But first, to that heading; when is a fault not a fault? That comment is not as facetious as it may appear. This story is a secondhand one and I was involved as little more than a spectator. Nevertheless, it is a story well worth telling if only because the solution was quite unexpected.

A friend of mine who is interested in computers decided to purchase two secondhand Pentium 133 computers on which to network his family business. He was assured that they were in good working order but when he set them up at home, neither would boot up – there was just a blank screen. However, if he switched them off and then on again immediately they worked fine.

This made him think that there was either something wrong with the setup or with hardware compatibility, so he started experimenting. First, he tried a succession of different monitors and found that some worked perfectly and others didn’t – even though none of the "faulty" monitors gave trouble on any other computer. So was there some incompatibility between the non-working SVGA monitors and the motherboards fitted in the two computers?

If so, the problem could be either in the energy saver function or multisync resolution but he could find no correlation between these features or why some monitors worked and some didn’t. In the meantime he asked me and several other people he knew if we had a clue as to why this was happening. But although we suggested various things to try in the BIOS, plus checking the links on the motherboard, he wasn’t getting anywhere.

Then, one day, he connected one of the computers to one of the monitors that originally wouldn’t work and suddenly it all worked perfectly. He repeated this test several times and it booted up every time.

It was then he realised he had accidentally forgotten to plug in the keyboard. No big deal – he plugged it in and the fault was back! Could it be a faulty keyboard? He acquired another keyboard and plugged it in and this time all was well. He checked the first keyboard on other computers and could find nothing wrong with it. So what was the difference between the keyboard that produced the fault and the one that didn’t?

This speaker terminal panel came with an unusual manufacturing fault, as indicated by the yellow circle.

Well, the former was a 101-key keyboard, while the latter was a 105-key Windows 95 type. As he had plenty of others around he tried a variety of keyboards and established that the problem was entirely due to the computers being incompatible with the older 101 keyboards (pre-Windows 95). Naturally, he was relieved to have discovered a cure although, strictly speaking, it wasn’t really a fault at all.

(Editor’s note: this problem is generally due to an incompatibility between the keyboard controller chip on the motherboard and the microcontroller in the keyboard itself).

Home theatre system

Young David, a 17-year old, was very proud of his Sherwood Home Theatre RV-4070R amplifier but he had become very concerned that it was intermittently cutting off. He could not think why it was happening as he was sure he was taking good care of it.

Eventually, he go fed up with it and decided to take his pride and joy to the "doctor’s" to get it fixed. To begin with, I had great difficulty in getting it to produce the fault and was about to dismiss it as being something external to the amplifier, when it finally failed. Eventually, after a lot of tapping, heating, cooling and various other tests, which young David really wouldn’t want to know about, the problem was narrowed down to the speaker connections.

It applied to the right channel only, where wiggling the speaker leads even slightly was enough to cause the amplifier to close down. Curious, I placed an ohmmeter across the terminals with the set switched off and inserted a speaker lead in the negative terminal. As I did so, the multimeter showed a dead short.

As shown in the accompanying photo, the spring-loaded speaker terminals were part of a panel, with the positive (red) terminals at the top and negative (black) terminals along the bottom. Initially, it was hard to see how this setup could possibly produce an intermittent short circuit, as the two terminals are well spaced.

It all became clear when I removed the terminal panel and took a close look at the back. As shown in the photo, the tinplate connectors from the terminals are all brought out along a common edge. The connectors from the red (top) terminals are towards the rear, while the connectors from the black (bottom) terminals sit closer to the panel.

Unfortunately, due to a manufacturing defect, the connector for the right channel negative terminal had not been pushed all the way down into the plastic moulding. Instead, it was loose and could easily come into contact with the positive terminal connector behind it.

And, as I quickly discovered, inserting the speaker lead only made things worse. When this was done, the spring-loaded plastic tab pushed against the lead which in turn pushed the metal connector backwards so that it came into contact with the positive connector.

Fig.1: part of the switchmode power supply in the NEC N3419. The 5-pin IC, Q801, is at lower right. It's heatsink had been loaded with five extra heat­sinks in an effort to control overheating.

Fortunately, the cure was a simple one – I pushed the connector all the way into the plastic moulding using a flat-bladed screwdriver and used a dab of super glue to ensure it wouldn’t come loose again. It was a rather strange fault but at least the repair was easy and it didn’t take too long to track down.

A set from the country

There’s often quite a bit of difference between service work in the city as compared to the country and that was brought home to me recently by this story.

City technicians generally have much better access to technical information and spare parts than their country cousins. In the country, it really is a case of sometimes making do and inventing solutions from limited resources. Obviously, if a country serviceman only has a 100μF capacitor and the circuit calls for a 47μF, the 100μF capacitor will have to do.

The Hayes lived in the country on a small farm and they had had a few problems with their 34cm NEC N3419 portable. This little set uses a Daewoo C43 chassis, is made in Korea, and is very popular with many brands. But it is getting a little ancient now.

The set had been cutting out intermittently for quite some time and though the local technician had tried his best on several occasions, it was finally brought in on a trip to the city.

When I removed the back cover, I was immediately aware of an array of four extra heatsinks which had been screwed onto the manufacturer’s original heatsink for Q801 (STR5412), 5-pin IC in the switchmode power supply. I also noticed several possible dry joints and there was brown goo everywhere, especially around IC 1502.

Despite all the extra heatsinks, they still became extremely hot when the set was running which explained why the set was intermittently cutting out. The reason wasn’t hard to find – the high tension was high at 120V instead of 103V, as shown on the circuit.

And that brings me to the real point of this story. The manufacturer has issued some modifications for this circuit, a fact that would be unknown to many technicians in remote areas. These modifications involve two capacitors, C811 and C808, both originally specified as 4700pF.

Anyway, I replaced the IC (Q801), removed the additional heatsinks, and changed C811 and C808 to 2200pF and .001μF respectively, as recommended by the manufacturer. I also attended to the suspect joints and cleaned up the brown goo.

Anyway, the set was now delivering the correct 103V HT and was running cool. The height had to be readjusted but that was that.

I don’t know how much extra life the additional heatsinks gave to this set but at least the bloke was trying with whatever came to hand. And he wouldn’t have known about the manufacturer’s modifications.

Finally, I would remind all those who work on this chassis to always change C434, a 10μF 160VW electrolytic capacitor on pin 4 of the horizontal output transformer. This will help avoid expensive pyrotechnics.

Akai TV receiver

Mr Keenan brought in his 2-year old Akai CT-21WA9AT 53cm TV set, complaining that it would switch off by itself after a while and that there was a white line across the screen. Considering the set was so new, I asked him if it might still be under warranty but it wasn’t.

With the back off, it didn’t take long to determine that the fault was in the vertical output stage, IC401. This stage takes its supply (Vcc3, pin 11) from pin 4 of the horizontal output transformer (T402) and was loading the horizontal output stage. The vertical stage was drawing too much current and this, in turn, caused Q403 to eventually turn the set off.

Replacing the IC fixed the problem but I was a bit nervous as to what had caused it to fail in the first place. I started by replacing the two electrolytics (C910, C912) in the power supply. I also checked the HT which was correct at 110V and then had a chat with a friendly Akai service agent that I know.

He got onto the Akai Service Guide CTV-042 Code 204 on the Internet. This suggested that three 0.1μF green mylar capacitors – C911, C424 and C351 – should be changed in the CT-2119AT series, to which this set belongs. He also showed me a few other modifications which, at the time, were irrelevant to the symptoms at hand.

I changed the three green capacitors but I must say I was rather surprised at the advice to do so as I haven’t previously had any problems with this type of capacitor. To be on the safe side, I also replaced a few suspect-looking electrolytic capacitors around the vertical IC. By now, I was feeling pretty confident, so I boxed it up and put it aside for soak testing.

Some time later, I decided to check on the precarious state of some of my shares. "Why not use Teletext?", I thought. The Akai was the nearest set to hand but when I hit the text button on Channel 7, it didn’t immediately show the Index on page 100. Instead, the display was a mixture of text from a variety of pages and the clock jumped from time to time instead of showing every second.

Now Teletext can be a rather temperamental feature and is highly dependent on a good quality signal (preferably with no ghosting and not too strong). With this in mind, I checked the reception on Channel 7 and it looked great. Just to make sure, I tuned in a UHF translator station and checked the text there too but again the Teletext was quite poor. I then tried another TV set, which performed perfectly.

Obviously there was a fault in the Teletext section of the Akai – but where to start? I thought I would begin with the easy things, like the RF AGC. To do this, it is necessary to put the set into the Test mode or Adjust Menu, by switching the set off and on again while holding the volume + and - buttons at the same time.

Pressing 2 on the remote keyboard brings on the RF AGC adjust menu which I then adjusted with + and - on the remote control, until the snow on the screen just disappeared. I then turned the set off and on again. This made no difference to the Teletext reception, which was slowly getting worse, with the clock not appearing for a very long time.

Mr Keenan then phoned to ask about the set’s progress and I told him that although I had fixed the two faults he had complained about, there was still a problem with the Teletext. My immediate impression was that Mr Keenan had known about this problem all along because he wasn’t the slightest bit surprised. Now Teletext hasn’t been a huge success in this country and most people can take it or leave it but not Mr Keenan – he definitely wanted it fixed.

And so I delved back into the guts of the set. The Teletext circuit consists of only two ICs (IC801 and IC802) and two transistors (Q801 and Q802). Surely it shouldn’t be that hard to fix! I began by checking the 5V rail to this circuit and it was spot on. I then checked that crystal X801 was oscillating at 13.875MHz, which it was. The CRO also told me that I had video all the way to pin 3 of IC801.

Unfortunately, there is no further technical information on the Teletext circuit (no block diagram or voltages), except for the adjustment of variable inductor T801 – this should give 2.5V on pin 28 of IC801. And that was my first clue – this voltage was low at only 0.5V and adjusting T801 didn’t make much difference, although the display became worse after losing its horizontal hold.

I also noticed other problems with text – in the mix mode, the text characters would lose their horizontal sync and tear. By delicately setting T801 I could almost lock it, implying a loss of some sort of sync.

I checked the two transistors and all the diodes and they were all OK. Frankly, I was running out of ideas. I went back to my Akai mate and he found that there was a modification involving an extra 1μF capacitor between pin 21 of IC801 (the 5V supply) and chassis. Encouraged, I hastened back to fit it only to find the set had already been modified on the PC board side.

There was nothing for it but to order and replace the components I couldn’t really check, namely the two ICs, the crystal, inductor T801 and varicap diode CD801. I decided to change the crystal because, although it may be oscillating, there could be something wrong with the amplitude.

While I was waiting for the new parts, I experimented with heating and freezing all the relevant circuits. This made no difference except that the problem was gradually getting worse. Eventually over $100 worth of parts arrived and I fitted them one at a time to try to pinpoint the culprit. Unfortunately, they made no difference.

Once again I pestered my Akai mate and this time he found a page of AASC Service Hints for this series with similar (but not exactly the same) symptoms. This suggested EEPROM IC602 – ST24CO4(SGS) – and the reader can imagine my frustration when it arrived and still didn’t fix the fault.

There are only about 50 components in this circuit and I had already replaced five of the main items – only 45 to go.

I started with the capacitors, especially the high capacity (104) brown ceramic types that often give trouble, and worked down. I also replaced the two electrolytics (C803 and C811) but nothing made any difference.

By now, this repair was no longer economic but I hated being beaten, especially after all my work. I now decided to measure all the resistors, starting with the highest values first and working my way down.

Everything was fine until I came to R806 33kΩ. It measured high at nearly 1MΩ. By the time I had removed it, it was even higher.

This, I felt sure, was it. I replaced the resistor, fitted everything back into position and switched it on. The first thing I had to do now was to retune T801 for 2.5V on pin 28. This time, the voltage was much healthier and it didn’t take much adjustment to reach the correct value. What’s more, the screen was displaying the full index and the clock was updating every second. Even in mix mode, the text was perfect and locked solid.

It is extremely unusual for a 33kΩ 0.5W resistor to go high in this manner. The problem now is what to charge for finding and fixing it. Whatever the figure is, it won’t be big enough!

Different standards

It is always interesting to see the varying approaches used by different countries to achieve the same result. With today’s multinational companies, it is not unusual to see European chipsets fitted in Asian TV sets, though often without large chunks of the technology within those chips being used.

Almost all Asian factories (Japan, Korea, Taiwan, Hong Kong and China) have the aquadag of the CRT connected to chassis, whereas Philips and other European sets use an above chassis design connecting to a beam limiting circuit. Aquadag, by the way, is the water-based metallic black paint on the exterior of the picture tube – it acts as a large capacitor plate, in conjunction with the internal anode, to filter the EHT.

One of the main problems when servicing TV sets is finding a convenient and reliable chassis reference point for the meter. However, one can depend on Asian manufacturers with their chassis connected tubes.

Recently, I had a Philips 2SSP1788/75R Symphobass come in with low contrast and brightness. The first problem was to work out what chassis was used and whether it was designed for Australia standards. It was built around 1991 and for those familiar with the Philips nomenclature, these points can be worked out from the model number.

I had to look it up and the Australian version is a G112S (a variant of the G110 series). Because Philips have used the above-chassis system in their colour TV sets since the K9 chassis in 1974, I knew from experience to expect approximately +15V on the tube aquadag with respect to chassis, depending on the beam current drawn at the time.

Based on this background, I went straight to this point and was not surprised to find it measured 0V. Following the path back from the aquadag to the beam limiting circuit was not so easy. The lead goes onto the CRT socket board (circuit C) and out via plug and socket 4P2/1S7 to D63. This then goes to plug and socket 1M7 (circuit D) and to the chassis end of the EHT tripler (pin 7) inside the horizontal output transformer (5901).

From there, the path follows the "AQUA" D103 line and this goes to TP43.

A number of circuits are involved with the beam limiting. First, there is TR7911, a protection transistor, and then the east/west circuit and vertical outputs, which are connected to keep a constant picture size during changes in beam current.

I was looking for a source for the 15V DC but with so many affiliated circuits, I wasn’t sure which was the significant one. A beam limiting line ran via plug and socket C74 back to the C circuit so I decided to start looking here first.

To cut a long story sort, I finally found that resistor R3970 (22kΩ) between the +34V rail and the beam limit line (C74) was open circuit. Replacing it restored the brightness and contrast – and the average 15V on the aquadag.

One can be lucky sometimes.