This is only a preview of the August 1991 issue of Silicon Chip. You can view 41 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:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
The case of the blowing fuses
I'm kicking off this month with a mystery story
from J. L., our regular contributor from
Tasmania. I've chosen this sequence because,
after reading his story, I had a chance to work on
a couple of similar chassis and make some
further observations. They may help a little.
What follows in in J. L.'s words.
Here 's how he tells it.
This is a story without any real
ending. It has had me thinking for a
long time and I still don't have acertain answer. I can make a convincing
guess as to the cause but I'm not sure.
See what you think.
The set was a General Electric
TC53L2, a 53cm model, fitted with an
Hitachi NP6A-A chassis. I don't think
that has anything to do with the fault
or the cure. It could have been any
model of any brand of similar vintage.
The set came in for an intermittent
"no go". This model offers two possibilities for this fault. One is a faulty
joint under the clamping screw for
the horizontal output transistor collector. The other is a loose connection
to the emitter of this transistor (it uses
a slip-on connector which is apt to
come loose). Either fault is easy to
cure and, in this case, I replaced the
slip-on connector with a hard wired
lead, soldered to the emitter pin. It
was a total cure - for that fault.
The customer returns
Unfortunately, the customer didn't
agree. He was back within a week for
what he said was "the same trouble ".
It wasn't though; this time it was the
main HT fuse, F903, a 1A time delay
type, that had gone open circuit.
.The fuse had blown gently. It wasn't
just broken - as through old age - nor
had it blown violently. But it had
blown through over-current and I
needed to know why.
This chassis uses an isolating transformer, T951 , feeding a bridge recti60
SILICON CHTP
fier and then a chopper type regulator
delivering 125V to the rest of the set.
Fuse F903 is between the positive
output of the bridge rectifier and the
chopper transformer, T902.
Apart from the unlikely event of a
short to chassis in the chopper transformer, a shorted chopper transistor,
TR906, is about the only thing likely
to take out the fuse.
The only other connections to this
rail are C910 (a 4. 7µF 400V electrolytic kickstart capacitor) and the R908/
R935/R909 network which supplies
HT to the chopper pre-drive transistor, TR904. Both these connections
TETIA TV TIP
Hitachi CEP288, CEP289
(PAL3-A chassis).
Symptom: Reduced height,
about 2cm of black at top ~nd
bottom of screen. No colour. The
picture can recover to normal after 10 minutes but the fault does
not appear to be heat sensitive.
Cure: C753 (100µF/25V electro)
open circuit. This cap is the input
to the filter on the 20V rail and its
loss causes the rail voltage to
drop, in this case to 14V. The rail
shows no unusual ripple, just a
lower than normal voltage.
TETIA TV Tip is supplied by the
Tasmanian branch of the Electronic Technician's Institute of
Australia. Contact Jim Lawler, 16
Adina St, Geilston Bay, 7015.
are quite high impedance, so a short
to chassis is an unlikely result if any
of these components breaks down.
So, after considering all these
points, I decided that it had to be a
leaky chopper transistor. An in-circuit test indicated that the transistor
was OK but, in the absence of any
other indication, I felt that it had to be
faulty in some way or other. So out it
came and in went another one.
I fitted a new fuse and switched on.
Up came a perfect picture - for about
an hour. The new fuse then failed just
as the first one had done.
I tried again and again but the fuses
lasted from only 15 minutes to an
hour before failing. I checked everything I could think of that might be
overloading the fuse but every voltage or resistance that I tested appeared
to be well within tolerance.
Thermal cutout
By this time , I was running out of
fuses, so I firkled about (good word,
that) in the junkbox until I found an
old 1.5A thermal cutout, rated to trip
at 3A. With this clipped into circuit,
the set ran for many hours without
any trouble.
As far as I could tell , nothing was
overheating, there were no excessive
voltages, and there was no sign of
incorrect picture geometry. The set
seemed to be operating perfectly
within normal limits.
With everything apparently normal,
I guessed that it would be OK to refit
the correct 1A fuse. But I was wrong.
It lasted only 10 minutes. So I fitted a
1.5A fuse and, as far as I know, the set
is still going strong.
I have racked my brain trying to
work out what could have caused the
trouble. I'll swear there was nothing
wrong with the set, yet it would not
work with the correct fuse fitted.
The theory I have come up with is
this. See if you agree.
Most domestic electrical equipment
is designed around component val-
lation and that he contact me at any
sign of overheating. Somehow, I don't
think we'll have any trouble.
Another explanation
.,
t.H~~prr
WMII-ZZPH
(&TC
~r~
The £use had broken. gently. It wasn't
just broken-nor had. tt blown vio~Uy....
....Bu.t it had blown lhrough overcurrent
,...._ and 1 needed to know why
ues with a ±20% tolerance. Some parts
have a closer tolerance but many are
20% because anything closer would
be unnecessarily expensive.
The law of averages dictates that
the tolerances in an average set will
be spread evenly between the upper
and l_o wer limits. But occasionally
there must be a set that gets a preponderance of plus components; or of
minus components.
Fuses, on the other hand, must always have a positive tolerance. The
designer must select a fuse value above
the steady state current in the circuit
to be protected. Quite obviously, a
negative value fuse would blow every
time the set was turned on.
We can assume that the designer
selected a fuse that would have a safe
margin over the steady current, but
not so much over that it would be too
slow to act in the event of an overload.
But what would happen if all those
20% component tolerances accumulated in the direction that increased
the normal current in the fused cir-
cuit? The set would still work normally and the fuse would continue to
carry the required current, but with
less tolerance to an overload.
And, finally, component values
change as the set ages. What would
happen if these changes accumulated
in the direction that added just marginally to the circuit current?
The fuse could no longer stand the
strain and would pop after only a few
minutes. I think this is what happened to the General Electric set that
inspired this story.
The problem is, what am I going to
do about it? The cause is probably the
accumulated result of a milliamp or
two of extra current in every resistor
in the set. Restoring them all would
be a prohibitive job.
On the other hand, the total current
being used by the set does not seem to
be enough to raise the temperature of
any part of the chassis, so is the slight
overload in any way dangerous? I
don't think so.
I have suggested to the owner that
he ensure that the set has good venti-
OK., so that's J. L.'s story. In answer
to the implied question as to what I
think, I'm afraid that, to coin a phrase,
"I dunno please".
J. L.'s theory is an interesting one
but I have some reservations about it.
For one thing, I question the 20%
tolerance figure. This was true in the
bad old days of moulded muck and
crude carbon resistors but 5% has
been a generally accepted figure for a
some time now. And, in most cases,
the product is well within this limit.
But OK, let's accept 20% for the sake
of argument.
I held this story for some time after
I received it, hoping that a similar
model would turn up on my own
bench. Sure enough, not one but two
came in - a TC53L2 (53cm) model, as
above, and a TC63L1 which is a 63cm
set using virtually the same chassis.
Unfortunately, they didn't help
much. The first thing I did, once I had
a chassis working properly (more on
this later), was to measure the current
through F903. This came out as
275mA for a black screen and 370mA
for a full white screen.
But even the full white value is
only a little over one third of the fuse
rating, with some margin due to the
slow-blow characteristic. On a steady
state basis, that doesn't fit in very
well with the 20% theory.
And that suggests a surge of some
kind. Two possibilities come to mind.
One is a switch-on surge which, while
not quite heavy enough to take the
fuse out immediately, weakens it so
that it lasts only a few minutes.
The other is an intermittent fault perhaps thermally sensitive - which
takes the current just above the fuse
rating, then clears itself in the time it
takes to replace the fuse.
Of course, it is easy to propound
such theories but quite another to
prove them. In practice, of course,
few of us have anything like the necessary facilities; nor do we always
have the time to tackle subtle faults
like this. So thanks for the story J. L.,
but I say again, "I dunno please".
A real swine
So what about the two General Electric sets on my own bench? One of
AUGUST
1991
61
Australian Made TEST
EQUIPMENT YOU CAN TRUST
SERVICEMAN'S LOG - CTD
SHORTED Tu
Built-in Meter to check EHT
transformers including split
diode type, yokes and drive
transformers.
$78.00 + $3.00 p&p
DEGAUSS
Strong magnetic field,
larger than usual coil with
multicore centre . Double
insulated for your safety,
also fitted with momentary on/off switch. 240V
NC 2.2 amps.
Just about as important as having a soldering
iron in your toolbox!
$75.00 + $10.00 p&p
HI-VOL1'A
Built-in meter reads positive
or negative 0-50kV. For
checking EHT and focus as
well as any other Hi-tension
v.Jltages.
$98.00 + $5.00 p&p
REMOTE CO T 0
TESTEH
(INFRARED OR
ULTRASONIC)
Designed to test
lnfrared
or
Ultrasonic control
units. With the
extension lead
you can also test
infrared units which cannot be placed in
front of the testing unit. Requires a 9V battery.
Output is via the LED diode and piezo
speaker .
$85.00 + $4.00 p&p
LASER DETE:.C
P 0
A new addition to the remote control tester.
Comparable with units costing $500 or
more. You can test the laser pick-up in
compact disc players.
$27.00 + $2.00 p&p
NEW PRODUCT
TEST TUNE
Save time and money with
this tester. Helps you determine if you have a tuner problem , an AGC problem or a
fault in the IF stages.
$250.00 (Tax inc.)
TUNER
IS
from $17.00 exchange+ p&p
Cheque, Money Order,
Bankcard or Mastercard
216 Canterbury Road, Revesby,
NSW, AUSTRALIA, 2212
~
(02) 774 1154
~
(02) 774 1154
62
SILI CON CIIIP
these, the TC63Ll, turned out to have
a real swine of an intermittent. It lead
me µp several garden paths and was
finally cured more on a brute force
than scientific basis. Nevertheless,
some of test routines are worth noting
for reference.
The customer used the set in a holiday cottage up the coast, and his complaint was that "it stops now and
again". In greater detail, it turned out
that the failure was fairly infrequent;
it would run perfectly for days, or
even weeks. And, when it did fail,
normal operation could be restored
by simply flicking the on-off switch.
This had been going on for some
considerable time but, while ever the
set responded to this simple treatment , the owner was prepared to live
with it. However, the day came when
it didn't respond, at least not immediately, and a goodly chunk of an interesting program was lost. That was
when the owner decided that something had to be done and it landed on
my bench.
Of course, it started as soon as I
switched it on and ran for several
days. And when it finally did fail,
and on subsequent occasions, I found
it difficult to make any worthwhile
checks. It would come good at the
slightest touch but I did manage to
establish that there was no 125V HT
rail out of the power supply when the
fault appeared.
I had no doubts that the fault was a
dry joint; this chassis is notorious for
them. These dry joints are found
mainly on the power supply board
and on the horizontal section of the
deflection boards.
And this is what makes it hard. The
chopper/regulator system in the power
supply board (TR903, TR904, TR905
& TR906) is driven with puls es from
the horizontal output stage. But the
horizontal system can't deliver these
pulses until the power supply delivers voltage to the HT rail. And the
power supply can't deliver this voltage until it receives pulses from the
horizontal system.
In practice, this deadlock is broken
with a kick-start system; a
multivibrator circuit consisting of
TR901 & TR902 which is activated
briefly from the bridge rectifier via
C910. At switch on, it delivers a few
pulses to the "pre-drive" transistor,
TR904, to get things started.
This is a fairly universal technique
and is all very clever. But when the
system collapses and there is no HT
rail , there is nothing to indicate
whether the fault is in the power supply or in the horizontal system.
After several abortive attempts to
get any kind of a lead, I settled for a
routine search for obvious dry joints.
I concentrated mainly on the power
supply board and, in particular, a
number of 2W and 5W resistors, such
as R908 , 909 , 935 , 928 and 924.
In order to aid heat dissipation,
these resistors are mounted clear of
the board, supported by short metal
tubes which go through the board to
the copper side, where they are soldered to the copper pattern. The resistor pigtails go down these tubes
and are, supposedly, soldered to the
them during the flow soldering process.
Unfortunately, this doesn't always
work. One problem is that the pigtails
are sometimes cut short and the solder doesn't reach them. But even when
the pigtails are full length, the bond
between them and the tube, or between the tube and the copper pattern, is often poor.
So one of the routine jobs with faults
like this is to go over all these joints
and resolder them. Having done this,
I checked the rest of the board and
resoldered a few other suspicious
looking joints just to be sure.
I ran the set for several days and it
behaved perfectly. But it had done
this many times before and I needed
more proof than that. By this stage,
however, some six weeks had elapsed
and the owner came in to check on
my progress. More specifically, he
wanted to see whether it would be
available for another stint up the coast.
I explained the situation and emphasised that I could make no claim
to having cured the fault. Nevertheless, he was keen to take it and give it
a try, so I said, "OK, but be warned".
It's not cured
That was the last I saw of it for
several months. Then, suddenly, the
owner turned up with it again. It was
TR903
2SC458rRI
PHASE
AMP
C917
0.0068
1 R914
1
180
1
J2
R940
220K
D
•;;:z.(H'r:-<r"U>- ·c&t',
'
i.
.
J
Fig.I: the power supply board for the GE-TC63Ll & GE-TC53L2 TV sets. The output from the bridge rectifier is
applied via fuse F903 (lower left) to isolating transformer T902 (top) which is switched by chopper transistor
TR906. TR901 & TR902 form a multivibrator circuit which is briefly powered up at switch-on to deliver pulses
to TR904 to kick-start the chopper circuit. TR908 & TR909 provide over-voltage protection & this circuit can be
disabled by disconnecting R941.
the same old story; initially, it would
run for a long period, fail , respond to
the on-off routine, and run for another long period. It had now again
reached the stage where it failed to
respond to this treatment, even after
repeated tries.
I switched it on while the owner
was there. And, yes, it was completely
dead. So I felt that, at last, I might get
to grips with it.
Another bonus was that it was not
likely to be needed again for several
months and I could take m y time. So I
put it aside for several days due to
pressure of other jobs. And that was a
mistake; when I did finally switch it
on, it came good immediately. Back to
square one.
This time I decided to try a different approach. The trick is to bypass
the chopper/regulator section of the
power supply and run the set directly
from the bridge rectifier. If it fails in
this configuration, the fault is almost
certainly in the horizontal system and,
in any case, with power still applied,
one has a better chance to track it
down.
It's a simple trick. The set is fed
from a Variac and a jumper lead is
used to connect fuse F903 to the cath-
ode side of diode CR908 (top right of
circuit, n ear T902). The Variac is then
wound up until the voltage at this
point reaches 125V - or a little less to
provide a safety margin.
I set everything up to enable me to
quickly do this and then left the set
running, waiting for it to fail. Eventually, it did and, what's more, it refused to start using the off/on technique. This was the ideal opportunity
for the bypass trick - I quickly connected the jumper lead and wound
the Variac up, whereupon the set leapt
into life. And it kept on going.
From this I deduced that the fault
AUGUST
1991
63
SERVICEMAN'S LOG - CTD
And. yes ... it was completely dead ...
was in the power supply and spent
some time going over the board again,
looking for an elusive dry joint which
I might have missed the first time. I
drew a blank.
There are a couple of other tricks
one can try in this situation. First, by
bridging capacitor C910 with a ·1kQ
resistor, the multi vibrator can be made
to run continuously, regardless of the
condition of the horizontal stage. This
allows the power supply to be checked
stage by stage until the fault is located.
But there is a point to watch here. If
the rest of the set is not drawing current, the over-voltage network consisting of TR908 & TR909 will shut
everything down. This can be prevented by temporarily disconnecting
R941.
A simpler trick is to repeatedly
switch the set on and off, to activate
the kick-start system, and use a CRO
(ideally a storage type) to check for
the pulses, stage by stage, up to TR906.
64
SILICON CHIP
This was what I did and I managed to
confirm that TR906 was indeed receiving these pulses.
Murphy's lunch
By now, of course, the set was running again. The next time it failed, I
again reverted to the jumper lead/
Variac setup. And that clinched it; the
set refused to run. The fault was not
in the power supply. (At long last I
had caught Murphy out at lunch).
I restored the power supply to its
normal configuration, then moved
over to the horizontal system. Unfortunately, we don 't have sufficient
space to reproduce the circuit, which
is quite extensive. It involves two separate PC boards; the Deflection Chassis
Board and the Deflection Output Chassis Board. The horizontal oscillator
(TR703) is on the first board, while
the driver stage (TR704) and the output stage (TR707) are on the second
board.
One of the nasty aspects of this part
of the set is the mechanical setup.
The Deflection Output Board is
mounted vertically on the r-ight hand
side of the set, copper side out. And
mounted on the copper side, supported on spacers, is a large heatsink
carrying the output transistor, TR707.
This creates a problem because all
solder joints under this heatsink are
completely inaccessible. And at least
two components in this area, R721
and R722 in the base circuit of the
horizontal driver transistor (TR704),
have a reputation for dry joints.
Another item obscured by this heatsink is a wire-wrap pin inserted from
the component side and soldered to
the copper pattern. The wire-wrap
lead from it runs to a similar wirewrap pin on the Deflection Chassis
Board. From here, the circuit runs to
the collector ofTR703, the horizontal
oscillator.
All of which is by way of background. Having exonerated the power
supply, I connected one CRO lead to
the collector of the horizontal oscillator (TR903) and another to the base of
the driver transistor (TR704). I then
switched the set on several times to
active the kick-start circuit. This produced a brief burst of voltage on the
HT rail, sufficient to active the horizontal oscillator and produce a short
burst of oscillation. So far so good.
But there was nothing at the base of
TR704; the fault was between these
two points. I was getting closer but
the exact cause still had to be found.
And considering its intermittent nature, it still looked like there was a
long haul ahead.
I lifted the heatsink clear of the
board and made a visual check of the
path, joint by joint: Each appeared to
be perfect but since this was obviously not the case, I went over each
one and resoldered it. That done, I
did the same around the oscillator
section, involving TR703.
And that was it; the set hasn't
missed a beat since, which was something of an anti-climax. Which joint
was it7 I can't be sure but I strongly
suspect one or both of the wire-wrap
pin joints; mainly because the solder
seemed to come away from these much
too readily.
And the TC53L2 set? It had a very
common fault involving the over-voltage protection circuit, TR908 and
TR909. I replaced both transistors and
that was it.
SC
|