Items Covered This Month Sony KV-W32MN21/BE32 TV set (SCC-J45A-A AG-1 chassis). Panasonic NV-SJ400 VCR. Philips 29SP1698/75R TV set. Selectronics SPI 1200-SS 1200W inverter.

Fix the roof, then fix the TV.

Mr Wilson (not his real name) is a barrister and lives in a beautiful harbourside mansion. Unfortunately for him, his mansion didn't have a perfect roof - something he didn't realise until, one evening, a bad storm forcibly brought it to his attention.

There was a leak and it was right over his beloved 7-year old widescreen 32-inch (76cm) Sony KV-W32MN21/BE32 (SCC-J45A-A AG-1 chassis). This set was the state of the art back in 1995 - the best money could buy - and it was totally consistent with Mr Wilson's "elevated" station in life.

When faced with this disaster, Mr Wilson moved as fast as he could to (a) switch the set off and (b) move it out of harm's way. But not knowing the extent of the water damage inside the "telly", he sensibly left it to dry out for a few days while the leak in the roof was repaired.

When he finally thought it was safe to reconnect it and switch it back on, his prayers went unanswered - after all, appeals to a higher court don't always work, not even for barristers. Apparently, there was a "phut" noise and a little "unmusical" electrical theatre before it all went silent. The only encore was a trickle of smoke and an unpleasant burning smell drifting up out of the ventilation holes at the rear.

Ever the optimist, he thought I could fix this in his home. Unfortunately, at that time, I didn't even have a circuit diagram. Still, I went through the motions of removing the back and inspecting the damage.

It was major - the water had leaked inside on a wide front from the deflection board (D) to the small signal and switchmode power supply board (A), plus a multitude of modules plugged into these along the way. Also, there was older corrosion from sea air along the top of the double-sided PC boards, particularly near the flyback transformer, so there was lots of damage.

Because of the likely cost of replacement parts, I was happy to declare the case hopeless and advised Mr Wilson to talk to his insurance assessors. To my amazement, it turned out he didn't have a policy to cover this. Not only that but he insisted that he wanted it fixed - apparently it had "sentimental" value and it matched his minimalist (!) decor and the colour scheme in the room.

I found his reasoning unconvincing but mine is not to reason why. And so, reluctantly, I agreed to fix it but I warned him that it could be costly.

It was a huge struggle to lug the 73kg (900 x 600 x 584mm) set to my truck but we made it without having to consult a cardiologist. Back at the workshop, I had another tussle to get it onto the workshop bench before I could finally tackle the destruction wrought by the water.

I started by dismantling the chassis, separating it into superficially damaged, severely corroded and damaged, and unaffected modules. It was soon clear that the power supply boards F1 and A were blown and that the deflection board was severely corroded and burnt.

I removed all the obviously burnt parts and then washed the boards to remove the corrosion. It was a long process, starting with scrubbing the boards with Nifti and then flushing with clean hot water. This was then followed by long periods of drying before using chemicals such as methylated spirits, CRC2-26, PC board cleaner and electrical cleaner to remove any residues and expel any remaining moisture.

I also had to "gouge" out areas where the board had deep ruts burnt into it. This process was tackled part time over a period of several weeks while the damage was diagnosed and parts ordered.

On the AC rectifier board (F1), R1602 (0.1Ω0.5W) had vaporised and D1602 - a TF5415 power switching SCR - had gone short circuit. Naturally, the F3401 mains fuse (5AT) on board F2 had failed as well.

I wasn't sure about IC1601 (the STR81159A rectifier switch) but it looked OK so I left it for the time being. Later, I decided it was too risky not to replace it, as it might do more damage to the rest of the supplies, so I ordered a new one.

Next, I moved to the A board and took a closer look. The power supply occupies only a small part of this module (towards the front of the set) but it is packed with components on both sides, many being surface mounted. I measured both FETs Q605 & Q606 (IRF1840G-LF, CONV-OUT) as being short circuit and I also checked D601 (a protection diode) and zener diodes D605, D606 and MA8180M. To be on the safe side, I decided to replace them, along with IC601, IC602, IC651 and IC652, plus some of the surface-mounted transistors which didn't look too good from all the corrosion.

On the deflection board there was a nasty mess near plug CN515, which supplies high voltage to the CRT board (C). R540 (a 1.8kΩ feed resistor for the +1000V supply) and R546 (a 0.47Ω feed resistor for the +200V rail) were both burnt and were open circuit. The corresponding diodes were OK but the PC board was charred. The whole board had a lot of corrosion on top as well but nothing that wasn't linkable or fixable.

The required parts eventually arrived and I informed Mr Wilson of my estimate. It always amazes me how rich people can whinge so much about bills when no doubt mine would pale into insignificance compared to one of his. Nevertheless, when he got over the shock, he agreed to go ahead.

Chinese puzzle

I fitted all the parts and reassembled all the boards. This was a bit like a Chinese puzzle and CN101 was hard to find, being tucked up at the back behind all the plastic support brackets. The problem was that the plugs are unmarked and often the colour doesn't match the socket. In fact, you need the circuit to prevent plugging the wrong plugs into the wrong sockets.

Finally, I rechecked all my work, braced myself and switched it on.

The main thing was that there was no melt-down. Surprisingly, I had fixed the major parts of the power supply and the set was working after a fashion. I heard the EHT come up, the sound was good and I measured +135V on R3501 at the rear of the deflection board. However, I still had no picture.

It was hard to see the CRT filament heaters, so I measured them with a true RMS meter. I also measured the +1000V and +200V to the CRT board (C). I then momentarily shorted one of the tube cathodes to ground and got a flash of a fully scanned raster. But what was more interesting was that just by touching the cathode colour!

The quality wasn't brilliant though and I began wondering whether the picture tube was suffering from low emission. When I switched the set off and then on again, the picture disappeared but it could be restored by just touching anyone of the cathodes.

This was bizarre but seemed to be consistent with a low-emission picture tube affecting the RGB cut-off circuit via CN703-7 "IK OUT". This line should be at approximately 2.2V.

If the tube was flat, then increasing the heater filament voltage should also increase the beam current enough to overcome the cut-off circuit. I did this and also varied the voltage on the IK-OUT line but nothing made any difference.

Next, I removed the CRT board and examined it carefully. It was then that I noticed that it too had suffered from corrosion. I thoroughly cleaned it, replaced some of the transistors, particularly in the red and green amplifiers (Q703, Q709 and Q702, Q708), and thoroughly checked everything else. Nothing changed so I also replaced Q714 (spot suppressor) and zener diode D714 but again drew a blank.

So how could touching the cathode of any gun switch on the picture? Was I injecting a signal, was it the extra capacitance or was I somehow increasing the beam current? I was still sure it was the IK-OUT line and I hooked up the CRO on and checked the waveforms on it. Without a picture, there was no waveform and its DC voltage was just a fraction lower than normal. Conversely, when the picture was present, there were large line pulses.

It really didn't get me anywhere though and I concluded that I was looking at the problem the wrong way.

Finally, I did what I should have done a lot sooner - I measured the voltage on every pin of the CRT, including the screen voltage to pin 3. This voltage was only at 225V whereas the circuit says the range should be 227-858V.

Having established that, I switched the set off and measured all the resistors around the screen control, including the control itself - all were correct. I then switched the set back on and ever mindful of how hard it is to align this control these days (see the service manual set-up adjustments), marked the position of the screen control knob. I then advanced it and as I did, the picture came up much better than before, with none of the colour bleeding that had previously been evident.

I experimented with a number of positions and finally settled for a voltage of around 430V, which gave the best picture.

So why didn't I follow the service manual's alignment instructions. The reason is that you have to: (1) set the picture to normal (define normal!); (2) set the video input to AV with no signal; (3) set the unit to Service Mode and turn off blue and black; (4) chase the BOF data of item number 1A (Auto Cut Off) from 00 to 01; (5) connect an oscilloscope to each of the cathodes and adjust (range 00-3F) 0E R-G, 0F G-G, 10 B-G Red, Green and Blue gain for a waveform that is 170V DC each above ground. And then it says "adjust G2 (RV701) volume to make the screen slightly bright".

Good, isn't it? You make all these careful accurate scientific adjustments to finally adjust it subjectively by eye! I skipped straight to the last line.

Anyway, this was finally all that was required to give it a crystal clear picture. I checked all its trick functions such as Text, Picture in Picture and 16:9, Dual Pictures, etc, etc and all were fine. I quite liked the feature where you could have one half the picture of a TV channel and the other half on Teletext (Ch 7 Supertext).

Mr Wilson was as happy as a barrister can be with the set - but he didn't like the bill!

So what happened? I speculate that rain water had got into the G2 screen control RV701 and caused a high resistance contact between the wiper and the carbon track, enough to drop the voltage outside the beam current circuit's capture range. Unfortunately, I didn't have the luxury of being able to dissect the sealed control further to find out.

Home handyman

A young unmarried couple brought in a Panasonic NV-SJ400 VCR for repair. When asked what the fault was, they kind of shuffled nervously and looked at each other before coming up with "no picture".

I didn't pay too much attention to this display of body language, until she qualified the statement by confessing that "initially there was no picture but now there was a something loose inside and something might be a bit bent . . ."

Eventually, after a bit of gentle persuasion, the full story slowly came out.

They were watching an old tape and the picture went snowy. They figured out, probably correctly, that the tape had dirtied the heads and so had used a wet-type head cleaner. Apparently, this didn't work, so they tried it again. When this failed, the boyfriend decided to open it up and "have a look".

All that took place only about an hour or so before they brought it to me! This was going to be good so I decided to have a good look inside while they were both still there.

Inside, I found that the automatic head cleaner was at a crazy angle, with part of its lever jammed hard under the master cam. Worse still, the take-up arm was missing altogether - that is, until the boyfriend reached into his pocket and sheepishly handed it to me - "is this important?".

This was already looking close to terminal for such a basic VCR and the end finally arrived when I noticed that one of the video heads was hanging loose from the upper head drum.

In the end, I told the couple that they had to go and buy new one. And so they left, wildly making all sorts of allegations at each other. My guess is that they are no longer a couple and that they will remain unmarried for some time yet!

Philips TV set

The Philips G110S chassis had a long and successful career in Australia in the early 1990s and was subsequently superseded by the G111S and G112. However, I don't see too many of the G112S, as it is the backbone of the more upmarket models, with advanced options and accessories. However, from my perspective, the three chassis are close enough in similarity to be able to use the same methods when it comes to repairing them.

I had a Philips 29SP1698/75R dropped in the other day with the fault marked as "No Sound". In fact, when I got it onto the workshop bench, the fault was more accurately "dead". The front LED was flashing rapidly in yellow (red and green).

A quick look at the chassis revealed that someone had put a lot of work into changing all sorts of parts (you could tell by the soldering). This meant that I had to be on my toes and look out for unusual problems.

When I said that the set was dead, this wasn't quite strictly true. Initially, you could hear the EHT static build up and then the set would die. It was as though a protection circuit had come on.

I checked the main +140V supply rails and the 5V out of the SOPS (Self Oscillating Power Supply) and they were OK. However, based on my experience with the G110S, I mostly suspected microprocessor IC7200 and the EEPROM (IC7278). As a result, I decided to replace them both with parts scrounged from a few scrapped chassis in my workshop.

The EEPROM was easy and I mounted it in an 8 pin socket for convenience. By contrast, the 42-pin high-density microprocessor was more recalcitrant but neither made any difference. The only real clue I had was the initial complaint of "No Sound".

I followed up on this lead and found that IC7266 (TDA8425) was short circuit on pin 4 - the +12V input. I also I found that R3277 - a 68Ω feed resistor - was open circuit, even though it looked brand new. Replacing both these parts brought the set back to life but only in a very limited way.

I could switch the set on with the remote control and the Green LED would stay on constantly. Additionally, there was a raster and there was hiss in the loudspeakers at high volume but apart from that, there was no sound or picture and no on-screen display.

Well, I searched high and low trying to find out what was going on. Eventually, some oscilloscope checks showed that signal was getting as far as the analog switching ICs on the AV module but no further. I then replaced the original EEPROM without result but was reluctant to replace the microprocessor because of the delicate work involved.

I still really couldn't comprehend why the set closed down when the sound output IC died. There are no protection circuit sensors in that part of the set - I can only conclude that the SDA and SCL data lines from pins 11 and 12 were feeding the information back to the microprocessor which closed it down.

By now, I was contemplating changing all the ICs on the AV module when I noticed that the microprocessor actually had G11oS stamped on it. It was then that the penny dropped - I had used a microprocessor from a G110S chassis rather than the real McKoy and they were not interchangeable.

Replacing the original microprocessor finally fixed the problem completely.

Now here is a contribution from one of our readers. It comes from A. P. of Kuranda, Qld. This is how he tells it . . .

DC-AC inverter

This story concerns a Selectronics SPI 1200-SS 1200W inverter, circa 1992, owned by John and Maria who live in Cooktown in far north Queensland. Their main source of power is a micro-hydro system which charges a bank of deep-cycle batteries and they used the inverter mainly to run their washing machine.

This particular model of inverter produces a PWM square wave, so it isn't really suitable for inductive loads like washing machines. Because of this, I wasn't terribly surprised when John told me it had "blown up" the washing machine about two years ago. On that occasion, they had sent the inverter to Sydney to be modified so that it was more suited to the task, although just what had been done wasn't made clear. Since then, it had operated the washing machine without any problems until recently, when it blew the main fuse (about 100A) on the battery bank.

The first problem I encountered when I got the beast into the workshop was how to try it out safely. If it could blow 100A fuses (and the user manual claimed it would limit its current drain to 450A!), I didn't want to connect it straight across the 12V truck battery I planned to use for testing.

I ended up jury-rigging a single strand of 16A household fuse wire in the battery positive connection. I then gingerly connected the negative lead to the battery - there was a brief chirp from the undervoltage alarm, a satisfying click as the reverse polarity protection relays changed over, and the fuse stood its ground.

So far, so good. However, when I set the inverter to the demand-start override mode, there was a buzzing noise. It sounded like the circuitry was under strain and the fuse immediately vaporised.

Figuring that this thing might need a bit more than 16A just to start up, I tried it again with two strands of fuse wire. Again the fuse was vaporised at switch-on.

By now, I had proved to my satisfaction that there was a fault and that it seemed to be persistent. It was time to get serious, so I took the lid off.

Inside the case was a whopping transformer, over which was mounted a single large double-sided PC board (component side down). There was also a large heatsink on the back of the case for the main MOSFETs, plus a smaller heatsink mounted on the side of the transformer for the secondary-side "power-recovery" MOSFETs.

I pulled the PC board out and began looking for obvious problems. However, there was no obvious sign of why the thing was eating 100A fuses for breakfast.

The next things to test were the MOSFETs. This inverter uses a total of 26 BUK456-50A MOSFETs to drive the primary of the transformer. As well, there are six further MOSFETs, type BUK437-500B, in a "power-recovery" circuit on the secondary side of the transformer.

I wasn't surprised when all the primary-side MOSFETs checked out OK - if any had gone short circuit, the inverter would be drawing significant current even before the demand-start circuit was activated. Conversely, MOSFETs going open-circuit wouldn't blow the fuse, although this doesn't mean that none might not have done so.

A more likely cause of the problem, especially considering the inductive load provided by a washing machine, would be that one or more of the secondary side MOSFETs had gone short-circuit. These are all mounted, along with their 10Ω gate resistors, on a tiny strip of PC board nestled in a heatsink mounted on the side of the transformer. Some phenolic insulating material was riveted to the base of the heatsink, hiding the MOSFETs, so I drilled out the rivets and began testing.

Minor discovery

And here I made a minor discovery. All the MOSFETs and gate resistors were fine but one of the gate resistors had not been properly inserted and one of its leads was only just touching the solder that covered the hole it was meant to go through.

This was easily fixed but I could not imagine that this was the cause of the fuse-blowing. Indeed, a quick test showed that there was no change in the inverter's behaviour.

At this stage I suspected that the present problem lay in the circuitry controlling the MOSFETs. I considered verifying this by testing the outputs of the MOSFETs without the transformer connected but since there would now be no feedback from the secondary of the transformer, I wasn't sure that this would prove anything. In addition, even if I found a problem here, I'd still have to find the fault in the control circuitry.

The main PC board in this inverter is mostly populated by inexpensive, readily available parts, so in the absence of a circuit diagram which might make a deductive approach possible, I decided on a shotgun approach: check or replace everything on the board.

I started with the resistors. This approach bore fruit very quickly when I found that R50, a 68Ω 1W resistor, was open circuit. There was no obvious heat damage to the resistor and I expected that this was the fault. However, when I replaced it and re-tested the inverter, its behaviour was the same as before.

I then checked every transistor and diode, the leakage and ESR of every electrolytic capacitor and the breakdown voltage of every zener diode. This revealed nothing so I checked the trimpots. The current limit trimpot, P7, is marked 200Ω but measured nearly 300Ω. I replaced it but didn't imagine that this would fix the problem. I was right.

Next, I started on the ICs. I was able to verify that the timebase, an M706B1 with a 6.5536MHz crystal, was producing a 100Hz square wave. I also verified the operation of two 4N25 optocouplers which provide isolated feedback from the 240VAC output to the control circuitry.

The other ICs couldn't be tested so easily without a circuit diagram, so I replaced them. There were three CD4093s, an LT3524 PWM controller, an LM335 temperature reference and an NE555 timer. Because the fault was not intermittent, and because these are all inexpensive parts, I replaced them all at once without reassembling the thing each time to test it.

If that had fixed it, we still wouldn't know exactly where the fault was. Fortunately for this story, the fault was still there.

By now, there really wasn't anything left on the PC board to check. So could it be a component that wasn't mounted on the board? Surely that massive transformer couldn't have a shorted turn, could it?

It was easy to test: I connected the output of a 9VAC 1A plugpack to the primary of the transformer and measured the voltage on the secondary. It was only 1.25V AC and the plugpack was pumping out 1.5A. Voila! I phoned Selectronics and they verified that I ought to be getting a much higher output voltage, with an input current of only 150mA at no load.

In retrospect, I should have tested the transformer a lot earlier but I had considered it so unlikely to be faulty that I hadn't bothered.

At that point, I declared the inverter a write-off. The price of new inverters has plummeted recently and a new transformer would cost almost $400. And after all that, it would still be just a PWM inverter. I have therefore advised John and Maria to save their pennies for a sinewave inverter to run their washing machine.

This all happened about two years ago and I have since encountered quite a few faulty high-power transformers, mostly in amplifiers. The humid environment here in the tropics obviously doesn't do them any good.

Share this Article: 

Digg this story

Seed Newsvine

Post to Reddit

Save on del.icio.us

Email