On 05/14/2012 07:59 PM, MikeS wrote: > > Oh, come on; unless you insist on using some really ancient and obsolete > versions (and can program them) even NMOS EPROMs still only draw more or > less the same power as the ROMs you're replacing, and we are after all > talking about replacing 3 chips with one. My programmer can do, among others, 2716 and 2532, the latter being handy for having fun with the CHAR-ROM since it doesn't need an adapter... I just don't like the idea of wasting power if it's not necessary for the circuit. > >> * I was worried that the delay of enabling the ROM and it's output >>> drivers would cause timing issues. On a 250nS 27C64, it can take up >>> to 350nS from !CE to data valid. That's compared to 120nS for !OE to >>> data valid. >> >> Where did you get the 350ns for a 250ns EPROM? > > Yeah, AFAIK the speed rating of an EPROM refers to the /CE to valid output > delay so it would be 250ns, but in any case the /OE to output delay is > always shorter AFAIK and sometimes substantially so. > > So, all in all, considering that a CMOS EPROM will almost certainly draw > substantially less power than the NMOS ROMs you're replacing and that using > /OE to select will give you a valid output more quickly than using /CE, I'd > say that using /OE is actually 'the better way' ;-) The only thing you gain by using _OE is that you run the external decoding for your select signal and the internal decoding of the addresses supplied to the EPROM in parallel. It will still take the time printed on the EPROM before you get any valid data though. It makes sense to do it this way if the sum of EPROM access time and external decoder delay would be too much for the frequency used. If that's not the case, it's pointless. Gerrit Message was sent through the cbm-hackers mailing listReceived on 2012-05-14 19:00:29
Archive generated by hypermail 2.2.0.