[sdiy] SV: 8254, was:Roland DCOs

[karl dalen]

--- Ingo Debus <debus at cityweb.de> skrev:
> We tried Intel, AMD and another manufacturer (forget which) whose '54  
> were even spec'd for an input frequency of 12,5MHz. We also tried the  
> non-CMOS version (8254 instead of 82C54). They all showed this  
> behaviour. The 8253 worked flawlessly however, but is only spec'd for  
> a much lower input frequency.

Peculiar! 
 > 
> > Also there are plenty of clones under different labeling,
> > altought 8253, 8254 was the system timer chip in 8088 286, 386 PC's  
> > for
> > quite some time. Inside those old PC's there are loads of these chips
> > in them particularly 8088-86.
> 
> Yeah, that puzzled me a lot. How come that chip never caused problems  
> in PCs?? I don't know for sure, but I suspect in a PC the chip was  
> not often accessed from the processor side.

I guess it was simply written once, for example to generate a
timed interrupt in square mode or narrow PW mode.I remember 
the brand uPD as quite a reliable timer clone.

> In my design the '54 ran  
> smoothly forever as long as the processor didn't read from or write  
> to it. That also made me suspect that tWC and tWG timing was  
> critical. Ah, memories come up again...

Usually in a DCO application the timer are set to toggle 50/50 and no
edge/flank manipulation, maybe that was enough so the timer didn't fuck around?

Reg
KD



      __________________________________________________________
Går det långsamt? Skaffa dig en snabbare bredbandsuppkoppling. 
Sök och jämför hos Yahoo! Shopping.
http://shopping.yahoo.se/c-100015813-bredband.html?partnerId=96914325