exxos's Atari, Amiga & retro forum

exxos wrote: 12 May 2023 11:57 Can you post a image of your motherboard overall..

All now with new chips installed :





The extra cables you see on the motherboard is the Marpet (SIMM) memory upgrade, back in the day I actually sent it to them for them to install and had 2.5MB RAM working. For this repair purposes I have detached the SIMM part from the machine (so 512K still active) and RAM test from Diag cartridge passes fine so would assume RAM/MMU all good here.

PhilC wrote: 12 May 2023 11:57 @jwd I've had some odd faults happen when a GLUE has decided to die, so might be worth swapping that and giving the socket its in a good test and clean as well.

That and MMU are the only chips not in sockets . . . :(

Unless there is something obvious like a bad solder joint or pin not correctly inserted into the ROM or CPU socket etc, then your only option is to continue to check the entire address and databus from the CPU to GLUE-MMU-ROM and see if anything is amiss. Also scrutinise all solder joints on the motherboard as well. I had one ones where I had to re-solder literally half of the board to get a machine working.

The Marpet cabling looks like a bit of a rats nest, have you checked for shorts around each location where a wire is soldered? If the wire starts to fail, there can be near invisible whiskers causing shorts.

Also, I'd suggest you hook up the ST to a PC with a null modem cable and fire up Teraterm, see what shows up on the terminal when you boot with the diag cart, sometimes you'll see something that isn't obvious with the diag cart.

Given it boots to a point, and also with the diag cart, that suggests the ROM access is probably OK.

rubber_jonnie wrote: 12 May 2023 13:30
Also, I'd suggest you hook up the ST to a PC with a null modem cable and fire up Teraterm, see what shows up on the terminal when you boot with the diag cart, sometimes you'll see something that isn't obvious with the diag cart.

Interestingly on bootup my ST screen has the known T6 1772 Bus Test fail but on the terminal it says something slightly different :

I6 Testing Bus Error
No (I6) error

Thing is there is no reference to I6 in the service manual as far as i can see (it only goes up to I5). Any clues from this error?

jwd wrote: 12 May 2023 15:07

rubber_jonnie wrote: 12 May 2023 13:30
Also, I'd suggest you hook up the ST to a PC with a null modem cable and fire up Teraterm, see what shows up on the terminal when you boot with the diag cart, sometimes you'll see something that isn't obvious with the diag cart.

Interestingly on bootup my ST screen has the known T6 1772 Bus Test fail but on the terminal it says something slightly different :

I6 Testing Bus Error
No (I6) error

Thing is there is no reference to I6 in the service manual as far as i can see (it only goes up to I5). Any clues from this error?

Not one I've seen before, let me think on it.

I6 error I have seen on my boosters which decode ROM, because it it decodes address zero, which is used for the bus error test. My decoders don't support that "feature" so the diagnostic comes up "I6 bus error handling error" Or something like that. Ordinarily in that case it is nothing to worry about.

However in your case you should not be seeing that malfunction. Does bus error actually switch on the CPU ?

As to why you get different messages between the screen and console I have never seen that one...

exxos wrote: 12 May 2023 15:44 I6 error I have seen on my boosters which decode ROM, because it it decodes address zero, which is used for the bus error test. My decoders don't support that "feature" so the diagnostic comes up "I6 bus error handling error" Or something like that. Ordinarily in that case it is nothing to worry about.

However in your case you should not be seeing that malfunction. Does bus error actually switch on the CPU ?

As to why you get different messages between the screen and console I have never seen that one...

I have seen RAM errors come up during boot that don't show on the Diag screen at all, hence why I am keen to suggest people also use the serial port during testing.

I can't recall, have you looked at the /HALT & /RESET signals on the CPU when this happens?

Also, as @PhilC said, it can be the GLUE, and I've had one or two bad GLUEs in my time.

We have life!!

Thanks for all the pointers on this, you guys have the experience to just point people in the right direction.

Checking all connections on the GLUE again I noticed that the RDY (pin 28) had no continuity to the DMA (pin38) RDY (if i put my big multimeter probe on the top of the GLUE pin). When i tested this continuity last time my big fat test probe must have touched the pcb below (which has continuity to the DMA) so that must mean the PCB track is fine just the pin > PCB isn't making.

Yep this one wasn't connected to the motherboard :





Anyways some careful soldering :hide:





No more 33 bombs and passes the diag timing tests fine :yay:

What i did notice before i found this solder break using the oscilloscope on the data/address lines (CPU) is that (without diag cart) the 33 bombs might appear on the screen but the machine is still doing stuff (data lines still pulsing etc.) then after about 7 seconds it gives up and the HALT is activated and everything dies. So the RDY is an input to DMA and if the DMA is still waiting for a ready to communicate and the rest of the machine can't speak to it then 33 bombs it is.

Hope this helps someone else, is it safe to say 33 bombs (needs high res as low res didn't show them all) = GLUE > DMA unable to communicate and therefore timeout.

I have put back all the original chips and still works, alot learnt and thoroughly enjoyed the journey (most difficult fix i've done so far).

Thanks again.


p.s. after studying the board so much I noticed that there had been alot of 'additional' non factory work done hence why nearly all chips have sockets and solder looks completely different on these sockets than the nice factory solder. Then it dawned on me this was a replacement as my original (late 87/early 88) had a single sided drive and the drive broke so my Dad sent it back to Dixons and they gave it back with a double sided drive, NO THEY DIDN'T, they gave me a bloody refurbed machine that Dixons had repaired with most of the chips replaced (including CPU!) and terrible soldering around the SMT GLUE so no wonder it became disconnected. This is something i had forgotten so again another positive from doing this repair - get my memory back!

Great to see you got it fixed @jwd but some of those other GLUE pins look like they need a bit of work too?