Email Address:

Lost your password?

This is the legacy website; please use the new website.

Serviceman's Log

Big TV sets can be a nightmare.

By The TV Serviceman

Items Covered This Month
  • APPLE m3502 21-inch monitor
  • PHILIPS 21PT128a/75R TV
  • PHILIPS KR6687T TV set
  • SONY KV-S29MN1 80cm TV

Things have changed a lot since I first took up servicing, even going back to the monochrome days but, particularly, since the advent of colour. I often used to hear, "My set is 25 years old, it has never been repaired and it is still going perfectly"

Well, frankly, I have yet to see a perfect 25 year old TV set that has never been fixed, but I am seeing a lot more 10-year old TV sets that have never had their backs off. To my mind, that is a fantastic salute to the manufacturers.

On the other hand, I have had several 1989 Mitsubishi TV sets recently with intermittent total or partial vertical collapse and the only fault has been faulty joints. The picture tubes were still in excellent condition.

There was a time when the only choice in acquiring a TV set was the size of the picture tube and design of the cabinet. Nowadays the list of options is enormous and many features are incorporated whether one uses them or not.

My own kids have never known monochrome TV and probably would not even know how to operate any old set without a remote control. It's hard to find a TV set today without AV inputs, while turret tuners and even pushbutton tuning systems have long gone the way of the dodo.

Click for larger image

So, what to do with a top-of-the-range TV set that is ten years old and cost over $5000 new? If some of the deluxe features fail, one can probably live without them but most times it is a general failure. So, should the set be abandoned, knowing that for a third of the original price, it can be replaced with one having many more features?

It is most likely that the total waste of the excellent technology which that cost heaps of money, which more or less impels the owner to get it fixed.

So it was with a 1992 Sony Kirara Basso that was delivered (fortunately) to the workshop.

The 34in/80cm set weighs 81kg and even the 29in/68cm version weighs 56kg - apparently due in part to the heavy iron support frames used in the Trinitron tube set-up.

I have written before on this series, using the G1 chassis. This particular model was an overseas version - KV-S29MN1 (SCC-F51B-A) - and the fault description was "dead".

When this set decides to spit the dummy on the power supply board, the result can be spectacular, not to mention expensive. When the pyrotechnics finish, it is best to replace all the following 16 parts: Q601, Q602, IC601, D626, D603, D604, D605, D606, VDR601, R606, R625, R626, C616, C618, C617 and C619. (Q601 and Q602 should be replaced with 2SC4834NS.) And if genuine Sony parts are used, the trade price is over $300!

Why does it blow? Apparently because four (blue) capacitors become leaky. In my case, these repairs were straightforward but when the set was switched on, all I had was a white line across the screen and no sound. Also, there was no AV in or out.

On the D board I replaced the vertical output IC503 with an STV9379 and electrolytic C570 with a 220μF 25V, the latter mounted on the PC board track side, between IC503 pin 4 (-) and chassis (+). This made no difference and I then replaced Q1801 and Q1802 on the VC board, plus C1802.

Others should be warned that there are two different versions of the VC board in this series, which are not interchangeable. I also changed all the electros on this module but still wasn't getting anywhere.

Despite this set having very short leads (the eight extension leads cost an average $32 each), the main chassis can be balanced, very precariously, on its side, to get limited access to the motherboard's tracks.

With the multimeter I confirmed that all the voltage supplies were correct; there are dozens to be checked.

Next, I concentrated on the jungle IC, IC3501 CXA-1464AS, on the A board. Despite having the correct voltages going in, there was no vertical output on pin 31. The oscillator was working on pins 32, 33 and 34 but absolutely no output drive. Reluctantly I ordered and replaced yet another expensive 40-pin high density IC but to my horror I found there was still no vertical output.

By now I had exceeded the estimate I had given the client and had to get back to him and report on my progress. Suffice to say he wasn't impressed, nor was he prepared to continue and said I could keep the rotten set.

This wasn't the best news I could have had; I now owned an expensive lemon, wasn't sure how to proceed and what to do to recover the costs and effort already invested. I put it aside until such time as there was a lull and also give me time to collect my thoughts.

Months later, I re-measured every pin on the 48-pin jungle IC and checked it off against the circuit. Everything was correct except for one pin. Pin 48, the SDA or Data Line, of the I2C digital control, was low at 0.64V instead of 5.6V. At last I had a clue that might possibly lead me somewhere. But this 5V line goes all over the set, to almost every IC.

It took an awful long time to discover where this voltage disappeared, by the very lengthy process of unsoldering each and every component on the rail until the 5V was restored.

This set has two I2C lines, one for the tuners and front end and the other for the rest of the set. It comes out of the microprocessor M module on CN1108 pin 31 from the two EEPROMS and pin 22 of IC005.

Logically, I started at the jungle IC, IC3501, and unsoldered pin 48 but it involved a very long time via very diverse routes. I won't go into how much time was wasted following just one red herring; the V-PROTECT line, which was (for obvious reasons now) giving incorrect voltages.

But I finally arrived at pin 14 of IC3503, CXA1315P, which was loading the 5V line. Of course, once I had found the culprit everything fell neatly into place. Pin 5 of this IC connects via CN1132 pin 10 (H-TRAPZ) to CN501 D Board and from there to pin 7 CN502/CN1801 pin 5 via R1816 27k, which all has to do with east/west and north/south correction.

Replacing IC3503 finally fixed all the problems at one go. My challenge now is to sell the set and recover my costs - but I must say it does perform very well and the twin tuner picture-in-picture is a treat.

Subsequently, I have had a few more of this model with miscellaneous faults. One had intermittent AV inputs which was due to faulty joints on the hinged connector between J and B boards CN2301/CN308 and CN2302/CN309.

Another set had total or intermittent no-sync. I checked PS01, 0.6A fuse, on the Teletext module V which supplies 5V to the text processor, IC02. Then I re-soldered several faulty joints on IC208 and IC209, 9V and 5V IC regulators, on A1 Board. From here I eventually found that the problem was IC502, the SBX1692-01 Digital Comb Filter on board B1. The main problem for that was the cost - $241.93 trade, plus freight.

Philips Anubis

My next story concerns the Philips Anubis type chassis. This chassis covers hundreds of models of TV sets and has spawned a large number of variations; eg, "Anubis A" to "Anubis S" and then "Anubis S AA" to "Anubis S DD". These are the ones I know about and I'm sure that other technicians could nominate more.

In my case, the service manuals for this one chassis occupy nearly one filing cabinet drawer. It also makes it very difficult to match the exact circuit with the set model, as the Philips Product Survey only shows "Anubis A", "B", and "S", etc and not the letters after that.

Against this background, here is a story about one of these sets. It was a 1997 Singapore-built Philips 21PT128A/75R with an "Anubis S DD" chassis. My customer, a Mrs Blossom, had brought this set in from a northern beaches suburb, so it was a little rusty. And although the fault ticket read "dead", it was actually pulsating very quietly.

I shorted out the base and emitter of the horizontal output transistor (7445), a BU1598CX, and then hung a 100W globe across the collector to emitter. Switching on lit the globe and the HT settled at 95V. Well, at least the power supply was good but it also suggested that the horizontal output transformer (5445) probably wasn't.

Hoping that was all, I ordered a new one. Unfortunately, because of all the variants, it is important to order the correct part number, which in this case was 4822 140 10557. And, as luck would have it, it is directly equivalent to the Spanish-made DIEMEN HR7815 transformer.

(Note: the 4822 140 10486 is no longer available for the 50cm Anubis S BB and 35cm CC and the HR7815 has to be modified to fit. As well, C2450 has to be changed to 680nF 250V.)

When the new transformer arrived and had been fitted, I was disappointed to find that the set still had the same problem. I then found that the horizontal output transistor (7440, BF422) was short circuit but replacing this still made no difference.

Using an oscilloscope, I found that in fact there was little or no output from pin 37 of the jungle IC (IC7225). A TDA8362 replacement for TDA8361 finally restored the EHT and produced an intense white raster. A few measurements then showed that resistor R3300 (10W), which feeds 160V to the video output stage, was open circuit. Finally, after a few adjustments I had a good clear picture.

I don't know how this sequence of events happened but I certainly felt unlucky to encounter so many faults in this set. Still, the end result was quite satisfactory.

Mr Ellis's Philips

I have repaired many Philips 2B-S chassis TV sets over the years but now feel that they really are getting too old. However, when it is quiet, I still occasionally take on a repair for this series, even though it is against my better judgement.

Recently, I foolishly took one on at a Mr Ellis's house. He had a 1987 28CT8893/75T or KR6687T and he was complaining about a "kink" in the picture, about two thirds up the screen. I figured that it had to be easy and was probably C2571 (100μF 63V), which does give trouble.

When I arrived, I removed the back and turned the set upside down. This way I get a manageable access to the PC board. I pulled the chassis out far enough to work on it and replaced the capacitor. I also spent a long time resoldering the entire chassis.

Click for larger image

An hour later, my cockiness had been completely knocked out of me. I still had the same fault even after changing the IC and all the nearby electros. I measured the HT to be spot on at 140V and I also measured all the voltages around IC7570. The only significant one out was pin 8, which the service manual has at 14V but which I read as 19.6V.

It's quite amazing, really, how one can work on these sets for 15 years and yet never notice some of the changes that have been made to the model over its life. In this case, the 26V or 27V supply rail from pin 4 of the horizontal output transformer had been changed. Diode D664, resistors R3647 & R3646 and fuse F1646 had all been changed from the circuit diagram - the diode had another strapped in parallel, R3647 had been deleted and R3646 reduced from 2.7Ω to 1.5Ω. The fuse was now an 800mA button type.

I checked all these and they were all OK. After a lot of mucking about, I found that by shorting L5646, R3648 and R3646, the symptom would disappear completely and the supply voltage would increase by 0.4V. Interestingly, this also made IC7570's pin 8 voltage rise to 20V.

Well, I searched my soul but I couldn't see why such a small increase could fix this symptom. And as there were no signs of stress anywhere, I decided that I would allow this "bodgie fix" to remain. After all, the set is now 15 years old and is already well past its use-by date. But at least Mr Ellis would get a little more life out of it.

I returned the set to Mr Ellis, made a nominal charge and warned him that he was probably working on borrowed time. I think he understood.

She's apples

Some years ago, while reading the classified ads, I spotted a 21-inch Apple computer monitor for sale for just $50.00. Although unsure of the connections or resolutions that Apple computers worked on, I thought it was worth the risk as a potential standard PC monitor or at worst, a video monitor.

Click for larger image

Unfortunately, when I arrived to inspect it, I discovered that I would not be able to see it working - there was no computer or even a lead to connect it to the computer. The only sign of life was the green power LED on the front panel.

However, I did discover that the deflection and EHT sections were probably OK, by observing the collapsing raster when it was turned off (any problem here would have made it too expensive to repair).

In the end, I bought it for just $40.00 (I haggled). It was worth a gamble - the cheapest 21-inch monitor at that time cost over $2000.00.

On the way home, I called into a couple of local Apple agents to try to get a video lead, as this monitor only had a 'Sun' socket. The first agent I called into specialised in secondhand as well as new Apple and PC hardware. He could only locate a secondhand cable with a 'Sun' connector to individual BNC connectors, which meant that it would have to be modified (ie, by fitting a VGA connector in place of the BNC connectors). I wasn't very keen on paying the $40.00 that he wanted for this secondhand lead but he challenged me to do better and refused to negotiate, so I left empty-handed.

The second agents sold just new Apple equipment but said they could obtain an original secondhand lead from a customer for only $200.00! Hiding my shock and regret for not getting the first lead (I couldn't possibly go grovelling back to the first shop), I calmly asked about obtaining a circuit diagram for the monitor but was told that this was not available.

By the way, I have since discovered that WES Components (Phone 02 9797 9866) now sell 'Sun' type leads and adaptors for just $24.50 (now that I don't need one). However, they still have to be modified as Apple didn't use the standard configuration.

Arriving home, I hit the Internet in search of as much information I could find about the Apple M3502 21-inch monitor. It turns out that the 'Sun' connector is also known as a 13W3 video connector - see Fig.1.

Click for larger image
Fig.1: signal input connector as seen from the rear of the monitor

Apple's 13W3 standard differed from the normal, swapping the red and blue. However, as there doesn't appear to be any standard for the sync connectors, Apple's configuration is as good as any.

The pins of interest are A1-blue, A2-green, A3-red, 2-Vsync and 6 -Hsync. I decided to hard-wire an old VGA lead from a wrecked monitor to the back of this socket. Fortunately, there is enough room between the circuit board that carries this socket and the one behind it to drill a hole large enough to take the VGA cable's original mount (I had to nibble out a square hole).

The other end of the video lead is the standard VGA connector on the lead and did not have to be modified. Its pin attachments are as shown below.

Signal Type : Analog
1 - Red 6 - Red Return11 - ID Bit
2 - Green 7 - Green Return12 - ID Bit
3 - Blue 8 - Blue Return13 - Horiz Sync
4 - ID Bit 9 - No Pin14 - Vert Sync
5 - Self Test10 - Ground15 - ID Bit

The pins of interest are 1-red, 2-green, 3-blue, 13-Hsync and 14-Vsync. The earth pins are left as is.

The next step was to set up the computer. Hooking the monitor up to the computer did produce a picture but with no vertical or horizontal sync and only half height. It was time to hit the Internet again.

This time I discovered that the Apple M3502 monitor has a fixed resolution of 1152 x 870 pixels and a horizontal frequency of 67.5kHz. This is very close to the PC standard of 1152 x 864 pixels and 68.7kHz. The vertical refresh rate is 75Hz and the dot pitch is 0.26mm, which is pretty good even by today's standards.

Changing the computer's display resolution and scan rate to 1152 x 864 at 75Hz did lock the display in vertically (sort of) but it was still only half height and out of sync horizontally. I was able to determine, though, that there were a multitude of problems relating to the sync, blanking and linearity.

Because I had no circuit diagram and because the monitor was seven years old, I decided to replace all the electrolytic capacitors in the power supply, EHT and deflection circuits (about 50 capacitors in all). I didn't replace the main power supply reservoir electros as they all checked OK and were expensive. About half of the 50 electros replaced were way off value and this cured most of the faults except for the main ones - still no horizontal sync and a half-height picture.

Careful inspection of the picture revealed that the vertical oscillator was running at twice the rate it should have been. I removed the small signal deflection board, checked everything but could not find any faults. As a precaution, I replaced all the electros but this had no effect.

Click for larger image
Fig.2: standard 15-pin VGA connector.

I desperately wanted a circuit diagram but couldn't justify the money wanted by the circuit diagram sellers on the web. In the end, I carefully re-examined all the information I had acquired and discovered that instead of using standard positive-going sync pulses, this monitor required negative-going sync pulses. Fortunately, Windows lets you specify the pulse polarity and all graphics cards support either - all I needed was a suitable driver.

I downloaded and installed the driver development kit from Microsoft and started trying to create a driver for the monitor without much success. I had since discovered that the monitor could also operate at 1024 x 768 at an 85Hz refresh rate, so I incorporated this setting into the driver as well but it still refused to work.

In fact, the sync looked exactly the same on-screen as before and checking the sync pulses coming from the graphics card with an oscilloscope revealed that they were always positive-going, no matter what was specified in the driver.

I checked the spot in the Windows registry where monitor resolutions are stored and discovered that they were as they were supposed to be, according to the driver (that is negative sync for H and V). So why were the polarity settings being ignored?

In the end, I abandoned this approach after discovering various software packages that let you change the sync pulse polarity on the fly. The best package I found was a shareware program called "Power Strip" and it does a lot of other stuff as well, including keyboard shortcuts (necessary if some rogue program changes the screen resolution and you can no longer "see" the screen to change it back because it's out of sync). With this package installed, the monitor instantly locked in and I had full height - all that remained was to reset the internal presets for an optimum display (this monitor had been well twiddled by someone before I came along).

One day, while pondering the meaning of life and why the monitor driver didn't work, I happened to test an Osborne MPV 1024 NI 14-inch monitor. This worked fine at a resolution of 640 x 480 but on 800 x 600 or above, there was no sync and a half-height picture. I spent some time looking for a fault before it dawned on me that I had seen this problem before.

A search on the Internet revealed that this monitor works on positive sync pulses for 640 x 480 but uses a combination of positive and negative sync pulses for the other two resolutions.

In addition, I managed to find a driver for this monitor (created by a user, not Osborne). I changed a couple of scan rates slightly to comply with the monitor's specifications (the driver must have originally been created by trial and error) and then installed it on my computer.

As before, Windows just ignored the sync polarity settings (as per the Apple monitor). However, the "Power Strip" software permits you to specify different polarity settings for the horizontal and vertical sync pulses for each resolution and so the Osborne worked perfectly.

The Osborne monitor was about the same age as the Apple monitor and I started wondering about this. All PC monitors since about 1995 work with on positive-going sync pulses, so it appeared as though Microsoft had decided that all subsequent versions of Windows (ie, from about Windows 98 on) would just ignore the sync pulse polarity settings required by these older daggier monitors.

To test this theory, I installed Windows 95 on another computer (my machine uses Windows 98), installed the Osborne driver and found that the Osborne monitor now worked perfectly on all supported resolutions. What's more, so did the Apple monitor with the driver I had created all those months ago.

Click for larger image
Fig. 3: two of these sync pulse inverters needed, one each for horizontal and vertical pulses.

After using the Apple monitor with "Power Strip" for some time, I subsequently upgraded my operating system from Windows 98 to Windows 2000. Unfortunately, my version of Power Strip (ver. 2.78) would no longer work (version 3.0 and above may work with Windows 2000 but I haven't tried it).

In the end, I modified the Apple monitor itself by installing a couple of simple sync pulse inverter circuits (see Fig.3). This simply involved cutting the horizontal and vertical lines inside the monitor (pins 6 & 2 to the sun connector) and hard-wiring the inverter circuits in series. A 6V rail to power the circuit was taken from a plug connection on the main deflection board directly above.

These hardware modifications enabled the Apple monitor to work perfectly without any special software. In Windows 98/Me, you first set the resolution to 1152 x 864, then click the Advanced button and change the refresh rate to 75Hz. Windows 2000 and Windows XP are similar except that a different refresh rate can be set for each resolution.

Note that some lesser graphics cards don't permit you to change the refresh rate. Some don't even support 1152 x 864, while those that do don't necessarily support a 75Hz refresh rate at that resolution. Those that did work OK included S3 4MB AGP, TNT2 M64 (both PCI and AGP versions) and GeForce 2 cards.

Anyway, the old Apple monitor served me faithfully for about 18 months. I then sold it and bought a nice flat 17-inch monitor that supports every resolution up to 1600 x 1200.

Share this Article: 

Privacy Policy  |  Advertise  |  Contact Us

Copyright © 1996-2019 Silicon Chip Publications Pty Ltd All Rights Reserved