Atari Mega STE CPU Halt

[Roberto]

Hi everyone,
for about two months now (whenever I find the time) I’ve been trying to solve a CPU stuck in HALT issue on a friend’s Atari Mega STE.
The machine was working fine before someone installed a socket on UA2 to mount the GAL that adds HD floppy support. After that modification, however, the CPU ended up stuck in HALT.

When I received the motherboard, the first thing I did was remove the GAL socket to check for broken traces. I didn’t find anything wrong — the work actually looked clean. I also managed to talk to the person who did the modification, and he confirmed that he cut the pins of the original GAL to install the socket, so no hot-air gun was used.

Up to now, I’ve swapped several components with the ones from my working Mega STE, specifically:

CPU
TOS
All RAM banks
PSU
TT-SCU
GST-SHIFTER
FDC

I’m still not getting any output, not even from the Diag Card (serial port).
The screen stays black (monochrome monitor) or white (color monitor).
When I press RESET, I can see that something tries to happen on screen, but it remains black afterward.

I checked the clock: it is present on the CPU, but the HALT pin (active low) is stuck at 0 V, and there’s no activity at all on the data or address pins.
I also checked the RP5C15 IC recently and noticed that its pin 3 had a square wave with double the amplitude compared to the one on my working Mega STE. Since I had a spare RP5C15, I installed a socket and swapped it. But now there’s no signal on pin 3 anymore (it stays at 0 V). The only remaining square wave is on pin 17.

At this point, I’m not sure whether I should focus on the components around the replaced GAL (even though I already tested with a known-working one), or if it’s just a coincidence that the fault appeared right after that modification.

[frank.lukas]

Check the Addressbus to GND and also to VCC and to the CPU ...

ua2_mste.jpg

[Roberto]

Check the Addressbus to GND and also to VCC and to the CPU ...

Hi Frank, nice to hear from you.
I did continuity tests with a multimeter between the address bus to UA01, UA02 and UA03 and there is continuity on all the lines.
However, I found a discrepancy regarding the resistance between CPU bus addresses to mass. All CPU addresses are at 4.8 Kohm except A13 which has 116 kΩ. On the working Mega STE board, on A13, I always have 4.8 Kohm.
I tried to replace U202 (SN74LS05N) and UA1 (SN74LS30N) but without success.

[frank.lukas]

Check the Pullups RP22,RP23,RP24 all have 4,7k to VCC ...

For a test you can put a 4,7k resistor from A13 to +5V VCC

[Roberto]

Check the Pullups RP22,RP23,RP24 all have 4,7k to VCC ...

For a test you can put a 4,7k resistor from A13 to +5V VCC

I did as you said, that is:
I checked RP22, RP23 and RP24 that have 4.7K to +5V VCC ---> YES
I put a resistance on A13 to +5V VDC but unfortunately the CPU remains in HALT.
So I tested the continuity between all the CPU addresses towards RP22, RP23, RP24 and I saw that in the working card follows this scheme:

RP22:
Pin 2 --> A17
Pin 3 --> A18
Pin 4 --> A19
Pin 5 --> A20
Pin 6 --> A21
Pin 7 --> A22
Pin 8 --> A23

RP23:
Pin 2 --> A16
Pin 3 --> A15
Pin 4 --> A14
Pin 5 --> A13
Pin 6 --> A9
Pin 7 --> A10
Pin 8 --> A12
Pin 9 --> A11

RP24:
Pin 2 --> A8
Pin 3 --> A7
Pin 4 --> A6
Pin 5 --> A5
Pin 6 --> A4
Pin 7 --> A3
Pin 8 --> A2
Pin 9 --> A1

Then I did the same thing with the card not working and I noticed that there is no continuity between A13 and pin 5 of RP23.

[Roberto]

I managed to find the track interrupted, which happened to be right under the hoof that was put by the other person. Now there is continuity between A13 and pin 5 of RP23. But it still doesn't work: black screen and CPU in Halt.

[frank.lukas]

Diagcart and a Nullmodem cable ..., any output?

Check all Datalines and Addresslines to the CPU from the Combo Chip.

[Roberto]

Diagcart and a Nullmodem cable ..., any output?

Check all Datalines and Addresslines to the CPU from the Combo Chip.

Since the work had been done right at UA2 to install a socket for the GAL, I removed the socket once again to check whether there was any broken trace on that side of the board—and that turned out to be the case. There was a trace that looked perfectly fine visually, but when tested with a multimeter it turned out to be broken.
Thank you, Frank, for the advice you gave me.

P.S.: with a few small RAM issues 🙂

Another Atari is alive today.

MegaSTE.jpeg