Items Covered This Month Kikusui COS5060TM 60MHz delay oscilloscope Technics RS-1500US professional reel to reel tape recorder Sharp CX68GS TV set Philips CH685 TV set (KL9A-3 chassis) Superloos – a close encounter of the toilet kind Grundig Megatron M82-115/9 IDTV/P+P TV set (CUC1882 chassis) Philips 32FL2881/75R TV set (FL2-G chassis) Panasonic TX-51P800H (GP1 VP chassis)

I think I’m over servicing oscilloscopes. The only thing they seem to have in common with TVs is the CRT and even then the voltages are all completely different and difficult to measure.

Basically, you need a digital voltmeter that can read from -3000V to over +12,000V. Unfortunately, using a 30kV probe with an analog meter is rather difficult – especially when it comes to making contact with a tiny, inaccessible component without shorting anything.

Anyway, I had to fix a Kikusui COS5060TM 60MHz delay oscilloscope that had excessive uncontrollable trace brightness. I don’t know its age (I guess about 1983) but the look and feel of this CRO is very similar to that of the LG/Goldstar ones I have written about recently. It even uses a very similar Toshiba 150CTB31 CRT (16W).

However, unlike the others, there are no voltages or control descriptions written on the PC boards, the access is much worse and there is no circuit diagram in the instruction manual.

Once I began looking inside, it was hard to know where to begin. In the end, I decided to start my attack with a soldering iron and resoldered as many dry joints as I could find, especially in and around the power supply and EHT sections. I found one "doozy" of a classic dry joint to the CRT filaments but nothing that made any significant difference to the fault at hand. All the other controls appeared to work and the intensity control was having a slight effect, mainly due to the size of the sweep and focus which I put down to "blooming" from the excessive beam current.

Next, I tried to figure out where the CRT bias control was but was unable to be sure. There are three largish trimpots next to the CRT connection plug which I marked before seeing what effect they had. However, they seemed only to have an impression on the trace focus.

By now, I desperately needed a service manual and so I got onto the web and started looking. Google revealed no private copies and virtually no information at all but I did discover that the Australian agent for Kikusui was Emona Enterprises. As a result, I contacted their spare parts division and enquired about purchasing a service manual.

A few days later, I was informed that the complete service manual was available for $150 plus GST (three weeks ex-Japan). Unfortunately, this price was out of my league but a very obliging young lady at Emona managed to organise a deal for me that was acceptable. She could supply a few relevant photocopied pages for $25. I jumped at this generous offer and in due course not only received the complete circuit diagram but also three BAV21 diodes that Emona’s technical section thought might be the cause of my problem (CR835, CR837 and CR838 in the G1 circuit to the CRT).

Well, that was service with a smile! I removed and checked the three diodes but unfortunately could only find CR835 to be slightly leaky. Replacing all three made absolutely no difference to the symptoms. I now knew that the CRT BIAS control was RV821 but it too made no difference to the fault.

Another disappointment was the lack of voltages marked on the circuit, although I managed to confirm the few that were marked as being correct. Using a high-voltage probe, the cathode measured -1.74kV, the first grid -1.76kV to -1.77kV (depending on the intensity control) and second grid (P1) -1.16kV (Focus). The third grid (Geom/Accel) was +45V and the fourth grid (Astig) was +63V, which was roughly what I expected.

G1 with respect to the cathode was varying between -20V and -30V but when I connected a DVM between these elements, the voltage range was different, this time varying from -8V to -20V (minimum to maximum intensity respectively). Similarly, the second grid to cathode measured 580V with the probe and only 190V with the meter. I’m not sure why these discrepancies occurred but I suspect that the DVM was dragging the voltages down.

Unfortunately, I couldn’t measure the EHT as access was too difficult, the probe being wedged hard between the CRT and the front control panel. Any attempt I made caused severe arcing and threatened to destroy the circuits!

POSITION VACANT: ELECTRONICS TECHNICIAN Our company is offering a long-term learning and rewarding experience for a talented Electronics Technician. Our valued clients provide us with a variety of service requests that change or are new each day, the successful completion of which is challenging and personally rewarding. The position vacant is well suited to one who enjoys designing, building and fault finding electronics equipment. A sound understanding of basic electronics theory and power supply topologies is essential. Using schematics, the suitable applicant will need to demonstrate their ability in efficient and effective navigation and fault-finding of complex electronics circuitry. As more than one-hour total travel time to and from our location in Hornsby has proven impractical in the past, only local applicants need apply. If you are prepared to apply yourself enthusiastically, with study, our company-provided regular training is of great benefit and the work environment stimulating. Find us at www.switchmode.com.au

Using another oscilloscope, I could see the "Unblocking Signal" coming into the CRT control DC regeneration circuit. This varied with the front-panel controls but I had no definitive figures to work with. At this point, I noticed two circuits that were very similar, one for intensity and the other for focus, so I compared their voltages. It turned out that the focus control voltages were much higher than those for the intensity control circuit (95V compared to 45V) but was this significant?

Because the BAV21 diodes were obviously recognised as troublesome components by Emona, I decided to check all such similar diodes in both of these control circuits. To do this, I unsoldered one end of each diode and measured its leakage with x100kΩ ohmmeter but I couldn’t find any that were faulty. Similarly, I checked all resistors over 47kΩ and even replaced C846, C847 and C848 (1nF 6kV) in case they were leaky but I was getting nowhere.

By now, I was seriously considering abandoning this project when a colleague volunteered to look it over for me. It took him only an hour to find the culprit which turned out to be CR836, a BAV21 from the wiper of the CRT Bias control (RV821) which I had already checked. It was indeed leaky but how did I miss it? I could swear I had measured it to be OK out of circuit but obviously I had missed it (I must be getting old). Anyway, this fixed the fault completely.

I must admit that I am now highly suspicious of all BAV21 diodes and will in future always measure them completely out of circuit as they never go dead short. Instead, they always seem to go high resistance!

Interestingly, I went back in afterwards to see what voltage differences it had made and found that the only change was to G1. This now varied from -62V (minimum intensity) to
-43V (maximum intensity).

Ironically, in the course of measuring these voltages, some small sparks were produced as the meter probe touched the solder pads and suddenly the fault was back again. This time, it was CR835 that was faulty.

A real recorder

An enterprising client of mine purchased an old but beautiful Technics RS-1500US professional reel-to-reel tape recorder. In fact, he bought this "Isolated Loop", 3-motor, 3-speed (15, 7½ , 3¾ ips), direct quartz locked-drive, 4-head tape deck along with 40 10½-inch reels of tapes. In its day, it was probably one of the finest tape recorders made and he paid for it outright by selling some of the tapes on eBay.

However, he struck a catch – the deck was faulty! None of the "turbo-touch" controls were working!

Unfortunately, access to the circuits in this machine is very difficult. Those Japanese must have very small hands, because you almost need to perform keyhole surgery when fixing this deck.

When I finally managed to remove the power supply and main control circuit, I started by checking the main power rails (21.4V, 20V, 15V and 5V). These were all spot on so I moved on to the six microswitches (or remote control), which controlled the input NAND gates of seven logic ICs via 12 diodes. These in turn control the motors, plus an additional timer circuit.

I soon found that the STOP input rail (D3, D6, D9 & D19) was perpetually held low, so I checked these diodes as well as D15, D16 (from the timer), D17 and D18 plus TR8, TR24 and TR25 but they were all OK.

In fact, I didn’t get anywhere until I checked out IC7 (M53200P), especially pin 11 (LO) which is controlled by pins 12 & 13 (HI). This in turn is controlled by TR10 (2SC828) and D26 which was switched off.

Both TR9 and TR10 measured OK, which just left C17 (4.7μF 10V) on TR9’s collector. This was short circuit and replacing it fixed up all the functions.

Zapped Sharp

Mr Marsden brought in his Sharp CX68GS TV set that had been killed during a storm – or so he thought as he wasn’t actually there when it happened. Looking inside, I quickly found that R706 (1.8 7W) was open circuit and that chopper transistor Q703 (2SC4429) had shorted.

With a switchmode power supply like this, you have to check everything in order for it not to blow again – especially after a cataclysmic event like a power surge or lightning strike. Subsequent checks revealed that C715 (0.01μF), R711 (47Ω 0.5W), Q702 (2SC3807) and C716 (2200pF 2kV), the latter sporting a black burnt spot (and probably the main culprit), were all in various stages of destruction. I replaced them all with identical components, some of which had to be ordered in.

I switched it on and BANG – the front face of the chopper transistor blew off completely. Oh dear! I installed a new chopper transistor and replaced Q702 (2SC3807), 5A fuse F701, IC703 (SE120N) and the optocoupler (IC702), before rechecking everything I had already done. That done, I gingerly switched the set on again and . . . CRACK! – it all blew up again (and the parts ain’t cheap).

This time, an even closer recheck of everything revealed that only transistors Q703 and Q702 and the fuse were damaged. The 2SC4429 is rated at 1100V, 8A & 60W and the 2SC3807 at 30V, 2A. It seemed obvious to me that the chopper was failing and blowing its driver, so I decided to install a BU508A (rated at 1500V, 8A, 125W) instead of the 2SC4429.

This time, when I switched the set on, it came on perfectly with all functions working.

At that point, I decided that I’d had enough of the 2SC4429s supplied to us and sent back all the other unused ones I had ordered. Frankly, I think that they were underrated counterfeit copies of the original. In fact, if you open up one of these devices, the "chip" is much smaller than you would expect for a 60W device. Unfortunately, there are a few re-labelled counterfeit power transistors around these days.

After soak testing the set for a week, it went back to Mr Marsden. However, the very next day, he phoned and reported that it had died again. I couldn’t believe it!

Back came the set and I was expecting the same problem as before. However, I was relieved to find that it was only R705, the other sister 1.8 7W surge resistor from the bridge rectifier. That was some months ago and so far, so good.

The humming Philips

I’m still getting in ancient Philips TVs which their owners are reluctant to upgrade and the faults are becoming more weird and interesting.

Recently, a 1986 CH685 KL9A-3 chassis came in, its owner complaining of a loud persistent hum in the sound. I thought at the time it would be a doddle to fix but it turned out to be rather tricky.

In greater detail, the problem was 50Hz hum which was causing the sound to "quaver" at low volume. This set has a separate audio output amplifier board with its own power supply, so I checked the 33.5V and 15V rails for ripple and tried unplugging the input to the amplifier module to confirm that the problem was elsewhere.

I then backtracked towards the tuner, unplugging leads as I went, and noticed that the hum would vary as I moved the modules and their wiring harnesses. Aha, I thought, dry joints on these boards might be the answer but there were none to be found.

It was then I noticed that the hum was varying with picture content and that the objects on the screen were very bright and "contrasty". As a result, I checked the aquadag screening from the CRT to the neck board and found that uncoupling it removed the hum. I then clipped a jumper in its place and earthed it to the chassis.

The sound was now back to normal and I spent a long, long time trying to find out why this was so. This involved following the earthing to the aquadag back to the EHT and tripler sections but I couldn’t find out what was wrong. This wretched set was going to fight me all the way.

I then began to notice that the most sensitive board to movement was the 5-pin DIN input module which also had a small 15V power supply and an analog switching IC (IC7031, HEF4066). And although I couldn’t find anything obviously wrong with it, I did notice that unplugging the TV input into the IC also removed the hum. I checked the switching voltage from the control panel TV/VCR switch and noticed that unplugging it also eliminated the hum.

I now felt that the problem was around this area and so ran DC checks all around transistor Q7025 and IC7031. This turned to be quite fruitful as it quickly revealed that R3040 (22k) to the base of this switching transistor was open circuit, leaving the transistor permanently switched on. And that in turn meant that two analog switches from the external DIN socket were connected permanently.

So why did this resistor cause so much strife? My theory is that when it went open circuit, it meant that the switching line cable probably became inductive and was picking up 50Hz buzz from the CRT and passing that through into the audio output stages. But then again . . .

Flushed with success

I was asked by a friend just recently if I could give him a hand with some toilets! Apparently, he wanted me to help flush out some problems!

Now these wasn’t just any toilets. Instead, they were the public toilets for a very expensive architect-designed restaurant and were the latest when it came to fantastic features.

For example, the men’s latrine has an automatic flush and the glass hand "basin" was also completely automatic. The water was controlled by a sensor and piped around the face mirror and flowed onto a piece of glass before exiting into a trough. You washed your hands in the water from the mirror faucet.

But it didn’t end there. The toilets had clear glass cubicles and when you went inside, a sensor would make them go opaque for 90 seconds or until the sensor no longer detected any movement. There is, fortunately a fail-safe feature – in the event of a power failure, the glass remains opaque. Flushing was via a touch sensitive area on the glass wall. All this for a mere $100,000 and that was just for the glass panels.

Well, with all these high-tech features, things were bound to go wrong and they did!

What no-one had envisaged was that apparently some young women liked to go into the toilet and take a nap! The problem was that after 90 seconds the glass would clear and they were visible in all their glory!

This was, of course, the down-side. The upside is that if someone is drunk or ill, someone will see their plight and presumably call for help. Anyway we had to go back and see what could be done to make the sensor more sensitive.

The infrared proximity sensors are mounted in the roof, one above each cubicle. Initially, my friend specified very sensitive small-footprint sensors but it was felt by management that a wide footprint would better, even if it meant a reduction in sensitivity. Unfortunately, this led to the current problem but I guess that’s all part and parcel of the teething problems involved with new technology.

Anyway, we replaced each sensor so that the smallest movement of someone sitting on the toilet would trigger it. We also realigned the sensor with the crosshairs towards the front of the seat where one’s head would normally be. The theory is that if the occupant falls asleep, their head will remain in the sensor’s footprint and slight movements will continue to trigger it.

This we hope will be sufficient, otherwise we may have to contemplate fitting a horizontal infrared LED, mirror and receiver from behind the occupant.

So there you have it – a fully automatic, timed and hygienic ergonomic superloo. High-tech is everywhere!

Grumpy Grundig

I was recently asked to look at a 1996 Grundig Megatron M82-115/9 IDTV/P+P employing a CUC1882 chassis. Initially, I thought the "Megatron" bit referred to the extra-long model number but I later discovered that it refers to almost every aspect of this luxury TV, which is modestly described in the sales brochure as a "Megavision Monolith". And in case you’re wondering, the set came with a $A9000 price tag – and that was back in 1996.

The client complained that the convergence was poor and that there were lines across the picture. He also clearly implied that there wasn’t really much wrong with it that a competent technician and $75 couldn’t fix! Yeah, right!

When I opened up the 76cm set and examined it on the bench, I did notice that the convergence was out but the only controls were the purity and static convergence rings around the neck of the Toshiba A76KJJ96X98 picture tube. Fortunately, it didn’t take long to fix the convergence problem, though I have to say the 2mm dynamic convergence errors didn’t impress me in a $9000 television with no additional correction.

The main symptom, though, was still there and that was the diagonal patterning which, when the colour saturation was increased, became magenta stripes. I then noticed that the Picture-in-Picture didn’t have these stripes but the picture was ghosting. This combination of errors was a dead giveaway for a fault in the digital comb filter in the "features box".

I couldn’t continue without the aid of a circuit diagram so I contacted Grundig, who were extremely helpful, and discovered that an exchange features box was $385 trade. However, their technical support went beyond the call of duty and told me to look out for a crack in the tripler and to change all 26 electrolytic capacitors inside the features box (upgrading their values to 220μF 35V 105°C).

This advice was spot on. There was a crack on the concealed underside of the tripler (where the internal focus resistor is) and indeed all the electrolytic capacitors inside the features box were shot.

Replacing all these items fixed the problem – but at a lot more than $75. Ironically, after fitting the new tripler, the set would try to start but would stop almost immediately. This was due to your truly forgetting to reinsert a plug immediately under the tripler. It took a long time to find and correct this as it was so well concealed.

One very impressive feature of this set was the amazing menu system, which was a complete in-depth instruction book available in multiple languages, plus a series of complex test cards and a service menu that can perform service adjustments after dialling up the service code of 8500.

The beautiful Philips

I had another beautiful TV come in this week and I found it hard to believe how technically advanced it was for a 1993 model. This was a Philips 32FL2881/75R employing an FL2-G chassis and boasting a 16 x 9 widescreen, progressive scan, Dolby surround sound, picture-in-picture and 100Hz scanning, to mention just a few of its features. Of course, all these features came at a price, the set costing around $6000 11 years ago.

This particular set was dead and pulsating, so the first step was to check out the HT rail. This involved shorting out the base and emitter terminals of the line output transistor and hanging some 240V light globes (200W total) across the collector. When I switched it on, the power supply had no trouble delivering 140V into this 200W "dummy" load.

This was screaming at me "flyback transformer", except that when I removed the shorting jumper and disconnected the horizontal deflection coils, the B+ was restored and the set stopped pulsating. In the end, I decided that the line output transformer was still the most likely culprit as there was no discernible EHT.

This turned out to be correct. After fitting a replacement and aligning the double focus control, the picture and sound were once again excellent.

Where will it end?

With modern TVs, it is not unusual to include a set of software convergence and other adjustments for both PAL and other systems as well. However, we recently came across a convergence problem that initially had us stumped.

The set concerned was a Panasonic TX-51P800H (GP1 VP chassis) that came in with a faulty green tube with a focus problem. After replacing the picture tube and doing all the alignment adjustments, everything was looking great apart from a bizarre fault: the convergence was out only in the tuning menu and the double window option.

At first we thought that this was due to a faulty EEPROM or digital board. We were about to order and replace these items when the Panasonic Technical Officer whispered in our ears.

Because we didn’t have the service manual, we could be forgiven for not realising that the Tuning Mode Search Menu switched the set to Progressive Scan and there is a separate set of convergence controls for this. Similarly, the double window has its own set of controls too.

To fix this, you get into the "Service Adjustment Mode", select convergence, and then select the mode (system, menu, double window, etc) that you wish to adjust. Where will it all end?