"Where do I start?" - that's often a question I ask myself when faced with an unusual fault or an unfamiliar model, especially when there are few obvious clues.

Normally, I try to find the likely area and then proceed to concentrate on that. For example, if there is no line drive, I start by measuring the voltage/waveform on the collector of the horizontal driver transistor and go from there, each time halving the number of components to be tested.

Recently though, I've had a few cases which made this procedure futile because the faults could have been anywhere. I was therefore reduced to probing various circuits in the hunt for clues.

Items Covered This Month Samsung CB-564BV, S51A TV set Philips 25CE6270/10B TV set - CP110 chassis. Sony KV-20PS1 TV set Philips 41 GR8840/75B projection TV set

Samsung CB-564BV TV set

The first set I was faced with was a 1999 Samsung CB-564BV, using an S51A chassis. This set is somewhat unusual as it employs a widescreen 53cm picture tube. However, it is not a real 16:9 widescreen set - rather it has an 11:8 ratio although the on-screen menu does offer 16:9 (but it is overscanned). Apart from that, the set had no special features.

The set had died after a thunderstorm and the symptoms were that it would pulsate every three seconds, with no sound or picture. If allowed to remain in this mode for a while, you could actually see it trying to give a blurred raster momentarily before switching off. It was as though someone was switching it on and then immediately off with a remote control.

Unfortunately, I had to order a service manual in especially for this set and even then I couldn't get the right model. What's more, it didn't arrive until two weeks after I had started the repair.

I began my investigations by measuring the output from the chopper supply to find that all was OK. The voltages were much as I had expected and were similar to other Samsung models I had seen. In particular, the outputs I measured were at 130V, 16V and 12V and these rails were all steady, varying only with the load as the set switched on and off. In addition, there were healthy line pulses coming from the horizontal output transistor (Q403) and there was EHT.

These sets employ a complicated arrangement of switchable IC regulators. These are IC803 (KA78R05, 4-pin) and IC804 (KA7630, 10 pin) and they deliver +5V and +8V rails respectively. These regulators are also "tied up" with the power-on, reset, disable and protection circuits.

Having decided that the primary power supplies were OK, I decided to investigate how the set is switched on and off normally. After all, it appeared that the low voltage supplies were being turned off, possibly because of a fault condition elsewhere in the set.

As it turned out, the low-voltage supplies are controlled by pin 18 of microprocessor IC901 (Zilog 4202 SZM-503ATS). When this pin goes high, the low-voltage IC regulators turn on and supply 8V to pins 12 and 37 of the jungle IC (IC201, TDA8843). And that was exactly what was happening except that something was also switching it off almost immediately. LED LD901 (a combined LED) was also flashing yellow every three seconds too.

My next step was to disconnect the remote receiver from pin 19 of IC901 in case that was randomly switching the set on and off. It wasn't, so I checked the EEPROM (IC902, K24CO41) by substitution but it too made no difference. I then found that the SCL and SDA rails had digital noise on them and were at +5V DC.

Next, I desoldered pin 18 of IC901 and applied 5V to the track leading to it to see if the set would switch on and stay on. My reasoning was that this would at least tell me whether it was the microprocessor control circuit that was the cause of the problem.

The result surprised me - the set came on and then went off and stayed off until I switched it off and on with the main switch.

And so, despite all this measuring and testing, I still really wasn't sure where the fault might be. The service manual arrived and wasn't much help either. It showed all sorts of circuits I had hoped would help me, such as x-ray protection, but in the end I discovered these options hadn't even been fitted.

By now, it was beginning to look as though I would have to order and replace several large and expensive ICs to confirm whether or not they were working.

Finally, I solved the problem in a rather roundabout way. As I said earlier, if you left the set on, you could get a pulsating raster totally out of focus. However, the FBT focus control made no difference to it at all.

After examining this series of events more closely, I came to the conclusion that the picture was projected up and reflected onto the screen - ie, that the foggy picture I was observing was due to reflections within the picture tube. And that indicated that there may be a fault in the vertical timebase that was triggering an internal data fault signal.

I now concentrated on IC301 (LA7845) and its seven pins. There was +16V on pin 6 and -16V on pin 1 and there was a nice sawtooth appearing in pin 4 (VDM). However, the waveform on pin 2, which goes to the deflection yoke, was not correct. I was expecting its DC voltage to be around 0V but instead I was getting nearly 16V, which accounted for the raster not being on the screen.

I also noticed that there was no sandcastle waveform from Q302's collector and I was getting even closer when I discovered that there was no waveform (VDP) to pin 5 of IC301. Because it was easier, I spent some time measuring and replacing components around IC301 (especially C302) before moving the scope probe to pin 46 of IC2301. There was no signal coming out of this pin, so I desoldered it and checked it with an ohmmeter - it measured a dead short to ground.

At last I had some sort of a clue that a component was at least faulty but was it enough to switch the set off? I tried switching the set on again with the pin disconnected but it still pulsated.

However, as there was definitely a fault on pin 46, IC201 (TDA8843) needed to be replaced. The problem was the cost - $67.65 trade plus an ETA of 6-10 weeks delay! In the end I was able to scrounge a secondhand TDA8844 from a scrapped Philips TV and fitted that.

And that, as they say, was that. After refocussing, the raster was restored permanently in the centre of the tube. The lack of sound was due to it either coming on in AV mode or muted because there was no signal. I used a similar remote from another set which was enough to get into the OSD menus and go through the setup options before returning the set to the customer.

Philips TV set

The second story concerns a 1989 Philips 25CE6270/10B with a CP110 PZ1 chassis. Despite its age, I hadn't worked on this model before - probably because there aren't too many of them and they are all probably fully imported from Belgium.

This set belonged to an elderly lady pensioner who wasn't too wealthy. She complained that the set had previously been taking some time to come on and was now dead.

After examining it, I could see it wasn't due to dry joints and that fixing it was going to be expensive and probably not worth it. However, she definitely could not afford a new set so I decided to give it a go.

Back at the workshop, I replaced the failed parts, namely fuse F1652, bridge rectifiers D6657 and D6658, chopper transistor 7665 (BUT11AF), IC7669 (TEA1039) and resistors R3658 (120Ω) and R3659 (100Ω). This restored the sound and picture - but only just. The fault now was that the set was pulsating rapidly and the main 140V rail was fluctuating wildly.

Once again, I wasn't really sure quite where to start. Was the fault in the primary or secondary of the power supply, or was it in the horizontal output stages or even the east-west circuit? Another possibility was that it was in the microprocessor circuitry as the set still didn't always want to start, leaving error messages in the display.

I also noticed that the spark gaps on the CRT board would occasionally flash over at switch on. That left open the possibility that it could even be the tube that was faulty.

In the end, I decided that it was most likely to still be in the power supply. In particular, I suspected that it wasn't regulating properly.

I ordered a service manual and meanwhile worked on what I could. I started with the electrolytic capacitors on the secondary supply rails (+25V, +140V, +32V, +15V, +12V, +9V and +6V), replacing any that were leaky (C2670 and C2621 were particularly bad). I then substituted a 60W globe for the line output load by disconnecting plug R13 and connecting the globe between pin 5 of this plug and ground.

However, the 140V rail was still unstable and varied a lot with a 100W globe as the load. It could, however, be adjusted using VR3670, but it would not remain steady.

When the service manual arrived, I found some notes on some modifications. These involved removing C2657 from the base of the chopper transistor to ground (live side), fitting a 39Ω resistor across L5656 and changing C2661 from 1500μF to 2200μF. These modifications slightly improved the stability of the 140V rail and encouraged me to continue working in that area.

Next, I replaced the 7670 optocoupler (it is marked as a CNX62 on the circuit but a CNX82A was fitted), followed by 6.2V zener diode ZD66676 which is on the feedback reference line. However, these changes made little difference.

What's more, I was being continually frustrated with differences between the circuit and the set itself. For example, mine had an extra module with an SCR (TR7000) fitted on the 15V rail.

At this stage, I noticed that the set performed differently when it was hot compared to when it was cold. In particular, when TR7666 (BD337) in the primary of the Self-Oscillating Power Supply was hit with freezer, there was a big change in the set's performance. I removed this transistor (which in reality was a BC337-25, the "25" signifying a higher hfe) and measured it very carefully. I could find absolutely nothing wrong with it - there was no detectable collector-emitter reverse leakage and its hfe was 175, but I couldn't see that as being significant.

A new one measured 190 but I fitted it anyway, not expecting much. Fortunately, for some unknown reason, I was wrong; at last the set was stable and I could adjust the 140V rail exactly. It looked as though all my problems had come to an end, so I put the set aside to soak test.

But that wasn't the end of it - most of the time, the picture was fine but just occasionally, the set would bloom a bit when there were bright objects in the picture. What's more, when the set was switched off, it was not always starting up again - particularly in the morning. This got particularly bad when we had a bit of damp weather.

To troubleshoot this problem, I connected a meter to the 140V rail so that it could be monitored while the set was soak testing.

These symptoms continued randomly and sometimes it was very hard to even start the set at all, although the +140V rail was continuously spot on. Occasionally, however, the display gave an error number such as F3 (IC7840 microprocessor) but this varied.

What on earth was I missing? The spark gaps on the CRT board now arced nearly every time I switched the set on and yet the EHT remained constant.

Next, I measured the back-up battery and found that it was completely dead. This component failure has been enough to cause many strange faults in Philips TVs but no such luck in this case. By now, I was pretty well satisfied that the power supply was functioning correctly, so I decided to spend some time investigating the line output stages.

One possibility was that the insulation was breaking down and there was a momentary excess of EHT, causing the spark gaps to flash and the microprocessor to detect a failure. I followed up a lot of stray dead end leads, such as tuning capacitors and even the picture tube before I decided to check out the CRT earthing of the aquadag. It didn't seem possible there could be anything wrong with this circuit, as I could see clearly no less than two leads going from the aquadag earth strap to the CRT socket. However, my ohmmeter could not read a path to the chassis ground.

I didn't panic here either because Philips is one of the few manufacturers that used to keep the CRT aquadag at about 15V and use it as part of the flyback beam limiting circuit, as in the CP90 circuit diagram (remember the K9, K11 series?). My next step then was to examine the circuit to find out what was happening.

Unfortunately, the circuit diagram was a bit ambiguous when it came to this vital bit of knowledge. The CRT circuit clearly shows the aquadag to be grounded but does not show where. So where was the strap to the chassis ground? There wasn't one, so I connected a crocodile clip from the CRT earthing strap to the chassis metal work. Bingo! - all the fault symptoms immediately disappeared.

What had happened? I can't be 100% sure but it turned out that someone else had looked at the set previously and hadn't been able to fix it. So I can only surmise that they took the lead off and either lost it or forgot to replace it.

I couldn't find it in my heart to charge the lady for all the time it took to fix these faults.

Sony Profeel TV system

The Sony Profeel TV system was a big step into the semi-professional market, offering for the first time (in 1982) a component video system. It is all of course ancient history now, and most of it is now landfill.

For reasons that I refuse to go into, I got conned into repairing a KV-20PS1 (using an HF SCC-428A-A chassis). The fault was no colour and this is a multi-system monitor with a complex chroma B board. In the days when Arthur was a boy, the Sony PAL boards were hard enough to troubleshoot but here we are talking about an automatic multi-system job!

Fortunately, the decoders are split between SECAM and PAL/NTSC I and II, with IC301 (PC1365C) being mostly PAL.

Despite the writing being very small and faint, I was blessed by having a service manual you could die for these days. The circuit detail was excellent, showing waveforms, voltages and even the block diagram inside the IC - all in one diagram.

The first thing I was looking for was the PAL colour killer so that I could override it and see what sort of colour - if any - was getting through. This turned out to be RV309, which fed pin 13 to the "ID killer" - 5.9V for colour, 7.3V for monochrome. I marked the position of the control and then twiddled it but it had no effect.

I checked the voltage on pin 13 and it never dropped below 8.7V. I then checked the earthing for the control circuit and also checked R324 (220kΩ) but all was OK. I was about to suspect the IC when I noticed another connection to pin 13. Following its tortuous path, I got to R384 (100kΩ) and plug B-2 (pin 41). Following the harness further, I reached a small slide switch (SW901) on customer control panel H and marked "SECAM" and "AUTO".

In the "AUTO" position, this lead went nowhere but in the "SECAM" position, it was connected to +12V (B-2 pin 42). There was 8.7V on one side of R384 and 12V on the other - no matter what position the switch was in! Disconnecting R384 restored the colour and the voltage on pin 13 dropped to +5.9V.

Unfortunately, the slide switch is located in an extremely inaccessible place, beneath the picture tube. And you could see from the scuff marks on the front panel that this switch had been used a lot and was pretty worn. A replacement fixed the problem, although cleaning the old one would probably have been sufficient. Alternatively, if the client hadn't been using SECAM, it could have simply been disconnected completely.

Philips projection TV

One morning, a young lady phoned and asked if I would go to her mother's house and repair her "telly" as soon as possible. Naturally, I immediately asked for the brand and model number of the TV but the somewhat agitated woman was unable to give me any further information except to tell me that it was a very large flat-screen set.

"Is it a plasma set", I asked. She didn't know, nor did she know how old it was. She just wanted me to fix it as soon as possible because her mother was phoning her about it every five minutes and giving her hell. It appeared that television totally dominated her mother's life.

I "rocked up" at her mother's extremely expensive residence about 30 minutes hour later and was immediately impressed by its opulence and breathtaking views. Yep, this would have to be a plasma TV - they could definitely afford it.

The entrance to the mansion was on the top floor, next to a massive garage, and was opened via a security system. Walking down a grand marble staircase, I finally met a large woman who turned out to be the anguished mother. She led me down to the spacious TV room, where I was extremely disappointed to find an ancient projection TV set - a 1989 Philips 41 GR8840/75B using a G110-PTV chassis.

The mother, who was Greek and had only limited English, was extremely friendly and was delighted that I was at last going to fix her faulty TV ("the TV - shea no work!"). I started by checking that it was plugged into a working power point and that the aerial connection looked intact. I then tried the remote - but nothing happened so I tried the front-panel controls and managed to switch it on but could not determine what channel it was on straight away.

Because the room had large windows with no curtains and it was a bright day, it took some time before I realised that there was a picture on the screen. However, it was extremely dim. "Mama", it turned out, only wanted to watch her two Greek channels on Foxtel but it was extremely difficult for me to establish what these were from her limited English.

After a few desperate calls to her daughter, I eventually discovered what they were (46 and 51) and tuned them in on the set-top box by hand. However, none of the remote controls were working which seemed rather strange. These included the Foxtel remote, the Philips remote (for the TV) and a Panasonic remote for the VCR - none worked, not even with brand new batteries.

After some more phone calls to the daughter, it transpired that they were all working the day before. This was just too much of a coincidence.

Finally, the penny dropped when she managed to tell me that "they mussa work, I cleaned them this morning". Cleaned them? Well, actually it turned out she had washed them by completely immersing them in suds and water in the sink before drying them.

OK, so it's not the end of the world, except that the Foxtel and Philips remotes are manufactured with their outer case shells glued together. And that means that they are unserviceable and have to be replaced.

I explained this as best I could and I think she understood. I also said it was time to buy a new "telivis" since the old one delivered a very washed out picture. However, she was so ecstatic that her favourite Greek channels had been restored that I don't think she really took this on-board.

She proffered a new $100 bill to pay for the service call but I didn't have change and told her I would take a cheque. She understood this part exactly and next we were climbing the marble staircase to her garage. Inside the garage, she opened the boot of her immaculate Roll Royce, retrieved her cheque book and proceeded to write me one out!

It was all slightly bizarre and somewhat amusing. The faults with the projection TV were not insurmountable but she really could probably afford several plasma TVs.

The remote (RC5903/21) is cheap enough to order in and the dull picture is probably due to the heat transference liquid in the projection lenses. However, if you follow the full procedure in the service manual and order the coolant kit (4822 310 57233), you won't get much change out of $200 and it takes all day make the change-over.

The TV coolant fluid is optical grade (99.5% pure) 70% mono denatured ethylene glycol and 30% glycerol/glycerine and can be obtained from the Internet or by email from sales@matelectronics.com for only US$6.95 (Part No. RCA 212072-16) for a 500ml (approx.) bottle! However the freight is expensive (US$42.20).

I bought four bottles, though three is more than enough and you might get away with just two, which will cost you $A128 landed in Sydney. If you are careful - and care is the operative word throughout all this - you really only need one bottle per CRT. This stuff is not only poisonous and toxic but is also corrosive. If you drop some onto the PC board below you will probably ruin the set forever, no matter how hard you try to clean up the mess afterwards!

So changing it is a bit of a challenge. If you follow the service manual, you will remove each CRT, disassemble it, change the sealing coupler and O-ring seal, etc, and then reassemble it. You will then spend hours realigning them.

Alternatively, you can take obsessive care, do it in-situ, and raise the corners of the cabinet until the face of the CRT you are working on is absolutely level. You then remove four black 1/4-inch hex screws and take off the lens assembly before unscrewing four 5/16-inch nuts and removing the C-lens (round concave cup lens) to expose the fluid.

This fluid has to be removed completely - I use a large pipette to suck it out and finally use a cloth to get the last bit out. You will then find that the front of the CRT is discoloured and will have to clean and polish the tube front with "Windex". The amount of "dirt" will vary from colour to colour. The algae grow best (or worst) in the blue and green tubes, while the red tube is often not too bad.

When you are satisfied the cavity is pristine clean, you can pour the new coolant in, taking care to ensure there are no bubbles. After that, you have to reassemble the cleaned lenses in the reverse order. And that's it - the difference in the picture will be amazing.

Note, however, that if there is insufficient coolant, you will burn the screen and replacement tubes are extremely expensive. Fortunately, this procedure is probably only needed once during the set's lifetime.

The other common failure in this series of TV is the EHT splitter (4822 218 20809) besides, of course, all the usual G110 chassis faults.