[cctalk] Philips P2000C restoration

Philip Belben gave me a Philips P2000C luggable CP/M computer some time back which had not been well-stored. It took a bit of work to get it going again. Here's what I did... The basic desgn is a single-board computer with a Z80A, 64K RAM, 4K ROM (bootstrap and a machine code monitor), floppy disk controller, SASI interface and 3 serial port. One for an external printer, one for communications [1] and one to provide a 19200 baud link to the other main circuit board. This is an intellegent terminal with another Z80A, 32K RAM, video circuitry, keyboard interface and of course a serial port [1] Standards are wonderful, everybody should have one of their own... This serial port is on a DB25 connectorr with the normal RS232 pinout, but normal RS232 cables probably won't work. The reason is that Philips decided to support synchronous operation too. So the serial chp (Z80A-SIO) clocks come from pins 15 and 17 on the connector via level shifters. The baud rate generator (one channel of a Z80A-CTC) is level shifted and comes out on pin 24 of the connector. You therefore need to strap 15-17-24 in the cable plug for normal asynchronous operation. Getting back to the machine, as well as the 2 main boards, there's a switch-mode power supply, a Misubishi 9" green screen CRT monitor (Why not Philips, they were certainly making such things at the time), a pair of Teac FD55 floppy drives (of which more later) anf the'power distribution PCB' to link them all toghether. Oh, and trivial things like the keyboard cable and mains input wiirng. I have the Philips service manual which contains schematcs for the 2 main boards but not the rest. I also have the Teac service manual for the floppy drives. Obvious faults on first inspection were that the mains on/off switch didn't latch properly, there was a lot of corrosion, and the carrying strap was missing. The last is important as to carry the machine you put the keyboard over the front panel, then slot the strap end fittings in place which also retain the keyboard. I took the machine apart and found that the terminal PCB at the back had suffered badly from poor storage. So had the disk drives, the spindle bearings felt very rough. The aluminium chassis had surface corrosion. Screws were very rusty (but standard M3 and M4 parts are not hard to get). The rest didn't look too bad. Time to sort some things out. I traced out the schematics for the power supply, monitor and the power distibution stuff. There were some RIFA 'smokebomb' capacitors on the PSU board which I replaced before they did their antisocial act. Since the mains switch was out of action I coupled a suicide lead to the power supply input pins with a chocolate block and carefully powered it up with a light bulb in series. The power supply worked first time. Tried the monitor board, running it on the bench supply. This powered up too, the high voltages came up but were low. As I didn't have the deflection yoke connected this didn't worry me. So I put the monitor chassis, PCB and CRT bak togther and connected it and the terminal PCB to the units power supply. Powered up, the screen was full of odd characters. It was clear the terminal processor wasn't doing the right things. Some checks showed the data lines on the RAMs were not looking right.Well, a couple were, but not the rest. Cut out the old RAMs, most of the DIL packages fell apart (!), fitted sockets and new 4116s. Corrected one open-circuit PCB trace too. Powered up again ,it seemed to work. Tried connecting the main board. It powered up and even gave the right startup screen asking for a system disk. Of course no drives or keyboard at this point, but it was a good sign. Took the keyboard apart, took off all the keycaps and removed the dregs of many cups of coffee. Put the keycaps back on. The keyboard cable, right-angled 6 pin DIN plugs at each end, was a mess. Insulation crumbling off, green corrosion of the wiring. Fortunately the plugs are not moulded, so I could open them up, remove the dead cable and rewire with a length of 6 core screened. It's not coiled stretchy stuff like the original, but it's electrically fine. Time to sort out the mains switch. I took it apart. An internal, tiny, spring was so badly corroded that it fell apart when I touched it. Other bits didn't look great either. My junk box disgorged an electrically-suitable switch that was actually a spare for a TV set. Only problems were that the pushrod to fit the button onto was 1/8" square (the original one for the P2000C was 3mm) and the mounting was very different. A file cured te first poblem. Fortunately the switch mouting was a little plate screwed to the PSU mounting, so I removed that and milled a block of aluminium to mount the replacement switch. Soldered the mains harness wires to the new swtich. While the chassis was apart I measured up and made some suitable end fittings for the carrying strap. Oriiginals were plastic, I made aluminium ones. Not too hard in that the tongue that goes into the P2000C catch is 30mm wide by 2mm thick and amazingly a local-ish DIY shed had 1m lengths of 2mm aluminium strip 30mm wide in stock. Cut lengths of that, drilled and milled the hole to engage with the catch, fitted a metal block to retain the keyboard and an eyebolt into that to put the strap on. Now to reasemble the chassis. Fitted the mains wirng, keyboard connector, distribution PCB, PSU and monitor. Plugged in the terminal PCB and connected the keyboard. Powered up then reset while holding <esc> down. This runs a simple self-test of the terminal board. It failed with a memory problem. I found another bad conneciton, this time a through-board VIA. Soldered a bit of wire through that and the terminal board then passed the self-test. I temporarily fitted it to the chassis so as not to have too many bits hanging on wires. Put the main PCB on top of the chassis, connected the power, reset, and serial connectors. Powered up, got the 'system disk' prompt. Pressed <esc> then and was in the machine code monitor. I could display/change memory, etc. It was essentially working. OK, now for the drives. These are Teac FD55A, single sided 40 cylnder. I took them apart one at a time. Not just to the units in the service manuak, I also took the head-load unit apart (tiny torsion springs), the top front chasss (even smaller E-clips), the stepper motor (the front bearing could not be removed without possibly damaging thngs, but the rear came off easily with a puller so I fitted a new ball race here) and the spindle motor (again, new ball races fitted). Got the drives back togther. They ran nicely on the exerciser. Much more smooth than they were when I took them out. Connected them to the Microtest alignment unit and did the head alignment. One oddity was that both spindle motors were slightly slow (about 295 rpm, not 300) but a tweak of the pot on the motor PCB cured that Also set up the disk read VCO on the mainboard as described in the service manual. It was a little off, I am sure it would have worked, but I re-set it anyway. Cabled up drive 0. Powered up and put a 40 cyclnder boot disk in. It booted. DIR worked too. As did running a program off the disk. Unplugged things and removed the terminal PCB. Put the 2 drives in place, fitted their mountings and the chassis top rail. Fitted the main PCB and terminal PCB to the rear chassis plate. cabled everything up. Tried the machine again. It booted. I could format a blank disk in the second drive and copy the CP/M master too it. The copy then booted fine. All that remained was to fit the rear plastic panel and top cover. Stored the boot disk copy and the keyboard cable in the cubbyhole on the front panel and put the keyboard on. Clipped on the carrying strap. It's not quite over... I am pretty sure my strap end fittings are strong enough. Not so sure about the strap itself which is one that came with a sports bag. I may try to get something stronger. I was given a few floppies with the machne. The only one it will read is the 40 cylinder boot disk. Philips, you see made 3 versions of the machine. One had a pair of 40 cylinder single-head drives (160K each). The second had a pair of 80 cylinder double head drivs (640K each). The last had a single 80 cylinder double head drive and apparently you could fit a 10MByte wnchester internally. I know nothing about that really.. Confusingly, the manuals call the 160K drive 'single density' and the 640K one 'double densiry' for all both use MFM encoding. But I digress. My guess is that at least some of the unreadable floppies are 80 cylinder. It would be worth linking up an external drive to see. Time to hunt in the junk box again. Then there's the SASI port. One manual mentioned a hard disk unit to connect there, a 'Xebec board and 1 or 2 10M drives'. My guess is that the former is an S1410, the latter a pair of Shugart ST412s or similar. But it seems crazy to me to try to track down said parts -- the Xebec board has serveral custom ASICs on it, hard drives can headcrash. Or even worst to use a Xebec controller with a drive emulator -- why convert bytes to a curious serial stream on the Xebec board and then back to bytes to store in flash memory on the drive emulator, or vice versa. It would seem logical to simply make a thing that connects to the SASI port, accepts the commands set of said S1410 controller and stores the data in flash memory directly. Any suggestions as to how to do that? Finally, the terminal board has an external video output It's a 5 pn DIN socket, separate syncs and analpgue video (not composite). There is a mention of a 12" monitor in one manual, of course with no model number. Odd, I wouldn't have thought 12" was much of an improvement over the built-in 9" unit. I would have expected something larger to show a group of people at once. But making something to connect to that output is another project. -tony