This is an old post, preserved for reference.
The products and services mentioned within are no longer available.
As part of the multiway programmer for the DivMMC Future programmer for The Future Was 8 Bit, we needed a number of Spectrum boards.
OverviewBefore I start, here is an example of a complete, working +2, time for a quick tour. First off, these are referred to as 'Greys' because they came in a grey case. Not be to confused with the later +2A and +2B which came in a black case and had very different boards inside.
more infomtion here, including an updated version with some bugfixes).
Here are a selection of boards that should be able to be patched up to do the job. I numbered them as I took them out of the box, to keep track of things.
Brendan Alford 's ZX Diag V1.28. This is good at testing the lower 16K RAM, and does well to indicate the problem even with the faulty display RAM. It doesn't test the full 128K RAM unless you go through the menu, and I don't have a keyboard connected at this stage.
Paul Farrow's 128K RAM tester. This retests the lower RAM and also tests all the banks of RAM on the 128K.
tape version of ZX Diagnostics (loaded from SD card), which will test ROM and system RAM (and other things if you hold down keys - see the link above for full instructions). That ran a soak test for a while and didn't detect any faults.
ConclusionThe results of this is 34 faulty RAM chips (and plus another dozen still on board 5) and one bad ROM chip. History does not recall whether the parts originally missing were removed as they were faulty awaiting replacements or more likely removed working as the rest of the board was deemed faulty. These boards all date to mid 1986, and all the failed RAM chips are Korean made Samsung KM4164B. Is this just a bad batch that start to fail after only thirty years?