Amiga 4000 - 3.3V on address lines ?

[mortenkjerulff]

Hi Forum.

I am trying to fix an unstable A4000.
It has had a recap.

I am only seeing 3.3V on the address lines.

I would have thought that there should be 5V, but is this normal with a A3660 cpu card installed?

Without the cpu card installed the signal lines are at 4.90V.

The 5V lines read 4.90V across the PCB.

The computer will boot with Diagrom installed, and can perform a memory test with chipmem without errors.
But only when "cold". If I continue with memory tests, it will eventually start to display a lot of errors and in the end freeze and require some time before it is able to boot again.


Best regards
Morten

[jbenam]

Hello there,
is the PSU you are testing with a known good PSU? Was it recapped?

Do you have a normal A3640 to test with? (I guess not)

If you get almost 5V without the card, I fear there might be something wrong with your A3660. It almost seems like it's feeding back 3.3V on the 5V line. Is the A3660 tested and known working?

[mortenkjerulff]

Thanks for weighing in :-)

I have been testing with both the original PSU and a new ATX, with same result.

I bought the A3660 card brand new, So I guess it has been tested before shipment.
I have the old A3640. It has a broken pin on the connector and minor damage from leaking caps, but a new connector is underway, so I can try and test it.

No it does not seem like a coincidence that I measure 3.3V on the address lines when the 68060 runs on 3.3V...

[jlin_au]

To quote the M68060 User’s Manual in section 11.2.1 "Power Considerations" (https://www.nxp.com/docs/en/data-sheet/MC68060UM.pdf):

"The MC68060 operates at a supply voltage of 3.3 V, not 5 V. The MC68060 interfaces gluelessly to transistor-transistor logic (TTL) levels."

Note - Acceptable TTL input levels are between 0V and 0.8V for a logic 0 and between 2.0V and 5V for a logic 1.

So your 3.3V at logic 1 on the system address lines when your using the A3660 card is acceptable. The 68060 is a 3.3V device so can only produce logic 1 at up to 3.3V. The 68060 CPU is TTL level tolerant on its inputs pins.

[jbenam]

Huh, my bad - I was from mobile and I didn't see "address lines". I thought these voltages were tested on a +5V point.

[mortenkjerulff]

Thanks a lot !

Glad to know that at least the cpu card is not my problem :-)

[jbenam]

Have you checked if any ICs become abnormally hot? You said it starts erroring out after a while and it needs to cool down to start again.

Looks like something that is exacerbated by normal thermal buildup. I daresay it's either a failing IC or a micro-fractured track/via.

A high pressure fan might help with troubleshooting. You can try keeping cold different parts of the motherboard until you find out approximately where the issue lies.

[mortenkjerulff]

Alice is getting more hot, than the other ICs. Maybe 50-60 degrees celsius.

I tried to reflow the legs of Alice which seems to make the computer run a little longer before it halts (about 30 minutes instead of 5-10).
But after this reflow the bottom half of the screen output is a mess.

So I guess something is up with Alice, which probably makes sense, since the address lines of the chip memory is also connected to this IC.

I hope it is a bad solder joint, and will retry to reflow tomorrow as it is getting late here now.

[slaapliedje]

I've been having a hell of a time getting my CSPPC to be stable, I probably should start testing voltages at this point (though I did try it out in my other A4k, and it was just as unstable... though that one hasn't been recapped, and the normal one has.)

[jasonsbeer]

Not having looked at schematics...just going on what is being presented here.

Without the cpu card installed the signal lines are at 4.90V.

There are pull up resistors somewhere, otherwise these would read 0V. 4.9V is what you should read when the address lines go high.

How are you measuring the voltage? If you are using a scope and seeing 3.3V peak voltage, yeah, that is low. If you are using a multimeter, you are seeing the average voltage of an oscillating signal. In that case, 3.3V is within reason.

But only when "cold". If I continue with memory tests, it will eventually start to display a lot of errors and in the end freeze and require some time before it is able to boot again.

I would continue looking for bad solder joints as the joint may loose contact when the parts expand...could also be a chip issue. After issues present themselves, try putting some pressure on various chips at power on. See if you get any response. On a side note, you can test for heat related IC failure by putting an ice cube in a plastic bag and holding it on a chip.

Edit: Was the computer running better or worse after the recap?