C070523-001 not just layout silly

[exxos]

What does it do if you remove the ribbon cable ?

[Robson]

What does it do if you remove the ribbon cable ?

same pattern running

[exxos]

@Robson

This does not particularly help, but this guy has exactly the same problem as you with one of my own memory upgrades, and it is with the same silly motherboard by the looks of it.

https://www.exxosforum.co.uk/forum/viewt ... 851#p41820

The only thing I can really suggest is you try a proper benchtop power supply to power the 5 V rail, and maybe even change the MMU IC itself as I know of some in original ST's were bugged, but I never heard of bugged ones in STF/M type machines.. but its possible I guess.

[Robson]

It seems that the memory is no longer an issue (fixed few more problems) getting somewhere with the diag, but it fails again when it comes to various timers, like mfp, vsync, hsync.

I guess what I see here is corresponding to the T0, T1, T2, etc errors in the service manual.

It comes back on serial like this screen goes black after a flicker.

Have you seen this before?
I guess tracing those signals is next..

Capture2.PNG

[Robson]

from service manual

Capture3.PNG

[Robson]

Without the cart it draws me this, very slowly and starting over.

IMG_20200726_234429.jpg

IMG_20200726_234416.jpg

IMG_20200726_234427.jpg

IMG_20200726_234445.jpg

[exxos]

Still looks like ram corruption. Vertical lines is normally a bad data bit. Garbage is bad addressing. Your diagnostic cart output looks like wrong letters coming up, but would have thought that wouldn't really happen as it reads from ROM from the cart. That's more a CPU fault on the bus as it reads ROM directly. So maybe resolder the CPU next ?

[Robson]

Thanks. Actually I thought the same right after posting the photos. Cpu is socketed so try another and recheck pins in the next round. Glue socket is also something I have in mind to look at closely.

[troed]

Stable image = GLUE (sync signals), MMU (advancing memory pointer) and Shifter (reading bus and outputting pixels) are doing a lot of things right.

The garbage on screen looks like bus data, what would normally come if you display _after_ last available memory. So, maybe the MMU is unable to address any memory?

/Troed

[rubber_jonnie]

@Robson Have you looked at the data lines to the MMU from the CPU? My Mega 4 had on screen corruption, admittedly not identical to yours, but not dissimilar and my D2 & D4 lines had no connection to the MMU.

It looked like a RAM problem, but the bits changed every test I ran

You can either scope it if you have one, use a logic probe, or just power it off and use a continuity test on a multimeter to see if all the data lines from the CPU are connected to the MMU.

My Mega also had a CPU socket and something failed on those two lines, I simply jumped them from the socket pin to the next via along and it came back to life.