2000 green screen with kickstart, but ok with diagrom

[Castellen]

I've had this same conversation many times that DiagROM has a number of limitations around the memory tests, particularly that it doesn't detect addressing errors.  I've also had a few cases where DiagROM has showed all memory as good using the byte-by-byte (slow) memory test, though it fundamentally can't work due to an open circuit (floating) line in the data bus.

Haven't read that entire 2MB expansion thing in detail, but are you supposed to remove the original DRAMs?  If the original parts are still in circuit, and being operated in parallel to the 2MB expansion (which is a terrible idea), then defects in the original DRAMs such as addressing errors, will still cause you problems.


My suggestions would be to:
1. Remove the original 1 bit x 4 DRAMs if you're using the 2MB replacement memory.

2. The PLCC84 socket that you used on the 2MB memory board, was that a new one, or was this the one removed from the A2000 main board?  Don't re-use IC sockets; they're problematic at the best of times without recycling dodgy old ones.

3. Check all tracks and vias (again) in the corroded area, particularly the data bus around the pin 1/64 area of the CPU.  CPU data bus continuity is quick and easy to check since each line is terminated (pullup to 5V) with 1k or 3.3k on the rev6 boards, or 10k on rev4 boards.  i.e. you simply need to make sure you can see the pullup resistor between the +5V line and the data bus point you're checking.

4. Make sure that the two resistor networks near pin 1 of the CPU doing data bus pullups (I forget their designations) are all OK, sometimes battery corrosion causes parts of them to fail open or short circuit.  Though you won't get a green screen error if there's a data bus short, so less likely to be that.  Either way, replace the resistor networks if they have any visible signs of corrosion because they can cause problems in the future.

5. Check your work on the CPU socket replacement (again).  When I get these kind of repair jobs in, it's often a case of spending half an hour finding and fixing the original fault, then several more hours fixing all the additional PCB damage and bent IC legs that aren't in sockets from the previous repair attempts.

[Castellen]

I've been told you don't need to remove original dram (the adapter board is supposed to disable on board), but was also told it's possible a bad chip is causing it.

A problem you could be facing is that one or more pieces of the original DRAM is possibly defective, so even if it's "disabled" by the adapter board, it will still partly be in circuit; at least across the register data bus.  Depending what problem the old DRAM has, it could be corrupting something on the register data bus, which would of course cause a lot of problems.

About the only thing the adapter board could do to "disable" the existing memory is to disconnect the multiplexed address bus, column/row strobes, and write enable.  The register data bus will still be connected, which is shared by the new memory and half of the Amiga system.

This might not be the cause of the problems, but the easiest way to tell is to remove the old DRAMs to eliminate them as a potential cause.

If DiagROM shows the memory as working, but you always get a green screen with the standard ROM, most likely you're looking at some kind of address error.