I am beginning to wonder if the problem might be the ULA. The reason I say this is that when I set up the logic analyser to *stop* on a zero ROM address, it didn't. Which means it isn't resetting in the way I thought it was. When I traced the instruction addresses during boot up I could see lots of resets back to address zero, and assumed that somehow it was getting further into the sequence to show the copyright screen, but then resetting again, because the screen would display the "snow" after showing the copyright screen.
So if it isn't resetting then something may be just interrupting the video display, and the machine just thinks it is showing the copyright screen and waiting for input. This is further evidenced by the fact that the ROM addresses when it is in this cycle all seem to be in the keyboard input area, based on a PDF that documents a disassembly of the ROM that I found on the web. If that was the case I thought it might respond to keypresses, but I checked and it doesn't, and the same happens with a second keyboard from another machine. However, the ULA is responsible for the video signal and for the keyboard input, so it suggests the ULA, or something around it, could be the problem. I have a second Spectrum (with a different fault that I think is the memory) that is a later design. It has the 6C001E-6 ULA, the older machine (which appears to reset) has a 5C112E ULA. Looking at this page http://www.bytedelight.com/?p=41 the 6C001E-7 (not -6) ULA can be used on older machines. Does anyone know if, just as a test, it would be safe to put the newer 6C001E-6 ULA in the older machine? Fortunately both ULAs are socketed, so this would be an easy test. Thanks Rob
