Epson QX-10 Emulator

OK, I bought a QX10 at a radio rally (hamfest), and the seller threw in an extra CPU box which he said didn't work properly, but it might be useful for spares. Well, of course, ard being ard, I decided to find out just what was wrong with this spare machine. The PSU checked out fine, so I fired it up. It seemed to boot OK, but after a bit of screen scrolling the display got terribly corrupted. I forget the exact details, but I think the text appeared in the wrong part of the screen, stuff that had scrolled off came back in strange locations, and so on. It looked like a video RAM problem. Now the QX10 has a daughterboard containing all the video circuitry, it contains an 7220, a gate array for the actual dot shifting/combining (early versions have TTL logic chips for this...), a character generator ROM, a bit of TTL and 16 DRAM chips. According to the techref, the last can be either 4116s (16K) or 4164s (64K), there are links on the board to select which type are in use. I checked, everything was set correctly. The board had 4164s on it, all the links were in the '64K' postiion. So, being young and foolish, I replaced the complete set of (socketed) DRAMs, intending to put the originals back until I'd found the one that was playing up. No, that made no difference. Nor (as I said I was young and foolisH) did temporarilu replacing the 7220 and gate array with the ones from the good machine. And the ones from the defective machine worked fine in the 'good' machine. So I had to think about it. I decided to see what the RAM size links did, these are shown on the schematic. One of them connected pin one of all the DRAMs to -5V, or left it floating. Another connected pin 8 of all the DRAMs to +12V or +5V And another connected pin 9 of all the DRAMs to +5V or an address output Suddenly, it hit me. Not shown on the schematic, but present on the board, were decoupling capacitors. And pin 9 of the DRAM array had 8 or so decoupling caps, 0.1uF each, between it and ground. Fine if 4116s were fitted and that pin was a +5V supply. Not so good if it was a logic-level signal. That extra capacitance added a ridiculous delay to that address line. So I desoldered the capacitors. With all the original chips in the board, the machine sprang to life and worked fine. Presumably a previous owner had replaced 4116s with 4164s, either to get the higher video RAM capacity, or because one of the former had failed and he only had the latter. He'd correctly re-set the links, but hadn't realised about the decoupling capacitors. -tony