??? 01/23/07 20:48 Read: times |
#131324 - That doesn't change anything. Responding to: ???'s previous message |
Erik Malund said:
Yes, your example was 25 ppm
I am hammering out 5000 bits serially (software driven, NOT SPI or such) every half millisecond, that, in my book, is more than 25ppm. It's time for a "food fight." It's needed for its entertainment value. Anyhow, let dead horses lie, you just keep shopping in the slowpoke store, you have no risk of meeting me there. Erik Well, the saying is "let sleeping dogs lie." I don't know how you're bit-banging 10000 bits per second, but unless you're doing 20-30 channels at once, each at a different baud rate, it's not taxing your 100 MIPS MCU, and, even then, you could easily do it with an 8255 from a slow-cycle bus, yet still maintain your 100 MIPs rate, as the slow-down would be impreceptible. After all, just the external bus accesses would be slowed down. All the timing would still proceed at full speed. You keep ignoring that. I don't mind, though, as it gives me an opportunity to point it out. I don't care what YOU use, Erik. I just object to your blanket statement that anyone who'd even consider an 8255 for his parallel I/O is making a mistake. The application requirements may not demand a high-speed MCU. Old and slow might fit just fine, but even if the application require a faster MCU, with a slow external bus cycle they'd work in many applications. After all, you get 24 bits of I/O for about $4 in unit quantity, and about half that in hundreds. The I/O is as hefty as any of the MCU's one might be tempted to use. If I were to encounter an application, I'd not hesitate to use 'em, though I'm not looking for such an oportunity. Economics should, of course, play a role. If I know how the I/O has to work, and I'm not building a "general-purpose" board, I'd probably use discrete (74xxx) logic or a programmable part, just because they cost less. If I wanted the flexibility that it offers, I might go with the 82C55, though. RE |