The problem of programming ATMEL MCU's has been a frequentl and recurring topic of discussion in this forum.

Why is that?

In the past, beginning with the original Intel 8751, I've used the manufacturer's recommended circuit for programming the things and been done with it. Today, I use a commercial programmer, i.e. a Data I/O UniSite or one of several physically smaller portable units to do the job.

What is it about ATMEL's parts that makes them such a problem?

This particular circuit is one of a dozen or more very similar JTAG-capable programmers that I've seen. I've used similar circuits, both home-built and commercially manufactured, that program FPGA's and CPLD's. They've seldom been a problem of any sort. Why are these so difficult to use with ATMEL parts? Is it their (Atmel's) software? Is it the construction technique implemented by home-builders? What, exactly, is the problem?

Circuits like this rely on the parallel port, and, of course, everybody who produces one has to use a different arrangement of the signals. To cope with this, I once used a crosspoint switch in a DB-25 shell case to enable me to rearrange the connections between the parallel port and the target circuit. That, thoug it was simply an exercise in mental masturbation, seemed to work just fine.

The circuits that Hendryawan proposed are SPI-based rather than JTAG, but once again, they're very similar. Why do you suppose that ATMEL is the target of so much inventiveness with respect to ISP programming circuitry? Don't any of their circuits work well enough to satsify? I'd bet they do actually work.

RE