1) if your goal is to make a "Universal programmer" just for yourself, handling, say, 6 different chips, and you, by a miracle, manage to get hold of the algorithms for programming those, you can do it and with the time it takes vs the cost of a lowcost ready made programmer you will, hopefully, be able to make $0.01/hr. Assuming you, by a miracle, manage to get hold of the algorithms for programming those, I will almost bet it will be easier and cheaper to make 6 boards thaqn to attempt something 'universal'.

2) if your goal is to make a "Universal programmer" for sale to the unsuspecting public, then you will need to support at least 200 different chips from the onset and be ready for immediate response when a customer request support for a chip not in your library.

so, if 1) you will work and work and work and work some more to save less than $100 (cost of a lowcost ready made programmer less the cost of the components you need)

so, if 2) you will need to convince every uC and/or flash and/or EPROM and/or EEPROM manufacturer that you are a bona fide programmer manufacturer in order to obtain the parallel programming specs. Then you need to obtain samples of each and every device to test your programmer. In addition you need to be willing to work overnight to supply updated parameter files for your programmer every time a customer call you "I have chip xxx and need to program it, it is not in your library"

GIVE IT UP, GO ISP or be prepared for some REAL frustration.

Erik

PS 0f course, if you have raised $10M, hired a few people and started the Nimish parallel programmer company the above is moot. But in that case you have hired the wrong people or you would not have to ask what you have asked above.