Another busted Zeddie

[JonB]

Hi all, new member, first post here. Some of you already know me from other forums - StarDot, VCF and the like, where I use the same handle.

Anyway, on to my ZX81, which is showing a white screen and no K

It's been on the shelf for a while. When put away only worked when the 16K RAM pack was fitted. Some of my reading indicated that this is a symptom of the internal 2114 SRAMs failing, so that's on the list. Now it is a white screen whether or not the RAM is fitted.

What I decided to do is get it going stand-alone first; that is, without the ram pack. I swapped out the 2114s for known good ones (tested on the Superbrain) but this hasn't resulted in a fix. Then I reseated the ROM chip and checked the CPU clock line with a scope, all good though it does look a wee bit noisy. Next up was to connect the Zicon logic analyser to the CPU and see what it's up to. It's likely you don't know what this is - it's not a logic analyser in the normal sense, it is an ancient test device that you connect to a CPU using a huge clip, and it monitors the address, data and control lines, recording the state changes. Handy if the Z80 is directly soldered to a board (otherwise I'd use the Z80-ICE). Anyway... I set recording to trigger at 0000h (address bus) and powered up the Z. I can see that it's seeing FF on the datalines constantly and that means the CPU will be doing RST 38 constantly, if it is indeed working.

So, it could be the ROM (I consider this rather unlikely) or perhaps there is some address decoding in the ULA that isn't working.

I suppose the ROM can be tested in a reader (after desoldering the patch wires and bending the pins back down. Is there a dump of it anywhere?

As to the ULA, well I'm a bit stumped there. Does it handle address decoding for the ROM? Or is this a red herring?

[1024MAK]

Hi Jon. Welcome

Yes, often a white screen is a RAM fault. But it could also be a ROM fault or a Z80A fault. Or sometimes a ULA fault. All it tells us is that the ULA is outputting the white level voltage on its video out pin. The video signal is produced on pin 16 of the ULA.

I have to ask, did you clean the edge-connector before using your RAM pack?

If you have a clock signal on pin 6 of the Z80A, then it’s a good sign that the ULA is at least partly working.

The ULA has ALL of the glue logic including the RAM and ROM decoding. There is a schematic in this post.

The chip select for the ROM is on ULA pin 13.

Yes, there are dumps of the ROM. In fact, there are three known ROM versions. Failure of a mask ROM is not very common, but occasionally does happen. Or do you have a PROM or EPROM?

Mark

[JonB]

Hello Mark

It is a 64k EPROM (ceramic / gold, very pretty) on an Issue 3 board with a number of bent pins and patch wires attached; presumably so that it presents as an 8k ROM. It has "64k ZX81" printed above the TI logo and RT18301 part number. Oh, and "SINGAPORE" but I doubt that is pertinent! There is a blob of black paint, rather than a sticker, over the UV window. There is a picture of an issue 3 board on the Wikipedia ZX81 page, this shows a mask ROM with no patch leads that doesn't occupy all the pins of the PCB socket footprint. Good old Uncle Clive.

Cleaning the contacts? No I didn't, at least not at first, but they did look OK. I know, rookie mistake! But since then I have been running without the RAM pack. Real programmers use 1K... RAM packs are for quiche eaters.

So, I could desolder the CPU and fit a socket. Or, I could attempt to read the ROM (will need the patch wires desoldering and legs straightened). But I don't recall ever seeing a Z80 or EPROM get damaged, they seem to be very robust. Can the same be said of the ULA? I am thinking yes, as (correct me if I'm wrong) it is just a gate array programmed by popping fusible links, no?

Cheers
JonB

[1024MAK]

A ULA is not programmed. It is manufactured. With the final metal interconnection layer being determined by each customer’s requirements. They are more reliable than people think.

Z80A processors do occasionally fail, but are normally fairly reliable.

Very old EPROMs are known to loose their programming, with bits reading as ones instead of zeros, so called bit rot.

Mask ROM chips are very reliable.

Mark

[JonB]

Re: ULA, interesting and fair do's, I stand corrected (my default stance as you know full well! ). Glad it is likely reliable. As to the bit rot... Superbrains are well known for this. I'd forgotten about that.

So I guess next step will be to see if the ROM has anything in it. Out with the programmer!

[BarryN]

Re: ULA, interesting and fair do's, I stand corrected (my default stance as you know full well! ). Glad it is likely reliable. As to the bit rot... Superbrains are well known for this. I'd forgotten about that.

So I guess next step will be to see if the ROM has anything in it. Out with the programmer!

ULAs can and do fail, mostly due to running to hot, or someone plugging the power adapter into the wrong plug port. Old EPROMs can also drop bits, but are usually pretty reliable and can often be erased and reprogrammed. The Z80 CPU can also fail.

[Moggy]

ULAs failing mostly due to running hot?

What evidence do you have for that? The only temp test I have ever witnessed in 40 years of its existence had it running at less than HALF (62 deg) the manufacturers recommended maximum(125 deg) and have met NO-ONE who can show a ULA that failed via excess temperature and in that same 40 years of owning and using ZX81's never had one suddenly go because it felt a tad hot, but we've been through all this before over the years and as for the EAR and MIC ports they have capacitors in their circuitry so DC is well blocked and again never known damage caused this way.

Expansion port abuse whilst powered up and failing memory packs are the main killers, Memotech ones especially and the few failures I have experienced have been down to my own clumsiness in that regard not from any heat problem, they are a more hardy chip than urban myth gives it credit for.

[1024MAK]

ULAs can and do fail, mostly due to running to hot, or someone plugging the power adapter into the wrong plug port. Old EPROMs can also drop bits, but are usually pretty reliable and can often be erased and reprogrammed. The Z80 CPU can also fail.

Erm, actually edge-connector abuse is more likely to kill a ULA or Z80A than plugging the 9V power into the ear or mic sockets (both of which have DC blocking capacitors).

Yes, ULAs do occasionally fail. But then so do Z80s, SRAM chips and especially DRAM chips. A ULA is slightly more forgiving of the power supply pin voltage going outside the normal 4.75V to 5.25V specification because it’s core circuitry is fed via on chip regulators.

As I’ve said before, ULAs are designed and specified to run hot.

Although, just like all other chips used in Sinclair computers, no one expected them to still being used 35 to 40 years after they were made!

Edit: Moggy got in first!

Mark

[Moggy]

So two of us agree at least!

Credit to Mark though he's been saying this for years.

[Moggy]

Just as a side issue and I'm sorry to say this but the most electrically fragile ULA I have come across is one of the new remakes.

Accidentally leave an address pin unplugged after putting in a ROM chip then powering up- Boom goes the dynamite.

Any peripheral with the slightest problem or not quite seated correctly ditto and so on.

I have fried three of the buggers in circumstances that the Ferranti jobby would laugh at.