Pettester, which bits are bad in zero page?

[Hutch]

I'm working on an 8032 with bad system ram in both lower and upper banks.
Pettester v4 is passing the video ram test and and moving on to 0/1 page.
I know the lower bank is bad but I'm trying to figure out which bits/chips are bad, so I don't have to desolder all eight of them.

I'm reading now that it's writing 00 to 0100 and 01 to 0101 and displaying the actual byte returned, is that right?
But the values are changing at regular intervals, so I suppose it's re-writing different values?

Freeze framing the video, the first frame shows a g at 02, 10, 11, 18, 19
I thought maybe bit 1 is stuck on or bit 0 stuck off, but the pattern doesn't follow.

It's showing an f (0110) for byte 0 which should be 00 and a g (0111) for byte 1 which should be 01
But also shows . or G for byte 2 which is 02 (0010)

What's happening as the values are changing?

Thanks Dave for PetTester.

 

[Hutch]

OK. I found a pattern.
Bit 0 is alternating between 0 and 1 every other address. I don't think I've seen that mode before.
Bit 1 is stuck on always
and Bit 2 is sometimes on when it should be off and off when it should be on.
I'm assuming the test is repeating and sometimes reading back different values on bit 2, causing the display to change.

00 0000 = 0110
01 0001 = 0111
02 0010 = 0010
03 0011 = 0111
04 0100 = 0110
05 0101 = 0011
06 0110 = 0010
07 0111 = 0011
08 1000 = 1110
09 1001 = 1011
0A 1010 = 1010
0B 1011 = 1011
.....
AA 10101010 = 1010 1010
AB 10101011 = 1010 1011
AC 10101100 = 1010 1010
AD 10101101 = 1010 1111
AE 10101110 = 1010 1010
AF 10101111 = 1010 1011
B0 10110000 = 1011 0010
B1 10110001 = 1011 0011
B2 10110010 = 1011 0110
B3 10110011 = 1011 0011