Keyboard problem ZX Spectrum plus

[thanospl]

Hello to all friends here, my name is Thanos
First post here
I got my old Spectrum+ from the warehouse back in he house to play a little but after all these years sitting there I had problem with the Keyboard
Whatever i tried I never make to have all keys working, so I order a new one from Sintech, Germany
This is my third replacement so far , the second had very little use but it seems it went bad in time ,even without using it.
Anyway I put the new membrane but still have some keys not working.
My PCB connectors are not in the best shape so I 'm going to replace them (I have to order and replace them)
I find out that the Row 2 (KB1) was not making contact because of the KB1 connector, but I also have no keys on the A15 column but there is no fault of the connector.
I remove both of the Connectors and tried to make a test with a wire connecting Rows and Columns together for testing, and still I have no keys responding on A15 column (B, N, M, Symbol Shift, Space Not working)
I checked connectivity between the PCB lines of the connector and the Diode (D8 i think) and the A15 line to CPU , ULA and seems Ok,.

Do you guys have an idea what to look next
Thanks in advance

[Ast A. Moore]

It’s either D8 (Pin 8 of KB2), or, unfortunately, the ULA itself. There’s a slim chance that it’s an intermittent fault of the CPU; you’ll need to run an extensive RAM test to be sure.

Check D8 and make sure there are no cold solder joints anywhere on the A15 line.

[thanospl]

Hi and thanks for your reply

D8 seems ok (measurement on the PCB)
What I may forgot to say is that Spectrum Boots fine on the white Sinclair screen, whatever that means for the CPU and ULA health.

[thanospl]

How to run a RAM test, no enough keyboard keys to load any program

[Ast A. Moore]

How to run a RAM test, no enough keyboard keys to load any program

Yeah, that’s a tricky one. How long does the power-on test take? You can compare it to an emulator switching between a 16K and 48K models. The black paper lingers on the screen a little longer when you reset the machine on a 48K Spectrum than it does on a 16K one.

There are some hardware diagnostic solutions, but it’s probably cheaper to buy another Spectrum than one of those.

Is the ULA socketed or soldered onto the logic board? If it is socketed, I’d take it out and inspect the socket. Then give everything a good clean and reseat the ULA.

There’s a chance that the 74LS00 NAND gate is faulty (IC24). If it’s holding A15 low all the time, the ULA won’t detect any keyboard input on it. You could try replacing it, see if it clears the fault. That’s a jellybean part readily available from many sources.

Do you have an oscilloscope, by the way? It would be great if you could see if there’s any activity on A15 or if it’s permanently stuck low.

Oh, and what’s the logic board’s revision? What issue is it?

[thanospl]

How to run a RAM test, no enough keyboard keys to load any program

Yeah, that’s a tricky one. How long does the power-on test take? You can compare it to an emulator switching between a 16K and 48K models. The black paper lingers on the screen a little longer when you reset the machine on a 48K Spectrum than it does on a 16K one.

There are some hardware diagnostic solutions, but it’s probably cheaper to buy another Spectrum than one of those.

Is the ULA socketed or soldered onto the logic board? If it is socketed, I’d take it out and inspect the socket. Then give everything a good clean and reseat the ULA.

There’s a chance that the 74LS00 NAND gate is faulty (IC24). If it’s holding A15 low all the time, the ULA won’t detect any keyboard input on it. You could try replacing it, see if it clears the fault. That’s a jellybean part readily available from many sources.

Do you have an oscilloscope, by the way? It would be great if you could see if there’s any activity on A15 or if it’s permanently stuck low.

Oh, and what’s the logic board’s revision? What issue is it?

Thanks Moore for your suggestion

Board is 4B, 1983
It's true that the A15 was always low, I did use my Owon MSO to verify that.
No only was stuck at low , it was actually shorted to ground!!
ULA and Z80 are socketed so I start by removing ULA first to see if the short is gone. NO.
Then I did remove the Z80 and voila, A15 was not shorted, So the CPU did the short for some reason
To be honest I was surprised. I thought the Spectrum does a memory check at the boot, but i think it checks only the lower 16K of RAM since it does not use the A15 during that test.
So I'm looking for a new CPU hopefully at local market, any suggestions in CPU revision, brand etc?

Thanks again
Thanos

[1024MAK]

The ROM code does do a check. But it is a relatively simple test, designed to discover how much RAM is present. It is not designed to test for memory faults as such.

New CMOS Z80 CPUs are still available from the larger electronic suppliers.

Mark

[1024MAK]

@thanospl

Whereabouts in the world are you?

You need a Z80 rated at 4MHz. 6MHz, 8MHz and 10MHz types are also okay. The original that was fitted by Sinclair was a NMOS type. The more modern CMOS types work as well.

For a new CMOS Z80, the part number is Z84C0008PEG.
In the U.K. Farnell stock them. See here.
RS also list them, see here.
Both of these companies also trade in other countries.
Various other large electronic suppliers should also be able to supply new chips.

Or you can take a risk with the likes of eBay and other similar sites. Where both the older NMOS and CMOS types are often advertised. Beware however that some of these may be second hand / used ‘pulls’ removed from old equipment. Some as as marked. But some have been remarked with a higher specification number / code. So a 4MHz rated NMOS chip May be marked as a 10MHz CMOS type.

Mark

[Ast A. Moore]

I was surprised. I thought the Spectrum does a memory check at the boot, but i think it checks only the lower 16K of RAM since it does not use the A15 during that test.

As Mark said, it’s a very simple test, and if it detects fault in the upper RAM, it simply marks it as non-existent. The Spectrum will then operate as a 16K model. Since it was an address line that was stuck, and not a data line, the test didn’t “fail,” as far as the Speccy was concerned; it just didn’t do right in the upper RAM.

When you get a new CPU, you can swap the two in and out. You’ll then likely notice that after you reset the machine, the test takes shorter with the faulty CPU than it does with a fully working one.

[thanospl]

thanks Mark and others for your suggestions
So far I 'm replacing the CPU and see.
Sure I can easily find a Z80A on the net , but it takes at least a week to 10days for them to send it, so I was looking for the local Athens, Greece market
to find one.
What do you guys think for this Z80A clone I found? Do you think it will work fine on the Spectrum?
https://www.ebay.com/itm/80ACPU-Z80ACPU ... SwuzRXfP1S

let me know
Thanos

[1024MAK]

I’ve never see or tried this clone, so hard to say other than give it a try...

See also this site and here, also see the Wikipedia page on the U880.

Mark

[thanospl]

The Eastern Germany Z80A clone seems to work fine
Yes, boot to Sinclair logo takes a little longer, I need to refresh my Z80 assembly knowledge to find out how the RAM-CHECK Routine works and determine the top of the RAM
Not sure but I think it runs a little hotter than the ST i had but not sure.
I have no keyboard yet, I'm waiting for the keyboard PCB connectors to arrive, but I manage to run a testram app by shorting dedicated Columns and Rows on the PCB for the appropriate keys to run the app.
It passes the Ram test fine.

Thanks everyone for their help.
Thanos

[Ast A. Moore]

The Eastern Germany Z80A clone seems to work fine
Yes, boot to Sinclair logo takes a little longer, I need to refresh my Z80 assembly knowledge to find out how the RAM-CHECK Routine works and determine the top of the RAM
Not sure but I think it runs a little hotter than the ST i had but not sure.
I have no keyboard yet, I'm waiting for the keyboard PCB connectors to arrive, but I manage to run a testram app by shorting dedicated Columns and Rows on the PCB for the appropriate keys to run the app.
It passes the Ram test fine.

Thanks everyone for their help.
Thanos

Glad you were able to sort it all out. Cheers!

[1024MAK]

Good progress

Mark

[thanospl]

Unfortunately that progress was for a little while.
I don't have a real keyboard yet and I just making contact with wires between the two connectors to have a key-press
I probably make a wrong contact somehow and now I don't have any key on the row 2 (KB1 of the ULA)
Any press on that Row making either nothing, or reset. or strange patterns on the screen (photo posted)
I 'm afraid that i fried the ULA, what do you think guys?

Thanos

[Ast A. Moore]

Possibly, but first, put the old CPU back in and retest.

[thanospl]

I did. Does the same think (besides A14 is shorted to ground)