BBC B weird screen corruption

[Albuzz]

Hi all.. I’m a newbie to the forum but have done quite a bit of research before posting this but haven’t managed to find any posts that quite fit the bill, so please forgive me if this particular issue has already been covered.

I’ve recently taken my old beeb out of the loft to give it a walk down memory lane, recapping the PSU before switching on and after almost 40 years of being in storage was very surprised and taken back in time to hear the reassuring blurb beep… all seemed good until I started to play around with it.

In a nutshell, I get screen corruption in most screen modes (not 7) where printing on the top area of the screen results in various corrupted bit patterns appearing as ghosts to the original characters further down the screen. In some modes, clearing the screen followed by simply hitting return repeatedly moving the cursor down the screen results in ghosted bit patterns further down.

I first suspected a ram problem so ran a simple memory test that seemed to show no ram issues, so then used Martin B’s comprehensive test but again it always shows memory to be ok.

That said, On moving S25 to switch out the top 16k, the machine won’t boot at all, implying a memory issue?? If I remove the jumper then it runs ok in 16k mode, so I assume there is some kind of issue related to upper RAM.

I then wrote a program running in mode 0 to poke 65 into each memory location from #3000 to #8000. All seems good with vertical bars appearing from the top of the screen up to address #3FFF, but from #4000 onwards the vertical bars appear twice, once as expected corresponding to the address into which the value 65 is stored but also repeated 8192 bytes further down the screen from #6000 onwards but these vertical bars are faint by comparison to those at #4000, so appearing as ghosts to the original.

I should add that the corruptions appear differently in different modes but always with ghosted patterns related to the real characters on the screen. The corruptions are not random and are always exactly reproducible.

I haven’t done any IC swaps or other investigations yet as I don’t want to tinker with it more than necessary.

My knowledge of the Beebs workings is scant, but would I be right to suspect it to be an address line issue possibly stemming from the CRT chip? That said why would they appear faint by comparison to the original???

I do have a scope and a second Beeb that might be helpful to diagnose things.

Any pointers would be greatly appreciated

Thanks Alan

Added 26th Jan: I just noticed that when running Martin AB’s memory test in all modes it to 6, the are ghost bit patterns appearing faintly and intermittently at various positions usually progressing down the screen as memory is tested. They are easily missed except for mode 5 where the are a few random patterns in yellow that don’t move or disappear, then in the last phase of the test, the screen is filled top down with random white vertical bars but when they reach the yellow corrupted bit patterns they don’t overwrite them but seem to superimpose them so that the yellow bit patterns appear behind the white vertical bars!! Could Thisbe a fault with IC2?