Timex 2068 Common Faulty Components Question

[hamsterbasher]

Hi,

I recently acquired a broken Timex 2068. My experience is mostly with the older Commodore variety so Times/Sinclair is fairly new for me and I dont have any history of common faults to fall back on. Since this machine is in such good cosmetic condition I would love to get it working again.

When applying power there is nothing showing up on video. I replaced the power supply with a recommended 15V DC 1A, negative center supply and the voltages on it look good. Looking through the circuit diagram I have done some testing. The +15V feed and filtered output looks good and also on the 78L12 the 12V rail also looks good. Measuring the 5V rail at Q1 (D43C1) is however low at around 2V. I assumed that the transistor may be bad but disconnecting that supply and providing a clean +5 from my trust power supply - limited to 1A the voltage remains at around 2V and the amperage clips immediately which indicates to me that there is a problem elsewhere.

Does anyone have experience of the 2068 that can point me in the direction of possible IC's to desolder first? Would the memory be a good place to start?I tried the temperature check method but none of them seemed to be getting warm so thats not something that could help me. There are also a couple of PN2222 running on that 5V rail. Maybe replacing those would be a good place to start also? Any help is appreciated.

Thanks
Hans

[1024MAK]

Hello Hans,

I’m not familiar with the TS2068 (American model) or the TC2068 (European version). However, I can give some general help.

From what you say, it looks like there is a short circuit on the +5V rail (significantly lower voltage at a significantly higher current than expected). Did you just disconnect Q1? Is CR1 okay? Are the electrolytic capacitors C40 and C67 okay? No leakage, top or bottoms bulging and no distortion of any other part of the case?

Have you done a resistance test (no power to the board, multimeter on a suitable resistance range) between the +5V rail and 0V/GND? Is it consistent if you switch to a different range? What do you get if you use the diode test range (assuming your multimeter gives a result as a voltage)?

For you to notice a chip getting hot, you would have to have it on for a number of minutes, but if the voltage or current is (being) limited, it will take longer before anything gets hot. How quickly something heats up depends on how much energy is being converted from electricity to heat, plus the time it takes for the actual semiconductor chip inside the plastic package to heat up, then some more time for the plastic case to heat up (plastic not being a good heat conductor).

If you have access to a IR thermal camera, that makes it much easier

DRAM chip faults are a common failure in many 1980s and 1990s computers. And one failure mode is a low resistance on the power supply connections. Although this is less common with the DRAM that runs on only a single +5V supply (the older 4116 which requires three supply voltages is known for this failure mode).

If the board has any tantalum electrolytic capacitors, these are also known to fail short circuit. Ceramic capacitors can also fail short circuit, but normally only if physically damaged.

Of course, any semiconductor that is connected to the +5V rail could be the cause. But transistors like the PN2222 rarely fail unless abused (such as being shorted out). Having said that, they are the easiest to replace. So if you suspect any, desolder it/them.

The schematic I looked at is here.

By the way, the board does not need the microprocessor or any memory chips to be operational in order for a display to be produced (although the display would be a mess).

I hope that helps.

Mark

[hamsterbasher]

Thank you Mark for the information.

There are a couple of Electrolytic caps but they look OK still. I know that is not a guarantee that they are still good so I will unsolder them all and check with my ESR meter. I think I will order replacements anyhow as always good to replace these on old machines like this.

I didn't unfortunately have much time to dig deeper this weekend but I did unsolder Q1 and then lifted the positive leg on C40 and was still seeing the same 2.8 ohms resistance between GND and +5 rail - so its starting to look lik ethge Power supply is probably good but definitely have a short happening somewhere further downstream. I am not seeing any damage on the ceramic caps so for now will assume they are all still good. I also didnt notice any Tantulam's but will take a closer look next time also.

I think next I will unsolder all the ram chips and see if that resistance changes at all.

Thanks again

Regards
Hans

[1024MAK]

I think it unlikely that it does have any tantalum capacitors, but as I’ve not seen one in real life, I can’t be sure…

You don’t need to desolder all the DRAM chips, you could just cut the VCC pins. If you cut them halfway between where they exit the IC package and the PCB, you can bend them so they they don’t touch. Then if that chip is not the cause, you can bend the leg back and solder it back together again.

Oh, and before I forget. Before doing anything else, very carefully examine the board with a magnifying glass and a very good light. It could be a solder bridge, solder splash or two things touching that should not be.

I fixed a U.K. made Memotech MTX board once, all it was, was a solder splash that must have moved slightly due to heat over the years. It looked like it had been there since the board was made.

Mark

[hamsterbasher]

So I ended up having a few hours to spare tonight so decided to remove the memory chips. I have gotten pretty good at desoldering 1980's IC's over the past few years thanks to way too many VIC20 memory issues to remember so was able to remove all without any issues. Measured the resistance and now it was showing 13 ohms instead of 2 ohms. Tried applying 5V again to the rail but still a black screen and voltage now down to around 3V. However this time I noticed that the Z80 CPU was getting hot and pretty quickly. So quickly removed power and also unsoldered it cleanly and bingo - now I get 5V on the rail and a white screen with a greyish center area. Some progress at last.

I soldered in some sockets and started trying to put back some of the memory chips. Three of them appear to return the black screen issue but the others seems to be ok with only a very slight drop in the Vcc rail voltage. I also now get an alternating black and white pattern on the display which I think means the memory is being accessed so it would appear that some of the memory is probably good but at least three of them are also stone dead. No idea what the previous owner did to this machine but seems he did a decent job on blowing up some silicon.

So ordered a new CPU, some new memory and will also recap and replace some of the power supply parts and see if I can get this machine back to life. Thanks again for you guidance - it definitely helped get me this far. Will let you know how it goes when the parts arrive.

Regards
Hans

[1024MAK]

Vertical black and white stripes are typical of party working RAM in the ‘bank’ that provides the ‘video’ RAM.

Mark

[bola_dor]

I don't have it at hand right now but you can search for the service manual online. Follow the power supply troubleshooting.. I remember there is a "transistor " Q1 acting as a regulator that should have 5v at its collector.
Voltages are managed completely differently than in ZX spectrum or other clones

[1024MAK]

Yeah, I linked to the service manual earlier under the text ‘schematic’.

Mark

[hamsterbasher]

Ok some strange progress. Replaced the RAM chips and now all I get is the whole screen divided up into six fairly large black and white alternating horizontal bars. The picture is stable and unmoving, actually looks quite good for composite but even with the CPU removed all I get is this screen. The replacement Z80 arrived and tried that also but whereas before the CPU would cause the machine to pull 5V low and not to show anything now all I get are the horizontal bars.

I put an oscilloscope probe on a few of the the address, data and M1 lines on the CPU and there is definitely activity there so it does look like the CPU is trying to do something but all I get is that screen. Anyone ever see something similar on a spectrum?

Regards
Hans

[1024MAK]

Not horizontal bars, no, that’s not something that I’ve seen.

Having said that, the screen data for the screen is not a simple arrangement. The screen on a ZX Spectrum is divided into the ‘display file’ (pixel data) and the attribute data. The pixel data is sub-divided into different sections. Here’s an extract from the user manual:

The display file stores the television picture. It is rather curiously laid out, so you probably won't want to PEEK or POKE in it. Each character position on the screen has an 8x8 square of dots, and each dot can be either 0 (paper) or 1 (ink) and by using binary notation we can store the pattern as 8 bytes, one for each row. However, these 8 bytes are not stored together. The corresponding rows in the 32 characters of a single line are stored together as a scan of 32 bytes, because this is what the electron beam in the television needs as it scans from the left hand side of the screen to the other. Since the complete picture has 24 lines of 8 scans each, you might expect the total of 172 scans to be stored in order, one after the other; you'd be wrong. First come the top scans of lines 0 to 7, then the next scans of lines 0 to 7, and so on to the bottom scans of lines 0 to 7; then the same for lines 8 to 15; and then the same for lines 16 to 23.

The TS2068 expands on this to offer additional screen modes by using extra RAM. But I can’t describe how it does it, as I don’t have enough knowledge on this subject.

Anyway, if you have horizontal bars, my first thought is there may be a problem with the address lines. But before then, does pressing a key make the key click sound? If it’s similar to a Speccy, that would mean that BASIC is running, and you just have a display system fault. If not, it would mean that the Z80A is not correctly running the ROM code. As the display is the same with or without the Z80A fitted, the likelihood is that it’s the latter.

So back to the address lines. Best to have a printout/hard copy/photo copy of the schematic that you write / drawn on, so you can mark each address line as you test it.

You can start by testing each address line (A0 to A15) on the Z80A to see if the line is showing activity (at the very least A0 to A6 should be counting up each time /RFSH is low) and goes to valid logic low and high states.

Then with the power off, use the resistance range to test each address line to +5V, 0V, and to any/all nearby PCB tracks. Basically you are looking for any short circuits.

Another possibility is that the ROM is faulty.

Mark