520STFM: After replacing memory: White screen of death.

[stephen_usher]

Problem:
I desoldered the existing memory, added the three resistors (and also replaced the reset switch and added the blitter socket). I then added the Exxos 4MB RAM board.

After this I get the white screen of death.

Using an oscilloscope I see that the address lines are all active and look fine, the /RAS lines on the memory are running normally but there are no /CAS pulses on any of the lines, i.e. /CASL0, /CASH0, /CASL1 and /CASH1 are all high.

These lines are high at the MMU and the clocks, 4MHz, 8MHz and 16Mhz are all good.

Approximately every second ALL the data lines pulse, as shown here:

IMG_2027.jpg (225.47 KiB) Viewed 4024 times

Other than that there's no activity on the data lines. This is the same with or without the RAM fitted. This disappears if the reset switch is kept pressed in.

I don't have a pinout for the Blitter socket but I doubt the /CAS lines go anywhere near there so I don't think that this can be the problem.

Could I have killed the MMU somehow?

I'm at a loss now and need someone with a more in depth knowledge of the MMU behaviour. I've read somewhere that if it doesn't detect RAM it turns off the /CAS lines for that bank, but I don't see any activity on any /CAS line at all, ever.

Basically, HELP!

[exxos]

Why not just change the MMU and see what happens rather than trying to diagnose the thing ? I would suggest you use the diagnostic cartridge and see if that boots and outputs serial. Then at least you know TOS is running.

[stephen_usher]

Why not just change the MMU and see what happens rather than trying to diagnose the thing ? I would suggest you use the diagnostic cartridge and see if that boots and outputs serial. Then at least you know TOS is running.

I've had the diag cart in there all the time and there's nothing on the serial port.

I can see from the data lines it's not actually getting very far at all until it crashes and starts again. (Removing the cart makes no difference to the behaviour.)

Just in case it's the ROMs (very unlikely) I've just put your dualtos card in there. There's a minor change to the shape of the data line in that there seems to be an extra sharp excursion to +5V, but otherwise no change.

As for swapping the MMU, I could buy a new one from you but that takes time and it could be something else I can find in the meantime. I can think about the purchase over the weekend.

[exxos]

Have a look here for ideas ... https://www.exxosforum.co.uk/forum/viewt ... =78&t=2362

[stephen_usher]

I think I have a plan of action for today.

Firstly I tried an old MMU which I know is faulty as it was swapped out of my STM in 1989 by 3rd Coast Technologies after I couldn't get the memory upgrade to work. That showed precisely the same symptoms. Maybe Me and MMU /CAS circuitry just don't mix.

So, the next phase will be bringing down my other STFM and borrowing the MMU from that, hopefully not breaking that in the process.

[tzok]

I desoldered the existing memory, added the three resistors (and also replaced the reset switch and added the blitter socket).

That's why I always recommend doing one step at a time, and check if the computer still works.

So, the next phase will be bringing down my other STFM and borrowing the MMU from that, hopefully not breaking that in the process.

I'd rather put supposedly bad MMU in a known, working board, than the other way around.

Have you probed /RESET (pin 18) and /BERR (pin 22) signals on the CPU?

[stephen_usher]

I desoldered the existing memory, added the three resistors (and also replaced the reset switch and added the blitter socket).

That's why I always recommend doing one step at a time, and check if the computer still works.

I know, I know. I'm generally the same. That's why I tested the system endlessly after removing the Marpet. The only incremental thing I could have done though on that side of the board would have been fit the three resistors. (The reset switch is nowhere near this location.) The next step, which made the machine non-operational, was resodlering all the RAM and removing the solder from all the other RAM locations.

Fitting the socket "shouldn't" make any difference, though the Blitter does have access to the bus. Having said that, I tested everything I could on that side of the board after soldering it.

So, the next phase will be bringing down my other STFM and borrowing the MMU from that, hopefully not breaking that in the process.

I'd rather put supposedly bad MMU in a known, working board, than the other way around.

True. Good idea. I'll try that first.

Have you probed /RESET (pin 18) and /BERR (pin 22) signals on the CPU?

I've not done the /BERR line, I must admit. Given that there's a lot of address line activity the CPU is unlikely to be halted.

[tzok]

Have you thoroughly probed the MMU and Shifter sockets for continuity between its contacts and the board? Removing and inserting the "snap-on" MMU adapter could easily damage the MMU socket. It is important to verify the continuity between the MMU pins, and the back side of the board (not the socket contacts from the top). In one board I also had a white screen after installing a similar memory expansion, yet I haven't even probed the bus, because in my case it was enough to solder back one pin to re-enable the on-board RAM, and this quickly led me to the diagnosis, that the culprit of the failure was in a bad contact in the Shifter adapter-board.

[stephen_usher]

OK, it's not the MMU, that works in the other STFM.

/BERR is mostly low, except for one pulse approximately every second.

/RESET is cycling every half a second.

The memory expansion I'm fitting is the soldered in one which sits in the RAM area, mostly the upper 512K bank but with a couple of connections to the lower one (data).

[tzok]

If I reckon correctly /BERR is active low, so if you have it "mostly low", you constantly have the Bus Error. Does It pulse high only for a short while after the /RESET goes from low to high (/RESET is also active low)?