) and this was what was
loading the 12V rail. Why had this
happened? I have absolutely no idea.
Naturally, I checked the substitute
diode and anything else that I might
have done wrong. I drew a blank on
all counts.
So all I could do was replace these
two components and try again. I
replaced the main diode (D503) and
removed the substitute diode I had
shunted across it, then I switched it
on again. Well, at least the set was
“alive” (raster and hiss) but there was
still no 12V. Then I realised that the
2.2Ω fusible resistor had done its job
and fused.
I fitted a new resistor and tried
again – still no 12V rail. I went over
everything again, checking and double checking, but could find nothing
wrong. But I did realise that something
else had happened; no practical warmup period would now cure the fault
and it appeared to be permanent.
Well, that could be all to the good.
And, having checked everything else,
the main suspect now was the horizontal output transformer, unpleasant
though this thought was.
With the frequency and waveform
involved here, the only practical way
to check this is with a CRO. But even
here I had a problem. As I mentioned
earlier, I was working from a circuit
which had no waveforms. So I had
only a very general idea of what I
would find on pin 2.
In fact, there was a waveform
there and its shape was not unreasonable. But I had no clue as
to what the amplitude should be
and it was rather beyond my grocery bill mathematics to work out
what it should be to deliver the
required 12V.
But I did suspect that it was
rather low, which only supported
my impression that there was some
kind of weird fault in the transformer. I finished up disconnecting
it entirely and making resistance
checks on all the tappings. They all
showed continuity and appeared
to make reasonable sense, at least
as far as I could tell without any
precise reference.
42 Silicon Chip
Finally, I pulled the transformer
out and checked it on the shorted
turns tester. Again I drew a blank.
Nevertheless, I had now convinced
myself that the transformer had to
be the culprit.
Good news & bad
On that basis, the next step was to
check availability and replacement
cost. A call to the NEC service department produced a good-news-bad-news
reply.
The good news was that replacements were available and the bad
news was the retail price of $166. For
most sets, the cost would range from
about $60 to $100, so this was a real
shocker, particularly as there was still
a niggling doubt as to whether I was
really on the right track.
But I had more or less committed
myself now, so it was up the motel
proprietor. I rang him, explained that
the job was going to take longer than
we had hoped and that it was going
to be quite expensive. By the time the
transformer price, labour and other
costs were added in, the bill would be
over $250. Did he want to go ahead?
He thought about it briefly, then
said, “yes, go ahead”. As he explained
it, there were a couple of factors involved. One was the alternative cost –it
would cost a good deal more to replace
the set and it was an essential item.
The other reason was more unu-
sual. When the motel had been fitted
out, the cabinet colours had been
specially chosen and supplied to suit
the decor (it was a light cream colour
that was not normally available). This
could be difficult and expensive to
replace.
So I ordered the transformer, which
arrived in a couple of days, and cost
another $8 freight. And from there
it was someth
ing of an anticlimax;
I fitted it, switched on, and the set
snapped into life with perfect picture
and sound. Of course, I gave it a thorough workout, with a routine of on-off
cycles over the next couple of days.
But it never missed a beat and hasn’t
missed one since.
Unanswered questions
So that was it; a faulty transformer.
The set went back to the motel and I
had a happy customer, in spite of the
cost. But, as readers will agree, it leaves
a lot of questions unanswered.
For a start, what kind of fault was
it? Remember, it produced what appeared to be a typical waveform at pin
2, even though there was no DC after
the rectifier.
The best suggestion I can make is
that it was some form of high internal
resistance, intermittent, and probably non-linear in some way. In other
words, it was incapable of supplying
any useful current to the load but could
still produce a waveform of sorts on a
sensitive CRO.
Further to that last thought, it was
only when the job was finished and
the set back in the motel that I found a
colleague with the correct circuit.
And it is the appropriate portion
of that circuit which is reproduced
here.
The waveform shown for pin
2 is essentially the same shape
as that which I observed for
the faulty transformer. But the
amplitude is another matter. I
didn’t take as much notice of it
as I should have but, as I recall,
it was nothing like the 120V p-p
as on the circuit.
And what about the destruction
of the diode and the zener diode?
This is an even greater mystery.
My best suggestion here is that the
substitute diode I shunted across
the original was faulty and was
breaking down at high voltage.
OK, so it’s a long shot. But I am
sure of one thing – if one such
faulty diode existed in a batch of ten
million, it would finish up in my spare
parts stock.
The microwave oven
And now for the second spot of
frustration. This involves a complete
change of scene; from a colour TV set
to a microwave oven, and an intermittent one to boot. This was a first for me.
Until now, I’ve had intermittent faults
in every device I can think of except a
microwave oven.
It started with a phone call from
a regular customer and concerns a
Panasonic model NN-9859. This is
a combination microwave and convection heating type and, in order to
appreciate the problem, it may help
to describe the operating procedure,
particularly for the convection mode.
Having turned the oven on, the
required temperature is selected by
pressing an appropriate key, which
increments the temperature indicator
in 10°C steps. When the oven reaches
the preset temperature, the system
beeps and flashes the temperature indicator. The oven is then held at that
temperature.
The customer’s complaint was that,
having gone through this procedure in
the convection mode, the oven would
behave normally for about five minutes
and then shut down. If the start button
was then pressed, it would run for
another few minutes, then shut down
again. This procedure might need to be
repeated several times but, eventually,
the oven would come good and run as
long as needed.
I immediately enquired as to whether this also happened in the microwave
mode, thus suggesting a common fault
area. But he couldn’t say; they seldom
used the microwave mode, only the
convection mode. The microwave
mode was used on the odd occasion
to reheat a cold meal but then the
time needed was probably too short
to create the problem.
So I said, “bring it in and we’ll have
a look at it”. And so it finished up
on the bench. I deliberately avoided
removing the covers, so as not to disturb anything, but simply switched it
on, set it up for a couple of hundred
degrees, and let it run.
And it ran perfectly; not a sign of
trouble. I switched to microwave
mode, added a jug of water as a dummy load, and tried that. Again, it ran
perfectly.
Fig.2: this drawing from the service manual shows the top of the
Panasonic NN-9859 with the cover removed. Note the temperature sensor
below the circulation fan pulley.
I turned it off, let it cool for a couple
of hours, then tried the convection
mode again. And this time it did misbehave; it ran for a couple of minutes
and then shut down. The temperature
display was still showing the correct
value and pressing the start button set
it off again.
And, just as the customer had said,
I had to do this two or three times.
Then it came good and ran up to the
selected temperature. I repeated the
test in the microwave mode and it
behaved perfectly.
I let it cool overnight and repeated
the tests the next day. The result was
exactly the same as before; intermittent
on convection, perfect on microwave.
On the face of it, it looked like a nasty
problem. And it could have been, had
I not serviced this model and earlier
models before. Which is not to say
that I had seen this problem before – I
hadn’t.
But I had encountered a fairly
common fault whereby the display
panel would exhibit a string of eights,
which meant that the oven could not
be programmed for either mode. And
the reason? An open circuit oven temperature sensor.
So, while the symptoms differed, I
went straight to this sensor. This looks
like a ceramic encased resistor and is
mounted on a ceramic strip. This in
turn mounts over an opening in the
top of the oven, with the sensor below
it. The sensor connections consist of
two flat metal lugs, to which are con-
nected leads which run back to the
microprocessor.
In all the units I had seen before,
these lugs were about 75mm long and
were encased in insulating sleeving
which extended back over the connecting leads. They were also bent
parallel with the top of the oven. In
this oven, however, the lugs were
only a few millimetres long and the
connections to the leads were plain
ly visible.
And so was the fault. Instead of the
usual welded or staked connections,
these looked as though they had been
soldered. But there was little solder to
be seen now. The lugs were blackened
and the tinned leads simply wrapped
around them. The wonder was that the
thing worked at all.
For a start, soft soldered connections
on those lugs simply do not make
sense. The oven is programmed up
to 250°C and would commonly run
at up to 200°C, so the sensor and its
lugs would also be heated to that level.
Against this, the melting point of 60/40
solder is around 190°C or even a little
less, creating a completely incompatible situation.
My bet is that it was bodgie repair.
The original sensor probably failed
and some smart type salvaged a sensor
from a ditched oven, clipping the lugs
short in the process. He then attempted
to solder the unsolderable, creating the
ultimate in dry joints.
I can’t prove it of course but it’s the
best theory I can come up with. SC
June 1995 43
|