My first story this month is about a Teac EU-68St which employs an 11AK19 chassis. This is a 68cm silver-coloured stereo TV set, manufactured in Turkey around January 2000. I was surprised to see it come into the workshop, as it looked brand spanking new.

I could tell from the owner's face that he was extremely upset; it was no longer under warranty and it had failed so soon. However, when I took down the details of his address, an absolute water frontage on the beach, I wasn't so surprised.

The set was dead and it didn't take long to diagnose a shorted horizontal output transistor (Q605, BU2508D). This was due to an overheating horizontal output transformer (TR602), no doubt hastened to its demise by the layer of salt deposited by the onshore winds. In fact, most of the metal screening cans were already rusty.

Items Covered This Month Teac EU-68ST TV set. Blaupunkt MS-70-109.VTM TV set. Mitsubishi HS-E11(A) VCR. IBM Thinkpad T20 computer. Troubleshooting an Internet connection.

An order was placed for the parts and a service manual and they arrived some time later. When the parts were installed and the set switched on, I was faced with a new symptom - no vertical deflection. As a result, a new vertical output IC (IC701, TDA8351) was ordered.

This duly arrived and was fitted only to produce symptom number three: no picture, just a raster with retrace lines.

This time, I knew I would be faced with a difficult problem and that I would have to consult the service manual. Unfortunately, this document is somewhat disappointing and mainly consists of 29 pages of a detailed "Bill of Material List". The circuit and PC board layout diagrams were also difficult to follow, being rather poor photocopies.

Because the problems with this set had all started with the horizontal output transformer, I was looking for components that might have been destroyed by the breakdown. With that in mind, I started by checking the beam current and ABL (Automatic Brightness Limiter) circuit. This is easier said than done because the main PC board is small and carries unmarked surface mount components on the copper side and normal-size parts on the other.

As can be imagined, the assembly is quite compact, which makes it difficult to trace the circuit. Anyway, that circuit eventually checked out OK.

One of the vital symptoms I noticed was that there was no OSD (On Screen Display). This suggested that the problem was likely to be affecting the video output rather than the small signal circuits from the tuner/IF onwards. In an effort to isolate the problem area, I connected a monitor to the AV output socket and found, with some mucking about, that I could get a clear full-colour picture on it from the tuner.

Next, I connected a colour bar generator into the video output sockets but could not display that on either the TV set or the monitor, regardless of what I did with the remote control. I then measured horizontal pulses and the sandcastle on pin 41 of the jungle IC (IC401, TDA8844/N2) and checked all seven of the voltage rails from the power supply with the CRO, looking for excessive ripple.

I also checked the five voltage rails sourced from the horizontal output transformer and these were all OK.

My next step was to check for the blanking pulses from IC501 (SDA545XOTP) but I couldn't detect them. But when I unsoldered pins 23, 24 and 25 of IC401, they reappeared. This was enough proof for me that the jungle IC was crook, so I ordered a new one.

However, when it arrived, I was disappointed to find it made absolutely no difference. It was only then that I found that these pulses were there only when the OSD was on (although I couldn't see this on the screen) - pressing the menu button would produce the pulses.

From here on in, things just became worse. I was following one blind alley after another and pursuing all sorts of strange and wonderful theories. There are a lot of circuits in this set. And it's not obvious what all of them do. For example, I couldn't work out the function of Q425 and Q426; a possible spot suppressor - who knows?

I phoned Teac for technical support and they suggested that I replace C604 (.047μF) on the beam limiting line from the horizontal output transformer. I did this but it made no difference.

Back off mate!

To be honest, I was close to quitting and returning the set to the customer. This elusive fault was eating into my time and the repair was becoming uneconomic, even for a fairly new TV set. In the meantime, the customer was pressuring me to fix the problem quickly but it doesn't work like that with this type of problem. You need space and time and the pressure tactics don't work.

Knowing that it's probably an insignificant little part that's causing the problem doesn't help either. And neither does the customer's implication that I don't know what I'm doing -even if I don't!

And then, suddenly, there was a clue. I was running my finger across IC401's pins (the jungle IC) and, somewhere around pin 22 & D701, the picture nearly appeared. Encouraged, I renewed my investigation into this area, even though I previously could not find anything wrong with it. And I now noticed that although the DC voltage and amplitude waveform didn't change much, the width of the vertical pulse changed.

I traced the circuit back via Q701 and ended up back at the vertical IC (IC701).

At this point, I also noticed that the blank raster wasn't perfectly linear in the vertical deflection and neither were the retrace lines, especially at the top. So was the new vertical IC faulty?

Well, before I replaced it again, I measured the DC voltage and checked the waveform on each of its nine pins. When I reached pin 6, I noticed that this was at 20V or so, despite being fed via a 100Ω resistor (R704) from a 45V rail. This meant that another 20V was being developed across R704, which struck me as a much too high a voltage across a 100Ω resistor, considering the likely current level in this part of the circuit.

I removed R704, a 1% metal film type, and measured it. It was nearly 5kΩ. I fitted a new 100ΩW resistor in its place and the rest, as they say, is history.

The picture was fully restored, although it required adjustments to the screen voltage and focus settings. The OSD appeared as expected and when I investigated the menu system, I found I could switch the 'AV-OUT' from 'TV' and the other 'AV-IN' sockets. I think, in retrospect, I should have been more observant and followed the likely path of the sparking horizontal output transformer and the vertical output IC failure - but it's easy to be smart after the event.

Blaupunkt MS-70-109 VTM

I have had a few Blaupunkt sets in recently, all about 10 years old. The fact that I rarely see them younger than that indicates how reliable they are. I am pretty sure, too, that they are now all made by Grundig.

The Blaupunkt in this case was an MS 70-109 VTM, using an FM 500-70 (7 669 840) chassis (probably equivalent to a Grundig 29701-057). Like most German sets it is somewhat upmarket, with a remote controlled motorised swivel base, sub-woofer, Toptext and menu system, etc. And, of course, it was as dead as a dodo.

In our hot humid climate, most faults start at the set's high voltage end and create further damage from there. Its owner had lost track of how old it was and as this was the first time it had broken down, he just told me fix it as soon as possible.

I initially diagnosed a faulty tripler (yes, there are still a few about), a faulty horizontal output transformer, and a crook horizontal output transistor (BU508AG, T541). The avalanche of faulty parts then continued with the horizontal oscillator driver IC (IC500, TDA8140), electrolytic capacitors C541 and C507, the east/west modulator diode (D546), C573 (0.36μF), plus collateral damage in the small signal circuits - eg, IC7010 and D7012 on the east/west deflection correction module.

All in all, this set was very sick but this bloke was determined - he wanted it fixed.

I replaced all the above parts plus IC500 and a few extra horizontal output transistors. I used a BU508A for the horizontal output transistor and the extra ones were due to a few mishaps which I won't go into here. I also found some electrolytic capacitors on the deflection correction module that were in dire need of replacement and there were poor joints everywhere - not surprising given the age of the set.

Anyway, after a lot of kerfuffle, up came a first class picture. I soak tested the set for a few days and then told the customer to come and pick it up. I then thought that I had better check the Teletext - there was nothing and neither was there any menu system. And so, after a hasty rearrangement of the pick-up schedule, the set landed back on the bench.

From what I can make out from the circuit, all the menu and text action comes (naturally enough) from the Teletext module. However, I noticed that after the set had been running for a while, I could just see a faint image in the background. I also noticed that vertical hold was rolling slowly in the text and menu modes.

Without extension leads, it is hard to work close to these modules. Nevertheless, with the hairdryer switched to high (I nearly melted the set), I found that I could almost recover a fully working Teletext (actually Blaupunkt Toptext) and menu system. Unfortunately, I couldn't quite pinpoint the culprits by then applying freezer.

I next measured the voltages applied to the text module and replaced all the electrolytic capacitors on this board but this made no difference. However, I did find one thing - while checking the voltages, I decided to make more space for the probes by removing the AV module (socket board), which I thought was redundant for the time being.

Amazingly, removing this module restored both the text and the menu but the colours weren't right because there was a red caste all over the text picture. There isn't a lot on the AV module - mostly analog switching ICs and transistors. The RGB lines go to SCART socket 1 and are terminated with 75Ω resistors. And I found that hanging an extra 75Ω resistor across R233 restored everything.

It would have been easy to stop there - one resistor fixes the problem; go home. But this would smack of the stripped apron approach. There had to be a real reason for all this and I was curious.

The original resistor measured perfectly OK, so I removed and cleaned the SCART socket itself, as well as the plug that fits into the main chassis. However, these proved to be OK. The only other thing in common with this socket was the U DATA line which feeds the AV module, the chroma module, transistor T364 and socket P15. I decided to check this out more carefully and my patience was quickly rewarded when I found that T364 (BC548B) was open circuit. Unfortunately, a new one made no difference, much to my dismay.

Next, I checked the other transistors (T361, T366 & T363) on the motherboard and followed the line all the way to the control unit module and to IC830 (UAA2022). I found that unplugging BP1 also "fixed" the problem but that left a blue caste this time (this line is called U screen B). Plug BP1 connects pin 3 of IC830 with the rest of the circuit and replacing IC830 made no difference. I also checked T2817 and IC2860 on the text module by substitution but this achieved nothing.

By now, I was really quite tetchy over all this - I had come so far and fixed so much but this elusive fault remained. There wasn't much left to test on the U DATA line - only the chroma module and I felt it was highly unlikely to be that as the picture and colour were perfect.

Nevertheless, I persevered and replaced IC5200 (TDA3505), as pin 11 is directly connected to U DATA.

That was it! Apparently, pin 11 is the pole for the analog switch that switches between TV and text, and it wasn't toggling properly. It had probably been damaged by spikes from the tripler, along with all the other parts.

All the other temporary fixes were just red herrings. This really was a beast of a fault but it was worth "sticking with it".

Mitsubishi VCR

I had an interesting problem with a Mitsubishi HS-E11(A) VCR, employing an Fo deck. The tracking was out because a white plastic roller on the exit guide had seized on its shaft.

I found that by freezing the stainless steel shaft, I could free it so that it would rotate even if I just blew on it. But once it returned to room temperature I had to use a lot of strength to rotate it at all.

I asked a knowledgeable friend who does a lot of VCRs what he did in these circumstances. His answer was to wick in some 3-in-1 oil at the top and bottom of the roller. So I duly tried that but all I succeeded in doing was make a mess with oil everywhere. The guide was still seized. I even froze it and then applied the oil but it always seized again.

Finally, I tried heating it with a hairdryer for a long time until I couldn't touch the metal any more. The roller was seized hard but back at room temperature, it was beginning to feel a little freer.

I then cleaned it with "metho" to remove all the excess oil and lo and behold, it freed up completely. Well, this poses more questions than it answers.

Had the roller seized because the plastic had contracted with time, or had the lubricant (if it used a lubricant) dried up inside? By cooling it, the shaft diameter shrank enough to free it. So, when it was heated, did the shaft expand enough to stretch the plastic? Or was it just that the "metho" dissolved any dried out lubricant?

Sorry, I don't have the answers but perhaps someone else does. Anyway, it fixed the problem.

PS: a new roller assembly would require a lot more effort and money to install and align than the machine is worth.

Computer problems

I had two interesting computer problems this month, the first involving an IBM Thinkpad T20, Type 2647 (Pentium III-700) belonging to a student. Its operating system was Windows XP Pro v2002.

The problem was that it wouldn't boot at all, either from a floppy, a CD or the 9.3GB hard drive. Neither could I get into the BIOS setup. The only thing I could see was the opening Thinkpad flash screen with an F1 option to go into the BIOS or F2 to select a bootable device. Selecting either would cause the computer to go to a blank screen and hang.

My approach to fix it was to remove the power supply, the main battery and the memory battery and switch on. Obviously, nothing happened! But when I then replaced them all and switched on again, everything now worked. XP took some time to boot up but running CHKDSK/F on the next boot automatically fixed things.

How it had happened I could only guess. No viruses were detected later and a few files were found to be corrupted; eg, dsprov.dll, esscli.dll and framedyn.dll. There were also four bad clusters but nothing that couldn't fixed.

Internet access problem

The second problem actually happened to my own machine. It's pretty ancient (to match its impecunious owner) but it can just handle Windows Me. The problem I had was being unable to access the Comsec (Commonwealth Securities) website in order to flog some shares to pay the overdraft.

Unfortunately, each time I attempted to go to this website, IE5.5 asked if I wanted to go to a secure site before timing out and displaying a "Page Unavailable" message. But much to my frustration, I could browse other websites.

I then tried accessing http:// www.comsec.com.au by following links from other sites (such as Netbank) but got nowhere. Eventually, after trying lots of other things, I swallowed my pride and sought help from both Comsec and from my ISP (Internet Service Provider), Zip Australia (now Pacific Internet).

At their prompting, I upgraded my browser to IE6 (plus updates), deleted all my temporary files and history, reset my defaults and set the security level to medium. But it made no difference.

We then tried a few fancy tricks like running a trace test at DOS level:
c:/>tracert www.comsec.com.au> c:\trace.txt. This produced a route to this website over a maximum of 30 hops which I then emailed to Comsec and Pacific Internet. I tried entering http://203.202.9.126 (Comsec's IP number) directly into my web browser but that didn't work either.

I was then advised to try http://www.commsec.com.au/Public/Content/CMD-ServiceDifficulties.asp. This failed the "Secure Connection" test and indicated either an SSL (Secure Socket Layer) or proxy setting fault.

I messed about with various proxy settings but it made no difference. Comsec were very helpful and came up with a 5-page screed on how to set up your browser for secure commercial sites. I tried everything on the list but nothing worked.

Finally, David Luger, Network Development Manager of Pacific Internet, was the one to crack it. He noticed that I had an incorrect DNS (Domain Name Server) entry (a DNS server links domain names to IP addresses).

When you do a "ping" or a "tracert" for a domain name, the system shows the IP address it is trying to access. And mine was selecting the wrong one. In fact, he noticed that mine was always looking up an obsolete address.

Anyway, David Luger suggested that this might be due to some "Internet acceleration software" that I may have previously installed, with the system now storing an old lookup address. This was beginning to ring some bells with me.

To track down the problem, he told me to try this at the DOS prompt from the root directory: dir/s hosts.*. This came up with three "hosts" files: hosts.tpx, hosts.sam and hosts.

The first two are sample Seagate and Microsoft files but the last one - c:\windows\hosts - was created by Modem Wizard (the Internet acceleration software I had installed). A "hosts" file, by the way, is nothing more than a simple text file that stores domain names and their corresponding IP addresses. It allows the system to bypass a DNS server for any domain names that are listed, thereby providing faster lookup (in effect, the hosts file acts as a local DNS).

Unfortunately, this list was now out of date - at least as far as the Comsec website was concerned. So, on David's advice, I simply renamed this file and tried Comsec again.

This time there was no problem and I had full access to the secure site. So, thank you, David. Nice one.