When you say "universal", you don't really mean universal, do you? You probably have a short list of devices you wish to program.

First of all, not all universal programmers, in fact, NONE, are limted to that 40-pin ZIF socket. My universal programmer which really does seem to be quite universal, at least up to a point in time, has 48-pin ZIF, PLCC's up to 84 pins, SOIC's from 8 to 32 pins, PQFP's of a range I can't even remember, etc.

Secondly, what I've concluded from your remarks, you want to program a number (perhaps larger than three, but finite, nonetheless) of different MCU's, not a universal range of programmable parts. Consequently, what you need is a precise, to the nanosecond and millivolt, specification of what each of the MCU's that you wish to program requires. Until that is in your hands, and thoroughly understood, you have no chance of succeeding at fabricating a "universal" or even somewhat less general, programmer.

Surely you've noticed that the majority of complaints regarding programming of ATMEL or Philips/NXP MCU's have involved "home-built" devices. From that, you should conclude that it's not a trivial exercise.

Most MCU-programmers published, and, in some cases, produced and sold, by the chip manufacturer handle exactly one MCU. There's good reason it works that way.

Some third-party programmers have "issues" that surface from time to time, resulting in reduction of repertoire, due to damage to some of the parts, or simply failure to program them. In short, they don't always work as promised. Surely there's a reason for this.

Once you have the complete list of all the devices you wish to program, and once you have the complete set of programming specifications for them, you can set about to build a circuit that controls, pin-by-pin, the exact voltage and current to which the device being programmed is subjected. There are many ways in which you can do that, but until it's done, you won't be able to proceed.

I have attempted, in some cases with success, to build programmers for various small sets of devices. Over the long run, however, I've found it less trouble and less expense, simply to buy something that does what I need. Then, if it fails to do what I was promised, I have someone to whom I can legitimately complain.

As for the '245 ... you can replace it with 2 'HC125's and 2 'HC126's. Each bit will then be individually controllable and individually enabled in either direction. You may even find that you don't need to drive both directions ... You might even consider using programmable logic ... but then you'd have to program it.

RE