This is only a preview of the January 1997 issue of Silicon Chip. You can view 24 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 "Control Panel For Multiple Smoke Alarms; Pt.1":
Articles in this series:
Articles in this series:
Items relevant to "Build A Pink Noise Source":
Items relevant to "Computer Controlled Dual Power Supply; Pt.1":
Purchase a printed copy of this issue for $10.00. |
SERVICEMAN'S LOG
The fireball TV set from hell
It might seem over-dramatic to describe a rather
ordinary looking NEC 51cm TV set like this
but this was indeed a wolf in sheep’s clothing.
And it lead me on a merry chase to discover the
cause of the problem.
The set, an NEC N4840 using a
Korean Daewoo C-50 chassis, was
brought in by a young, brusque woman
who was succinct and to the point: “it
smoked, burned and then went black”.
I barely got her name and address
before she left as quickly as she had
arrived.
I was rather busy at the time and
a couple of days passed before I was
able to examine the set. It didn’t seem
wise to switch the set on immediately, as her description of the fault
suggested that there may have been
a fire. As such, it would be all too
easy to exacerbate the problem and,
in any case, there would probably
be obvious visual evidence of the
damage inside.
And so it was that I gingerly removed the back, carefully examined
the various circuit boards, and tried
sniffing for the telltale smell of fire.
But there was nothing. The set, which
I guessed was about seven years old,
was reasonably clean and everything
look OK. I especially examined the
flyback transformer and power supply circuitry but all was fine. Perhaps
the damage was on the inside of the
deflection yoke but that would have
to wait for the moment.
Eventually, I concluded that there
was nothing for it but to go for a smoke
test – in this case, literally. I plugged
the set into the power and switched it
on. At this stage, I didn’t know what
to expect but the result was something
of an anticlimax. The set momentarily
spluttered into life and then died – no
sign of smoke or flames or anything
dramatic yet.
OK, so where to start? My first volt-
age check was at the collector of the
line output transistor. This measured
103V which seemed reasonable and
so I switched the set off and fished
out the file on NEC/Daewoo sets. Unfortunately, the only circuit diagram I
had was an abysmal photocopy with
virtually illegible component values
and type numbers. However, by using
a magnifying glass, I could just discern that the B+ rail was indeed 103V
which meant that the power supply
was probably OK.
So why was the set dead? Well, maybe because a safety circuit had turned
off the horizontal oscillator. I initially
confirmed this by measuring the collector voltage of the driver transistor
in this stage – it was floating at the B+
potential, which meant that it wasn’t
turning on and off. What’s
more, a quick check with
the CRO showed that
there was no waveform
on pin 20 of IC I501
(TA8718).
that was going too high.
To test the latter theory, I decided to
try using a Variac to reduce the mains
voltage before switching the set on.
Of course, the input voltage can only
be reduced so far. The set is remote
controlled and if the input voltage
is set too low, the microprocessor is
starved of voltage and will not switch
the set on.
Nevertheless, I ploughed on and
swung the input voltage up to 110V
AC. Before switching on though, it
is necessary to secure the PC board.
Normally, this is held into the front of
the cabinet shell by pressure from the
back. As a result, if you try to switch
the set on without the back in place,
the pressure on the mains switch can
be enough to push the whole chassis
back inside the cabinet without the
power actually coming on.
To overcome this problem, I held
the rear of the chassis and the very
edge of the PC board with one hand,
taking extreme care not to touch any
of the parts or copper tracks. By this
stage, I was really beginning to feel
relaxed about the repair and that the
Tricks of the trade
Because the circuit
diagram was so poor,
I could not discern
how the protection
circuit worked or
even where it was.
Basicall y, there
were two possibilities to consider: (1)
either the protection
circuit itself was faulty;
or (2) the protection circuit was shutting down
the horizontal oscillator
in response to a voltage
January 1997 69
Serviceman’s Log – continued
symptoms had been over exaggerated.
I was wrong.
When I pressed the power switch,
the set powered up . . . and up . . . and
up, until there was a terrific “crack”. I
jumped away, partly in response to this
“crack” but mostly due to an electric
shock that I received from the two
places I had been touching the set. And
it continued to crack and spark until
I recovered my senses sufficiently to
dive for the mains wall power socket
and switch it off.
It would hardly be an exaggeration
to say that we had too much voltage!
Now I was going to have to be much
more cautious. Obviously the EHT
was far too high and it was arcing
everywhere – even across the plastic
insulation and onto me. The shock
wasn’t severe, except perhaps to my
wounded pride.
EHT checks
After a fright like that, it was time
for some heavy-duty armour. After
checking that the .0056µF high-voltage capacitor across the line output
transistor was OK, I reached for the
70 Silicon Chip
EHT meter and connected it to the
EHT output lead at the ultor cap (ie,
where it plugs into the tube). I also
connected the multimeter to the B+
rail so that I could monitor this voltage as well.
There was no way I was going to
touch the PC board again. This time
I wedged the chassis into the front
of the cabinet with an old defection
yoke rubber positioner and turned the
variac down 100V.
Wearing a rubber glove, I switched
the set on and watched the meters.
Interestingly, the multimeter showed
that the B+ rail initially rose to +103V
for a second or so and then continued
to rise even higher to over 200V (fsd on
the meter). Similarly, the EHT paused
momentarily at about 22kV and then
rose to over 30kV, at which point it
began to arc everywhere and I had to
switch off.
I was confused. Why was the B+
rail OK in shutdown mode and why
was it rising so high until shutdown
occurred? Right now, I didn’t have any
answers to these questions but there
was one other worry; all this arcing
was bound to cause more damage to
peripheral circuits.
To overcome this problem, I decided to disable the line output stage
until I had sorted out the problem
with the B+ rail. Fortunately, this
is easy to do; all that’s required is a
jumper between base and emitter of
the line output transistor. This done,
I switched the set on again and to my
surprise the B+ rail rose to its correct
value of +103V and stayed there dead
steady.
By now I was really baffled. The
only theory I could come up with at
this stage was that the power supply
was somehow breaking down under
load. To this end, I replaced switch
mode IC I801 (STR50103) and resistor
R806 (470kΩ) as I had had problems
with that going high in other sets.
I also replaced C814 (1µF 160V) as
it looked suspicious and connected
another meter across the output of the
bridge rectifier.
Unfortunately, that didn’t cure it.
When I removed the shorting jumper
from the base of the line output transistor and switched on again, sparks
flew everywhere. Reducing the variac
below about 90V killed the set completely, while between 110V and 240V
the voltage across the bridge rectifier
rose to 350V. And, as before, the B+
rail and the EHT rose well above their
specifications and the set often closed
down.
I did manage to reduce the arcing
a little by cleaning around the ultor
cap with CRC 2-26 and by cleaning
around the CRT board but it was still
very hairy. But obviously, this was
fiddling at the edges and had nothing
to do with the real fault.
The EHT stage
My next approach was to replace
the spike suppression ca
p acitors
around the line output transistor but
this only showed that I was still miles
off the track. About all I could do was
temporarily fit some larger values to
reduce the EHT to a more manageable
27kV while I checked the components
around the flyback transformer.
Eventually, it got to the point where
I began suspecting the transformer
itself. Perhaps an internal insulation
breakdown was causing EHT to arc
onto the B+ rail? It certainly seemed
that way, although the CRO only
showed oversize (but otherwise per
fect) pulses on the collector of the
line output transistor. Nevertheless, I
felt sure that I was on the right track
at last and ordered a new transformer.
When it duly arrived, I wasted
no time in fitting it. Unfortunately,
it made absolutely no difference. I
subsequently fitted a substitute yoke
without result and, by this stage, was
becoming thoroughly fed up. So much
for my initial confidence.
Logical thought
It was time for some logical thought.
The crux of the problem was what
caused the B+ rail to go high? It was
time to take a closer look at how this
rail is derived and where it went.
In summary, the B+ rail is generated from a switchmode power supply based on transformer T802 and
switching IC I801. And pin 4 of I801
is connected via a diode to pin 2 of the
flyback transformer (T402). I did some
voltage checks and noticed that the
B+ voltage got higher as it got closer
to the flyback transformer – even on
the same track!
How was this possible? By now, I
felt sure that some sort of weird voltage doubling process was taking place
and if it wasn’t the diode itself that
was at fault it had to be a capacitor.
So I began hanging extra capacitors
onto the B+ rail at different points in
the hope of changing something but
to no avail.
I was about to give up when I noticed
that the circuit shows an electrolytic
capacitor (22µF 160V) between pin 4
of the flyback transformer and earth.
But what really caught my attention
was that no internal connection to
pin 4 was shown. Obviously, this
was wrong – pin 4 had to go somewhere, otherwise why connect a
capacitor to it?
Fortunately, the circuit of an NEC
N4845 circuit (Daewoo C-900 chassis) is similar in many respects and
this showed that pin 4 connects to
a tapping on the flyback transformer
primary. I removed the capacitor and
immediately noticed that it was leaking slightly down the positive lead.
Could this be it, at last? I was desperate. I soldered in a new capacitor, held
my breath and switched on. Hallelujah
– it worked! The B+ rail stabilised at
+103V and the EHT settled at 22kV,
even with 240V input.
Unfortunately, all that EHT arcing
had created a couple of extra faults,
although these proved easy to track
down. First, the picture came up as
an overbright raster. This was due to
the 10µF 160V electrolytic capacitor
on the +180V rail to the RGB outputs.
It had gone leaky and pulled the rail
down to about 70V (the poor beast
had nearly exploded from its trauma).
Secondly, the set suffered top vertical foldover and there were obvious retrace lines. This problem was traced to
the vertical output IC (I301, AN5515)
which had been damaged. A new IC
restored the set to full health.
The set was soak tested for a week
before it was whisked away by its
unknowing owner. I must renew my
life insurance.
Computer monitors
The next day started looking distinctly “computerish”, as three monitors were dropped in by the local
computer shop as soon as I opened the
door. As usual, they were extremely
urgent and their clients wanted free
quotes. To cap it off, no faults were
specified which is often par for the
course but can cause problems if a
fault is intermittent.
I don’t really consider “free quotes”
as being fair as most of the work is in
the diagnosis and not the actual fixing. After all, if you go to a doctor, he
charges you for the consultation, gives
you no guarantee and then you have
to go elsewhere to buy your own parts
(drugs). In the circumstances, the best
I can offer are free guesses.
Now that I repair so many monitors, I have set up two old 286
computers with VGA cards running
a test program by Koenig, as well as
a 386 with Windows 3.11 running a
program called Wintach. I also have
another 286 with an EGA card for
older monitors.
The three monitors were all only
two years old and were 15-inch digital
non-interlaced SVGA types. Two were
Moebius CM15VDE models and the
other a WEN JD156B. I began by connecting them to my three computers
and switched on.
One Moebius was initially working
OK, while the second one was giving a
“pink” picture. The WEN, on the other
hand, was completely dead –well, almost. I decided that the “pink-picture”
job would be the easiest and tackled
that one first.
The back was held on with two
screws on the bottom and two plastic
lugs at the top that are awkward to
unclip. Once this was off, I unsoldered
the metal screen over the CRT board
(PWB1787). It was obvious that the
problem was no green so I examined
this board for dry joints, glue, corrosion and cracks but all was OK.
The fault had to be somewhere in
this vicinity because the cable from
the computer connected directly to
CRT board, with sync pins 13 and 14
going off to the motherboard.
Next, I considered the possibility
January 1997 71
that the fault was in the cable itself.
With this in mind, the DB25M plug
was carefully examined for broken or
bent pins, with particular emphasis
on pin 2 (the green input). I could
find nothing wrong. I then checked
for continuity between pin 2 and the
CRT board plug (P502) at pin 3 and
again all was OK.
Voltage checks
My next step was to make some
voltage checks around the CRT board.
First, I checked the voltage on the
green cathode (pin 6 of the CRT socket), then the red and blue cathode
voltages (pins 8 & 11). The latter both
measured about 70V, whereas the
green cathode voltage was at 60V. This
was rather puzzling – I had expected
the green cathode voltage to be higher
than the other two, because the green
gun was being cut off.
Because these voltages were not
unreasonable (after allowing for grey
scale adjustments), and because there
were no signs of any distressed components around the LM2419T RGB
power amplifier IC, I concluded that
the problem was back around the decoder IC (I501, MM1203). It was time
to fire up the CRO.
Immediately, it was obvious that
there was no signal on the green channel. There was no sign of a signal at
the input to the decoder IC or even
where the plug connects to the CRT
board. There just had to be a short
somewhere that was pulling the green
signal down. To test this theory, I
shorted the red input to the green one
and the red immediately dropped out.
Similarly, when the blue input was
shorted to the green input, the blue
dropped out.
An ohmmeter test between the green
cathode and ground subsequently
confirmed the existence of a short.
All I had to do now was track it down.
I began my search by checking all
the decoupling components to the
green input but they were all OK.
However, when I unplugged the connector to the CRT board, the short on
the board vanished. Obviously, the
problem was either in the cable or in
the DB25M plug.
I suspected the plug at first as this
Fig.1: the NEC
N4845 circuit
(Daewoo C-900
chassis) is similar
in many respects
to the N4840,
particularly
around the line
output stage. Note
the capacitor
connected to pin
4 of the flyback
transformer.
72 Silicon Chip
is often abused. Unfortunately, it is
directly moulded to the cable and
wiggling it while checking between
pins 2 & 7 with an ohmmeter made
no difference – the two pins remained
shorted.
Adjacent to the DB25M plug is a
cylindrical assembly – probably a
ferrite ring core – then there is a metre
of cable before it goes through a plastic
clamp on the back of the monitor. After that, about 15cm further on, there
is an earth clamp around the striped
cable braid, then another ferrite core
before the plug to the CRT board. It
all looked OK and nothing I could do
would clear the short.
Getting a replacement cable probably wouldn’t be easy, so I tried one last
gamble – I connected a variable power
supply across pins 2 and 7 and wound
it up in the hope it might burn off the
short. It didn’t work; the current rose
to 5A (the supply limit) with no sign
of the short melting. But what was
interesting was that the cable became
warm only as far as the entry clamp
but no further.
That just had to be the location of the
short. I removed the cable, ringbarked
the outer sheath on either side of the
clamp marks and carefully opened
the braid. To cut a long story short, I
eventually found a small nick in the
green signal cable which allowed the
inner conductor to short against the
outer braid.
After that, it was a simple job to correct the fault and refit the cable. And
that fixed the problem – the green was
fully restored and the display returned
to normal.
Two to go
By this time, Moebius No.2 had
decided to show its fault which was
a very dark display. On the bench, the
tube filament read only 2V instead of
6.3V RMS, so all I had to do was find
out why.
I traced the source of the voltage to
the +6.3V rail off the main chopper
transformer and it measured OK all the
way from there to a plug designated
P501a-1. From there, it went to P001-2
on a small “power saving” board and
then from P001-1 to the CRT socket
board. And the 4V was being lost on
the power saving board.
The power saving circuit includes
transistor Q003 (2SD667). The 6.3V
rail goes to its collector and the output to the picture tube filaments. The
An hour later, I had another
look at it only to find that it
was dead and that the power
supply was oscillating again.
Obviously, my choice of a substitute line output transistor
hadn’t been a good one. There
was nothing for it but to order
the correct transistor. It arrived
within a week, was duly fitted
and fixed the problem.
It had really all been a piece
of cake so far. Now for the really
difficult part – the “quotes”.
The three monitors had taken
nearly all day in labour time
and estimates of $82.50, $90.00
and $155.00 were given for each
job in turn. The first two were
accepted readily but the owner
of the third monitor baulked at
the cost. Later, on discovering
the cost of new one, she chang
ed her mind and decided to
proceed with the repair.
base is controlled by IC I002-6 (MC
14551BCP). As well, there are two
other transistors, an SCR and a second
IC (I001, HA17555). I checked Q003
and it was OK
Because the set had worked initially,
it appeared that the fault might be heat
sensitive and so I decided to try the
freezer approach. And I was rewarded
with instant success – when I sprayed
C001, a 470µF 16V electrolytic, the
picture returned to normal.
Replacing the capacitor made the
cure permanent and a soak test revealed no further problems. So two
down and one to go.
The WEN monitor
Fortunately, I had dealt with WEN
monitors before and already knew
about their energy saving functions.
In greater detail, this model will shut
down when not connected to the computer and will also shut down under
software control. However, this one
was almost totally dead when connected to a computer, the only sign of life
being a high pitched whistle.
Fairly obviously, that high pitched
whistle was coming from the switch
mode power supply which was closing
down because of a short circuit. On
the bench, I managed to locate the line
output transistor (Q404, 2SC4924) and
found that it was shorted. But it wasn’t
going to be that easy.
First, access to this transistor is very
poor. There are two side PC boards
and getting at the transistor mounting
screws from the lefthand side involves
removing the chopper FET (Q801)
and its heatsink, as well as C304 (a
2200µF 35V electrolytic). After that,
the transistor can only be reached by
moving the CRT board which is glued
securely to the CRT itself.
In fact, the CRT board required
considerable force to prise the socket
off the neck of the tube. Fortunately,
I managed to do this without breaking anything but I cannot say I was
impressed. Nor was I impressed with
the general quality of the soldering on
any of the boards.
The next challenge was to come up
with a suitable line output transistor.
My catalogs only went up to 2SC4700
and, as with the two Moebius monitors, I had no circuit and no data. The
nearest I could get lay my hands on
was a BU508DFI which was worth a
try. I fitted one and reworked all the
dry joints I could see. When I switched
it on, there was power and EHT but
still no picture.
I had postponed tackling the CRT
socket board because it was enclosed
in a metal screen. When I removed it,
I saw that its solder joints were even
more horrendous than in the rest of set.
Anyway, resoldering the CRT socket
connections restored the picture, so I
replaced the covers and put it aside
to soak test.
Another monitor
Later that same afternoon, another
monitor came in. This time, it was
a Videocon 14-inch mono VGA unit
(model T-14MS31) and, according to
its owner, it was smoking.
When I opened it up, I found that
two electrolytic capaci
tors had exploded, leaving small bits of paper
everywhere. Fortunately, I found the
metal/plastic covers and was able to
identify their values. One was a 2.2µF
100V bipolar capacitor (C523), used as
a yoke coupler, while the other was
a 1µF 160V electrolytic (C522). The
former is hard to get, so I fitted a 2.2µF
450V electro and a 1µF 250V electro
and switched on.
The picture was good, so I cleaned
up the gunk that was all over
everything, reworked a few suspect
joints, fitted the cover and left it to
soak test. The next day, after it had
been soak testing for a few hours, there
was a loud bang, followed by a hiss.
It had blown up again, destroying the
same two capacitors.
This time, I chose a high-current
2.2µF 400V polypropylene capacitor
for the yoke coupler and replaced
C522 with the same type as before. I
left it to soak test for two days before
calling the customer and telling him
that it was ready. He was grateful for
the speedy repair but I did wonder if
the service cost was worth it for an old
SC
monochrome monitor.
January 1997 73
|