FIXED Spectrum +2 grey RAM fault but RAM is good

[Rabs]

Yes, this is a tricky fault. You said earlier that you had changed the Z80. Did you try the removed chip in another Spectrum?

Mark

No just swapped for a CMOS Z80 I had lying around, problem persisted then swapped back.

Obviously this looks like a RAM fault but I tested the RAM IC and swapped for a known working. The stuck key warning is also an indicator since the bit is stuck low.

The PCB is in good shape and nobody else has attempted a repair, so I am assuming it is all good track wise.

Hmm!

Basically I think it will be by a process of elimination but I think something is dragging D1 low.

[Rabs]

But there is no screen corruption when the Diga Runs. So the Z80 is reading the ROM ok and writing to the RAM and the ULA is reading it OK.

[Rabs]

Is there, at any point any screen corruption?

RAM in positions IC25 to IC32 provides the RAM for the screen display. If IC31 / bit 1 was faulty, or there was an open circuit fault in the D1 data line, you will get screen corruption.

So no screen corruption.

So is it possible that the faulty RAM IC is actually in the other 64k but shows up as a fault on IC31? Like somehow D1 in the other 64k is holding the line.

[1024MAK]

Unlikely given that resistors are used to divide the data bus and the two groups of DRAM are controlled differently.

I’m wondering if the ULA is receiving all the control signals from the Z80.

Mark

[Rabs]

I’m wondering if the ULA is receiving all the control signals from the Z80.

Mark

I have checked MREQ, RD and WR at the ULA and they all look stable.

What is worse, I have seen the lower RAM test pass and on one occasion I have seen it flag IC28 bit 4.

Unfortunately I am wondering if the ULA has a fault.

[1024MAK]

If the fault is or has been intermittent, poor IC socket contacts, dry jointed solder joints, and PCB track faults are still possibilities.

Mark

[Rabs]

If the fault is or has been intermittent, poor IC socket contacts, dry jointed solder joints, and PCB track faults are still possibilities.

Mark

Indeed, that is what I am afraid of. But the board is clean and in good visual condition and no previous work has been completed.

As a precaution I reflowed and cleaned the Z80 and ULA sockets, which are both of the bouble leaf variety and no change.

I would 98% of the time the RAM test just reports bit 1. About three times it has also listed bit 4 and twice it passed (you can imagine I fell off my chair when that happened )

Think I will summarise and then pause this while I think about what is next.

Thanks for the help by the way. As usual you are very helpful.

Rob

[Rabs]

In summary.

ZX Spectrum +2 (grey) displays typical RAM failure pattern on power on.

Diagnostics ROM indicates IC31 bit 1 faulty.

Removed IC31 and tested off board, and passed. RAM is good.

Socketed IC31 and replaced with a new 4164 but faster speed.

Problem still persists.

Other actions taken but still not fixed.

1. Removed, socketed and replaced Amstrad 40058 Multiplexer with ZX8401
2. Removed, ROM
3. Removed, socketed and replaced IC18 bit 1 RAM on second bank
4. Swapped Z80 out
5. Cleaned and reflowed sockets for Z80 and ULA
6. Checked continuity for RAM and Z80 control signals. No breaks found.
7. Scoped data and Z80 control lines, no obvious faults

Other symptoms
1. Diagnostics ROM reports stuck key which sometimes clears
2. There is no screen corruption when running the Diagnostics ROM
3. Occasionally the RAM test also reports that bit 4 has failed and has once or twice passed.

PCB is in good visual condition and no previous work has been undertaken.

Other things left to try (in no particular order just a desperate list of things I can think of)

1. Replace ULA.
2. Remove AY
3. Remove Joystick controller
4. Replace all RAM

Right
Beer

[Rabs]

Bit of progress.

The clue was a stuck key on bit 1 but also bit 4.

Both of these bits from part of the joystick interface.

So I removed the joystick controller IC and all RAM tests now pass and no stuck keys.

Not booting yet but guess that is because I need a joystick controller.

[1024MAK]

The joystick chip is believed to be the same as the chip in the Sinclair Interface 2. Hence it's not believed to be essential. The other thing of note, is that the data bus outputs can only drive each data line low (like open collector/open drain).

Well done on making progress

Mark