Post

I’ve had a dead Executel 3910 in for repair for months, attacking it on and off. It’s the reason I bought my monster Agilent 16702A logic analyser and learned how to use it. I fixed the initial issue and the Executel worked, then it didn’t. This was a crash I’d not seen before. 1/7

The issue was video, the Executel uses the Plessey MR9735 as a video chip, it’s from the TELEVIEW viewdata chipset from Prestel days. One big fault with Executels is the video loses the sync signal at the MR9735 and I’ve never discovered why. 2/7

I have THE machine that STC took around the world to demonstrate the Executel. It’s different. It has different ROMs, a different video chip and a different resistor layout to every other Executel I’ve come across. What did they know? This machine is an early one with more bodge wires... 3/7

The demo machine uses non-interlaced RGB while all the other production machines use interlaced. This makes no sense to the Viewdata people I know. Testing with Tandata Prestel terminals that also use MR9735 and non-interlaced RGB show that all the dead chips work fine in non-interlaced mode. 4/7

The board is the same between machines though.. So I could copy the ROMs, set the board up the same way and it’ll just work, right? No. Not in this case anyway. Cue a long period of troubleshooting and analyser tracing. And the occasional swear. 5/7

Weirdness keeps happening. The machine works but the video doesn’t init - it beeps correctly and the tape seeks. With the Agilent I can see that the code is working the same way the demo machine does, so this is hardware. I continue annotating the EPROM dumps using my Executed source code. 6/7

Eventually I discover the 74LS32 that controls the /RAMSEL and /RAMWR signals is half dead, so those signals never get to the video chip. Today I swapped it (11B, fact fans), and the CACKLE I just did... FIN

Great stuff! Now for the 'Faulty time' issue!