Mike's PCjr Page

jmetal88 wrote:Alright, it's a terrible video, but here it is.

Doesn't look terrible. Nothing like seeing it actually work! That's very cool!

jharre wrote:

jmetal88 wrote:Alright, it's a terrible video, but here it is.

Doesn't look terrible. Nothing like seeing it actually work! That's very cool!

Thanks! When I said it was 'terrible' I was referring to the fact that I couldn't hold the camera steady, had to type with one hand, and am mumbling so badly I can't understand some of my own words, haha.

I just placed a Digikey order with two goals:

1. Get this thing working with a crystal oscillator instead of whatever's generating the internal signal. Hopefully this leads to a stability increase and fewer key presses missed by the PCjr.

2. Possibly get the code running on a cheaper 8-pin PIC. I mean, aside from programming, which can be done by taking the chip out of circuit anyway, I only need three accessible data lines on the chip. I went ahead and ported my code as best I could to my selected candidate before I ordered it, and it still fits in memory, so I think I'll be okay to run it when it comes in.

No new upload yet (I kind of want to wait until after my Digikey order comes in so I can see if it works on the new chip) but I just implemented some rudimentary error detection. I can't tell if it's working or not yet, all I know is my output is still what it's supposed to be. Basically, I put a time-out on the PS/2 read routine because for whatever reason, every once in a while it was getting off by a bit. So it's supposed to reset the read routine and flag an error bit if the read bit counter gets stuck above 0 for too long (and then it's supposed to re-request the last scan code from the keyboard). I may try to implement parity detection eventually as well if this doesn't seem to be enough.

I just had a great game of Paku Paku on the current code, with no error beeps heard from the PCjr at all, also getting several levels farther in than I usually do (and a score about 10000 points higher than my last score). This makes me think using an external clock crystal is probably going to fix pretty much any of the issues I've been having. I'm also really interested to try out the code on the cheaper, smaller PIC12F617 when it comes in in a few days, though.

So are you willing to make+sell kits and/or assembled units for this? I'd buy two of them today- just name your price!

For a production kit, I'd want to use the 8-pin ICs. I've done a little research on parts cost, but not enough to be able to quote you a price yet. I know a place I can get PCBs done for cheap, but I have to order 10 of them at a time.

Just got (hopefully) past an issue where the PIC wasn't fully resetting if a power off was too brief. I enabled the Power-Up Timer, which is supposed to hold the PIC in the RESET state until a certain number of clock cycles have passed. Then I started running into issues with my time-out activating after nearly every key, but since then I adjusted it for the actual maximum time it should take a scan code to read (I had done it based on a 16MHz clock since that is what I'm moving to, but I'm still using an 8MHz clock at the moment) and it seems to be doing well again now.

Hargle: Since you seem to be the main person interested in actually buying one of these, do you want me to include an ICSP header on the board, or would you just as soon remove the PIC to reprogram it if need be? Since I plan on moving to an 8-pin PIC I would have to add a multiplexer on the programming pins to keep from messing with the PCjr or PS/2 keyboard if I put an ICSP header on board. Not that that's much cost (about 60 cents a board I think) it's just that it is an extra chip and I'm not sure if it should go on the production board if I'm the only one who might need it.

Regarding price: If I order enough parts to make, and assume I can sell, four of these, I'll have to sell them at about $25 apiece to make back the money I spent. To make my money back on just two of them, I'd have to do $35 apiece. This is assuming you want features such as the proper AMP connector with the proper metal shielding on the outside. I could make it a bit cheaper if you just wanted the plastic 6-pin connector, say, encased in heat shrink tubing or something. And it looks like without the ICSP header it'd knock off about another dollar. A hand-wired version like the one in my demo video would be quite cheap to produce, but I'm assuming nobody actually wants one like that.

EDIT: I was also toying with the idea of just making the board as skinny as possible (obviously no ICSP header in that case) and putting a right-angle 6-pin connector on one end. That would be the REALLY cheap yet still somewhat professional way to do it. Assuming you don't mind a 5cm board sticking straight out the back of your PCjr that may or may not partially block the Light Pen port.

I'd buy at least one @ $25.00. Trust me.

Mike