Faulty Spectrum 48k 3b

[payty]

Hi,

I am trying to fix an issue 3b ZX Spectrum.

On power on it displays the Sinclair Logo.Keyboard working, sound, etc.

There is a problem with the upper memory.

I've bought the computer from somebody and I do not know if it has been modified or not. It has sockets on the chips that drive the upper memory.
Ram chips are OKI, IC25 and IC26 are 74LS158 instead of 74LS157. But there does not seem to be any soldering attempts on the underside of the PCB.

It cannot use all the upper memory, print peek 23733 returns 127.

I have changed IC 23, IC 24 with same chips as original and IC25 and IC26 for 74LS157. Also changed all ram chips with 4164.
Now print peek 23733 returns 191.

The capacitors have been changed, ULA swapped with another one. Nothing happened, same result.

Any advice? What should I test in order to make it work?

The strange thing is that if I poke the last byte of memory that should be usable, and the next one, it seems to work. Print peek on the same address returns the correct value, so memory does not appear to be actually faulty. I've tried several values ( including 10101010 and 01010101).

[Firefox]

What wire-link jumpers are fitted? (Back edge of the PCB between the edge connector and the cassette jacks.)

[payty]

Well, it looks like there are no wires for chip type selector.

I think that this computer was sold as a 16k model and somebody has tried to upgrade it without those wire links.


I have installed the wires and now it's working properly.

Thanks a lot!

[Firefox]

Wow, that was a lucky guess!

I remember seeing a similar fault in one of JoulesPerCoulomb's repair videos, though in that case I think the jumperless floating signal was causing it to randomly switch between Oki and TI memory modes, causing weird errors with written values not being the same when read back, or something.

Glad your machine's working properly now, anyway!

[1024MAK]

When the ROM does a RAM check, it repeatedly accesses the same area of RAM (the ‘upper’ RAM), so any unconnected inputs to the multiplexer chips will pick up noise. Whereas when you get BASIC to write or read from the upper RAM, it’s only one access with a relatively large period of time (as far as the electronics is concerned) between accesses.

Mark

[payty]

This computer is working well so far. But I have tried to change some of the upper ram chips in order to test some 4164 that I have bought on ebay.

Original chips are OKI 3732L
The chips I am trying to check are Mitsubishi and are labeled M5k4164NP - 20.

I have a test ROM module with the ROM from Retroleum. The thing is that most of the 4164 chips appear to be faulty, with the computer either not starting at all or the test ROM finding unstable values returned by the ram.

What could be the problem in this case? The chips are faulty, or there is something else?
For this test I have kept 4 OKI chips and the other 4 are 4164.



The computer is working well with all the original OKI chips in place, no errors no matter how much I test it.

[1024MAK]

The M5k4164NP - 20 may be too slow.

DRAM chips rated 150ns or faster are recommended.

Mark