it IS possible to interface a number of 8255's via an 8255. Just because one can, as you say, doesn't mean that one should.

However, one CAN belittle the 8255 but that doesn't make it useless. It still does what it always did. It also lives on eval boards, and in textbooks, which is how it finds its way into queries about it here.

It's come up in queries enough times, to warrant discussion and it's still incorporated in a number of eval boards, though I can't comment on their quality, and, of course, it's still readily available in Asia, not to mention that Toshiba (?) apparently still produces it. Lots of people have history with it, though most dislike it because they never really learned how to use it.

If you don't want to have the all-too-common trouble programming ATMEL parts, which, BTW, are also quite available for God only knows what reason, in Asia, but probably because no serious user in other parts of the world will put them into his design, and you want to hook up an 8x51/52 type as an 803x, then it's little extra trouble to set up an 8255 to replace the ports that are lost to the external memory interface circuitry.

Quite honestly, I've never admired the i8255 because it didn't have enough output current to drive a significant load, and, back when I considered it a candidate for application, it cost more than the TTL with which I replaced it. However, that TTL was never reconfigurable as whatever else the 8255 could have done.

Nevertheless, I see no reason to write off the 8255 out-of-hand just becuase it's an "old" device (I'm getting pretty old myself, doncha know) since it still works if used as intended.

I'll wait to see whether anyone can come up with a good viable replacement for it that's availble for and applicable to the requirements of various Asian users. What I'd expect is something that can operate with one-time setup, not a serial configuration that takes multiple microseconds to transfer data to the device and then takes considerable time to return an interrupt, if it should occur. I'd expect data to appear at the outputs or in the accumulator of the MCU no more than a microsecond or maybe a microsecond and a half after the MCU sets about to write to the port, or read from it, which is how quickly the device would transfer data if attached to a 12MHz 8031. I'd also expect it to handle hardware handshaking, which the 8255 can do.

Now, is that a big enough challenge for all you knee-jerk 8255-haters?

I'm not saying this CAN'T be done, but nobody's pointed out how it SHOULD be done.

RE