> Assuming you mean my original mod from year(s) back then you shouldn't touch > the ground line! The MOD is simply to reduce the +Vcc to the UART ONLY by > cutting the +Vcc track between the decoupling disc-ceramic capacitor and the > UART and patching a *forward* biased general-purpose silicon diode from the > +Vcc solder-pad of the disc-ceramic previously mentioned and the +Vcc pin of > the UART. > This just happens to reduce the UART's logic-zero output to below the critical > point at which it previosly gave false 1's. (Just one of those flashes of > inspiration that worked first time;-))
I'll check with Martin... it may be the same mod and I'm getting mixed up, but it didn't work before we did it, and only worked when we had restored the cut track *and* had the diode in place as you describe as well... How odd.. > I did *TRY* connecting the other SAM-BUS 0V to the COMMS-INT 0V circuit and > I believe it helped when the diode fix wasn't used but wasn't 100% reliable, > whilst the diode fix did make it 100% reliable with or without any other 0V > extras:-) *grins* Don't you just love it. I wish that Bruce had just done the ASIC and everyone else had mulled around on the board -- the sam would have been released much earlier and perhaps would have been less buggy in these respects -- no disrespect to bruce there though. > Oh Si, An alternative to using FiFoFull UART interupt you *could* do RxRDY > and whilst in it's ISR Disable all other interupts and sit there reading the > Fifo until empty or software ring-buffer fills.... If Ring-buffer full occours > first then try issuing a UART command to de-assert MPO manually then hopfully > there'll be enough Fifo positions left for the sender to react and stop > sending > and not lose chars ie a PeeCee Null-modem sender, Modems are ok as they honour > RTS flow-control properly. > It should overcome the interupt latency forced on us by having to determine > who caused the interupt via the asic status reg and then the UART ISR... Funny you should mention this -- last night I connected the SAM to the PC and saw what happened -- lost chars all over the place due to the PC's dodgy hardware flow control. All I can say is... What a chip the IM26C91 is! > If the above works then it might allow SAM to receive at 115200baud I > mentioned > in a previous message about commix;-) Hmmmmm........ If only /my/ SAM (as opposed to Martin's) had a working keyboard and a working drive. I don't want one of the dodgy ones from West Coast though... Actually, we're trying to get the High Density drive working soon... we've made the PCB for the SAM Hard Drive (v0.2) now, and we'll post laser-printed acetates to you (jonathan) as well as a guide to what chip's where) in the near future... I assume that you /can/ photoetch pcbs? :) Simon
