If I've berated you at all it's been because you frequently ignore what the person making the request for information asks and recommend something else instead. In this particular case, the original poster made it clear that he understands some of what he's doing. Clearly, he doesn't understand it well, but telling him to do something else probably won't help him.

Apparently you've overlooked that the Dallas chips, i.e. the ones with on-chip program store, e.g. that DS89C430 he previously mentioned, are, in fact, ISP chips, though they won't, IIRC, allow you to program the bottom 32 kBytes with IAP. Since that's not his stated goal, I don't suppose it matters.

I fundamentally agree that some of the SiLabs devices may work as well as the Dallas chips he's mentioned, but those aren't the ones with which he's asking us to help him. However, what I recommended was that he use the ISP feature of that Dallas part, which simply requires a serial port. While the parallel programming circuit and algorithm was published in the original DS89C4x0 datasheet, I've never recommended it above the serial (RS232) version, since the serial version is so simple.

I'm still convinced that when a guy asks how to implement external memory, utilize an 8255, or do something else outside the current mainstream, it is a disservice to him to tell him to use a component other than what he's asked about until after you've told him how to do what he wants. Once you've told him how to do things in what you consider the wrong way, since it should still work properly, you can recommend an easier way, since it should become apparent to him that your way is better. Your frequent, "Don't do what you want, do what I want!" is not helpful.

RE