80-Bus News |
September–October 1983 · Volume 2 · Issue 5 |
| Page 28 of 67 |
|---|
IC17 pin 2 – top right-hand corner of the board – and hard wiring it to pin 6 of IC9 – its immediate neighbour). He then tried ORing /M1 and the /XROM signal in an LS32, and using that to drive 1C17/2, which was where failure set in. His problem lies in the accumualtion of delays within the circuit. Let’s try a quick total up:
| Clock falling edge to /MREQ low | – 85ns max out of Z80 |
| /MREQ delay through buffer | – 12ns (?) |
| /XROM decoding delay thru IC47 | – 25ns (typ) |
| Delay thru’ LS32 | – 14ns (typ) |
| D-type data setup time (IC17) | – 20ns |
| TOTAL | =156ns |
Time available from clock falling edge to rising edge that clocks the wait state generator = 125ns (4MHz system in an ideal world). 156 is greater than 125, so failure! In practice the /MREQ delay is unlikely to be the full 85ns, and in fact his letter indicates that the /WAIT input does just make it low before the rising edge of the clock, but obviously not soon enough. The overall delay must be reduced, and one way is to do the address decoding separately (i.e. duplicate the PROM) and remove the dependence on /MREQ. The new PROM would be gated by /M1, and the LS32 would be discarded, the decoded output of the second prom being connected directly to IC17/2. (Alternatively an LS155, LS138 or similar decoder plus a few gates could be used in place of the PROM.) The timing figure would now be:
| Clock rising edge to /M1 low | – 100ns max |
| Clock rising edge to Address valid | – 110ns max (NB 10ns Worse than above) |
| Delay through buffers | – 12ns |
| PROM (or logic) decoding delay | – 25ns |
| D-type data setup time | – 20ns |
| TOTAL | =167ns |
Time available to rising edge of clock, 250ns. – plenty to spare!
Notice we have gained in two ways, first the decoding delays have been reduced
as the LS32 is no longer used, and secondly everything is now refered to the
rising edge of the system clock and we have an entire clock cycle available,
rather than just half a cycle that was available previously.
Hopefully the above is of some use if all else fails, but I would first check the circuit modifications done for the 2716s in the hope that matters can be cured there.
I only seem to have covered one topic this time, but at least there isn’t a mention of discs anywhere in it. Remember, the above is fueled by your letters, so write!
| Page 28 of 67 |
|---|