Marcel Kilgus via Ql-Users wrote:

> As I've understood your chip is 3.3V and the lines (or GND) is too
> noisy so that logic level 0 could be interpreted as 1 before the lines
> settle

Yes this is one of the aspects, but it's not a 3.3V versus 5V issue. 
(The relevant logic levels are the same.)

The rise and fall timings of the SGC (and to a smaller degree GC) 
are too fast for the QL mainboard, which creates too much crosstalk 
and ringing. Even without QL-SD, illegal levels occur and can be 
seen on oscilloscope. The QL itself only works because it's hardware 
is too slow to "see" these illegal levels.

> because the chip samples them faster as older chips would, is
> that about right?

Roughly. A 400 MHz chip inside an ancient 8 MHz design.

> Would it help to put a 74-style Schmitt-trigger in
> front of ROMOEH or something like that?
-> qlromext-sch.pdf, see U4 ;)

Unfortunately this was not good enough to get the SGC working.

I guess the best solution would be another QL-SD design with an old, 
slow PLD. Second best: Experimenting with serial resitors in all 
lines. A different place (e.g. external ROM port) might also help, 
but it is hard to predict.


QL-Users Mailing List

Reply via email to