Does Anyone Recognize This Pattern?

[1024MAK]

Update

I can now confirm that pin 39 on the vLA module is not used (I asked the producer of the module). So the only thing a ‘scope will pick up here is noise

Mark

[TMAOne]

Yep, noise is what I had. But also at ULA 32, so the replacement ULA has been killed.

Not sure how the computer and I managed to do that, but after some generous support from Charles, I have a second vLA82 on its way through the post. Efforts to revive this computer will have to wait for the mail systems to favor me with delivery.

I've been learning a lot with this rescue attempt, and hope the time and expense will ultimately be worth it in the form of a working Spectrum. Maybe I could have bought a working one cheaply, but where's the fun in that?

Thank-you Mark for all the invaluable input. I will of course pick up this thread whenever there are new developments.

[TMAOne]

Is this progress? It certainly is different,...

Soooo, I got the 2nd vLA82 in the mail, and had no real issues replacing the socket and putting it in. But rather than the crisp black and white stripes I had (if only briefly) before, this is what I get now:

The image is blurry--that's not the camera. There are a couple of lines of small (character sized?) colored cubes in a couple of stripes amid the white boxes. Some of them blink at a cycle of about 1 second.

I wish I knew if I'm closer or farther away to getting this working,...

Any suggestions of what to look at first before I start in again?

Ian

[1024MAK]

I’ve never seen an image from a ZX Spectrum that blurry before

I trust that your 7805 has had the heatsink refitted to it?

First off, can you please check the supply voltages: +9V input, +5V, +12V, -5V.
Then, check the connections for your video cable. Is the signal line okay? Is the ground / screen okay?
Then check any section of the video circuit where any work has taken place, looking for dry joints, broken/cracked tracks etc.

The actual screen image is typical of there still being a fault whereby the Z80 CPU is not running the ROM program, possibly due to a lower RAM problem, upper RAM problem, or a problem with the Z80. It’s hard to tell, is the border white (sometimes looks grey)?

The coloured / flashing (blinking) squares are the ULA displaying the data it gets from the lower RAM.

Mark

[rcade]

Note that 4116 RAM chips will all fail nearly instantly and fatally without the proper -5v.

[TMAOne]

Sorry for the delayed response. Life tends to get in the way of working on this. And I think it's time to give it up. More on that presently.

To answer the specific questions and areas of concern first, yes either the heat sink was fitted for testing, or a clamp-on auxiliary was used and power-ups limited to about 10 seconds or less.

Supply voltages are good with the exception of the -5V line, which is still hovering around -3.7V. Lower RAM sockets had been tested for continuity or shorts; the chips themselves have now been tested in another device and are all fine.

The video cable is fine. I wasn't terribly concerned about the blurriness, as I had experimentally replaced output transistors with rough equivalents, and in the re-cap effort made some temporary substitutions of 33 uF when 22 uF was called for. (Getting parts from China is particularly difficult in these days of Covid19.) I would have addressed the blurry picture once there was something good to look at.

I do have another Z80 CPU, which I substituted, to no real effect. There was no white border in the blurry picture that I could see.

However,...

All bets are now off. The video output has disappeared again, totally. I was not doing anything at the time--it just came up dark. Hoping it was an output component, I put the original transistors (which tested good out-of-circuit) back in place. Nope, no help there. Had a look at CPU pin 6 (clock)--there is no signal.

I don't know how, but something about this board seems to be killing the ULAs.

Given that:

• The motherboard had been over-volted by the original defective 7805, causing who-knows-what secondary damage;
• The lower RAM is good, casting doubt specifically on the upper RAM which eBay tells me is expensive or impossible to replace;
• There are certainly multiple problems with this board; and, worst of all,
• Precious ULA/vLA82's are somehow assassinated by this board even though the main voltage is now benign,...

I think it's time to give this up as a bad investment of any more time and expense.

I now have another Spectrum 48K with an unknown problem, which I haven't looked at yet. I'm going to shift efforts to this 2nd one and consider the 1st to be a parts donor.

I hate to give up, but in this case it seems prudent.

[TMAOne]

A lot has happened since I last posted to this thread. I'll try to review as I go so that any interested reader need not go back to re-read the previous posts.

Mark and others had generously offered some great advice to assist me in attempting to revive an old Spectrum 48K I discovered in my possession. But my efforts had been hampered by a few factors, both known and unknown at the time.

For one thing, I have since discovered that my "best" multimeter has some sort of bug or design flaw. It will not read the voltage of the -5V circuit correctly.

No, I didn't simply have the probes the wrong way 'round. Both the meter and myself are somewhat smarter than that. It also seems to read every thing else correctly, including the 5V directly from the 7805 (in either direction), and everything else I can throw at it.

The 4116 -5V supply line was correct all the time; I just didn't know it. I'm at a loss to explain this "limited bug". The revelation was a long time coming, but has been grudgingly verified by an oscilloscope and 4 other meters. The 4116 voltages are fine. (I've sheepishly purchased a replacement multimeter.)

Also, my misadventures with the aftermarket ULA were something of a red herring. The original ULA was working, even though I've had bad luck with the vLA82's. The original is now functioning properly in the 2nd broken Spectrum 48K I mentioned in my last post.

I am happy to report that after some upper RAM replacement, this 2nd unit is now working fine!

So after a couple of months "off" from this project, my attention returns like a stubborn dog with a bone to that first Spectrum, an Issue 4B. I can't help it. I just hate broken computers.

Here is it's current "status":

• The 7805 is good.

• It has been re-capped.

• I have proven by swapping, the ROM, and the CPU.

• The ULA currently in the machine, while having flawed video, does allow a functioning machine to boot correctly. (The above photo was taken with a 100% ULA.)

• The 8 lower RAM chips have been tested good in another device. The voltages are (sigh,...) correct. Continuity and short tests, as outlined by Mark, prove out. I'm as sure as I know how to be (without actually seeing it boot), that the lower RAM is good.

Questions:

• Am I correct in assuming that if the lower RAM is good, and the upper RAM has flaws, that the machine will still boot? (That seemed to be the case with the 2nd unit. I ran memory scans of the flawed high RAM from the working low RAM in order to determine which chip to replace.)

• Failing that, how does one "test" upper RAM? Can this apparent complete lack of activity even occur as a result of RAM fault(s)?

• Any suggestions where to probe for activity, (which seems to be sadly lacking on this unit), in order to determine where the fault lies?

Thanks in advance,
Ian

[1024MAK]

I have since discovered that my "best" multimeter has some sort of bug or design flaw. It will not read the voltage of the -5V circuit correctly.

Digital multimeters having strange quirks is not unheard of. A manufacturer of expensive professional multimeters was caught out by a large former nationalised U.K. company where staff discovered that when testing on a certain item of equipment, one model of their multimeters gave different DC voltage readings depending on which way round the probes were connected to the circuit... But the older model of multimeter from the same company worked correctly.

Anyway, now you know, please label the errant meter to remind yourself in future. The +12V and the -5V supplies in a ZX Spectrum will have some high audio frequency ripple and noise on them, maybe this is upsetting the multimeter?

Questions:

• Am I correct in assuming that if the lower RAM is good, and the upper RAM has flaws, that the machine will still boot? (That seemed to be the case with the 2nd unit. I ran memory scans of the flawed high RAM from the working low RAM in order to determine which chip to replace.)

No. It’s not as simple as that (it’s a Sinclair!). Quick answer: only sometimes. Long answer: the test that the ROM code does at power on (and at reset) is only to try to work out how much RAM there is. It’s not designed to properly test the RAM. It may however decide that there is less than 48K of RAM, or that there is no ‘upper’ RAM (due to one of the ‘upper’ DRAM chips or the associated multiplexer or glue logic chips being faulty) in which case if the ‘lower’ DRAM is okay, it will start up and try to run as a 16K machine (or with whatever RAM it decided is available).

Whatever it decided, the Z80 machine stack will be placed near the top end of available RAM. If the RAM in this area is faulty (e.g. one or more of the ‘upper’ DRAM chips are faulty or the multiplexer/glue logic for these is faulty) then data (which will include return addresses) stored in this area will be corrupted. When the Z80 pulls a return address off the stack, goodness knows where it will now jump to when trying to continue executing. The result can range from a ‘frozen’ machine, to a ‘loading screen’ and countless other variations.

Now if it has decided that it only has 16K of RAM (and this RAM is good), the machine stack will be in the ‘lower’ RAM area, and hence it will now work as a 16K machine. It will execute BASIC commands correctly, run 16K games and be rock solid in operation.

• Failing that, how does one "test" upper RAM? Can this apparent complete lack of activity even occur as a result of RAM fault(s)?

Your photo does indicate that it may be a fault with one or more of the the ‘upper’ DRAM chips (or associated multiplexer or glue logic chips)...

When powered on or reset, does the screen always go black with a white border and with faint vertical red stripes in the black area? Turn up the brightness, contrast and colour and be ready, it all happens rather quickly...

Without a diagnostic ROM (which can diagnose many RAM problems), there is simple way to fool the machine into thinking it has no ‘upper’ RAM and therefore it’s only a 16K machine. That is, we temporarily disable the ‘upper’ DRAM chips. Please note this only applies to issue 2 through to issue 4S boards. Issue 5 and issue 6A boards use different circuitry.

To disable the "upper" 32k of RAM, on a issue issue 2 through to issue 4S board, take a piece of wire and connect pin 5 on IC23 (a 74LS32) to +5V. (link).

I suggest you make a temporary solder connection with an SPST on/off switch in series. Start with the switch set to on. Then if you get to the copyright screen and BASIC works, after that, you can turn the switch off so that some simple BASIC commands can access and therefore test the upper RAM.

This is done because the ROM code tries to see how much RAM there is at start-up. After this, the BASIC ROM will ignore any RAM that is not present (or in this case, is disabled)(well, unless you reset the limits using CLEAR n). But the BASIC POKE and PEEK commands have no limits and work across the whole Z80 address range.

Double and triple check that you have the correct pin on the correct chip, or you WILL damage something. This temporary modification will disable the "upper" RAM (when the switch is ON), as the CAS signal won't reach the RAM chips.

Note that if a faulty DRAM IC is jamming one of the data lines, this won't really help.

• Any suggestions where to probe for activity, (which seems to be sadly lacking on this unit), in order to determine where the fault lies?

Yes, if you have a logic probe (or are willing to make one) or have an oscilloscope.

First off, what signals do you get on the following pins of the Z80 CPU?
Be very careful, if you short the probe between adjacent pins you may damage one or more of the chips.

Pin Function
6 clock (3.5MHz square-ish clock signal)
16 /INT (should be high but pulsing low at approximately 50Hz for PAL or 60Hz for NTSC)
17 /NMI (should be high)
18 /HALT (should be high)
19 /MREQ (should be pulsing rapidly)
20 /IORQ (should be pulsing)
21 /RD (should be pulsing rapidly)
22 /WR (should be pulsing)
24 /WAIT (should be high)
25 /BUSRQ (should be high)
26 /RESET (should be high, except for a very short time after power is applied or go low then high as the reset button is pressed and released)
27 /M1 (should be pulsing)
28 /RFSH (should be pulsing)

Then test address lines A13, A14 and A15. These give an idea of which memory the CPU is accessing.

Mark

[TMAOne]

Awesome post as always Mark.

I had not heard of bugs in digital multimeters before. It caught me out completely. I replaced every component of the -5V circuit, and when that didn't work I pulled all 8 4116's thinking there must be an internal short in the chips. It was only when the bare board STILL didn't read correctly that I tried an oscilloscope to "see" what was going on. What's this? -5V exactly as ordered? Oy,...

They say to never put to sea with two compasses. Take one or take three. I had been putting all my faith in just the one without cross-checking with other tools. Never again.

I haven't had the time yet to get completely into this, but that method for temporarily cutting off the upper 32K and restoring it after the boot is complete, is exactly what the doctor ordered! I wired in the switch, and it worked--the computer booted to 16K.

Now to load in my little RAMSCAN program,... awk! The keyboard doesn't work. Tried two keyboards--it's not the keyboard itself, something on the computer is cooked. I can't access the far left or far right keys, which unfortunately include the double quotes and of course Enter, so I can't load in or run any programs.

That's as far as I got tonight. I'll have to debug that little issue before I can load a program and scan the upper memory. But I think the back is broken. If it boots on 16k, the memory march should tell me which RAM(s) to replace.

Thank-you Mark. You're a gentleman and a scholar and I owe you a pint or six.

Ian

[TMAOne]

The problem with the keyboard turned out to be the ever-more-quickly failing ULA I was using in the unit. It will alternately accept input in the "J" column (for LOAD), or in the far column (for Shift-P double quotes, and Enter), but never both on the same boot-up. Stymied.

The one good ULA I own, does not seem to work in the still-ailing computer at all. There is no video output. But it's happy enough in the working machine. (I have no explanation for this duality.)

Since I have also had some bad luck with aftermarket ULA's in the ailing machine, I think I'll leave well enough alone again, and simply celebrate the fact that I do have one unit working 100%.

Now if I could only convince it to let me win once in a long while,...