On 12/17/2013 09:33 PM, email@example.com wrote: > > On 2013-12-17, at 20:35, Gerrit Heitsch <firstname.lastname@example.org> wrote: > > >>> Those I have don’t, but as Gerrit corrected me - it makes sense: there was no intertwining on the original PATA (typical PC crap engineering - like with most PC stuff when compared to the rest..) >> >> Gets even better. There were even IDE-Cards that omitted the 74LS245 buffers between the ISA slot and the IDE connector. That means that the data bus from the ISA slot was directly connected to the HD over a 40wire cable with no shielding between the signals whatsoever. >> >> >> The funny thing was, it usually worked. > > Like almost everything PC for about two decades - it mostly, usually worked ;-) > > At least I believe the other end was not connected directly to the CPU pins like in the SX? No, only the ISA bus which should be buffered from the rest of the system, especially since it runs at a lower speed once the 80286 passed 8 MHz. >> I sometimes wonder if one could improve stability if one removed the pullups for A12 to A15 and replaced them with a 74LS125. The inputs connected to +5V, the enable signals to AEC. That way A12 to A15 would no longer slowly float up to +5V when VIC takes over the bus but get pulled HIGH quickly. Maybe the same for R/_W. > > BTW - I wondered why is it done like that in the first place? I mean the address lines, not R/_W. Well, once the CPU is taken off the bus via AEC going LOW, there is no one left to drive the upper address lines. VIC and the 74LS373 only supply up to A11. Everyting else stays in the multiplexed area between the 74LS257, VIC and the DRAMs. So you have A12 to A15 floating. This should not confuse the PLA when AEC is LOW since _CS for the Char-ROM does not depend on A12 - A15 with AEC LOW, but you don't want open signals, just in case. Simplest way to fix it are pullups. Cheaper than a 74LS125 for sure. >> Another detail I'd take a closer look at would be the 7406/pullup combo that tristates the output the 74LS257 multiplexers when it's VIC's turn. Replace it with a real driver (74LS04). > > That “nice” 180 on the OC output. Right - another interesting idea. It was probably a good (read: cheaper) solution and a good use for the otherwise unused 7406 gate. Yes, and on the older boards R31 was 1KOhm and it still worked. Oh, BTW, it seems 180 Ohm is a bit low, not too long ago I had 2 C64 boards with the same fault, the OC driver in the 7406 that handles the 180 Ohm pullup was dead. Both 7406 were made by NatSemi, maybe they at a dodgy run at one time. I agree that it was probably using that leftover driver and a low value pullup or use another IC, so they used the leftover driver. With 180 Ohm that alone eats more than 20mA when pulled low. Thats more than the power draw on a CMOS EPROM. Gerrit Message was sent through the cbm-hackers mailing listReceived on 2013-12-17 21:03:44
Archive generated by hypermail 2.2.0.