I've fixed the problem with this colour board.
First was to update the CPLD - That 'nearly' fixed it - the frame glitch now only occured in slow mode.
Out with the 'scope. First I checked the +5V supply - reading good, however, there was excessive AC noise! Well over 2V PP
and even more in bursts that seem to be synchronised with the frame pulse.
The fix was: Add 100n and a 10u tantalum bead under the 6847 (there was unpopulated positions for them). Next - the colour board uses one pin from IC42 on the main atom board, however, connects with the whole right hand row of that IC. As the contacts were sloppy I replaced it with a turned pin socket. Then I connected a link, on the colour board, from what would be the +5V supply to IC42, to the +5V supply on the colour board. And, finally, added a second 10u tantalum bead near to the 3V3 regulator. See photos.
The scrolling is now rock solid.
this may be useful for anyone else having similar problems. The colour board version is 5.10c.
Dave H.
and at some time during the next 12 months we will be moving house to somewhere on the South Coast - so the less I have to pack the better - I most likely won't have such a large workroom 
However, I'm seeing this problem - (starts at 1MHz switching to 1.7MHz then back to 1MHz). I'm waiting for a reply from Phill to see if there is a CPLD update. The board version is: 5.10c (not sure of the .jed file version that it has).
