Zane H. Healy wrote: :-) On a related note, is there such a thing as a homebrewed cross-reference showing which EPROMS can be treated as the 'same' from a programming point of view? For instance, my programmer knows nothing of Mitsubishi EPROMs, but does acknowledge many other manufacturers. I had no luck earlier trying to program a Mitsubishi 27218 part as an AMD EPROM of equivalent size, but it worked fine by telling the programmer that it was an Intel 27128 - presumably the timings vary slightly between manufacturers, but it's often possible to find something that's "close enough". (If anyone else has an old Elan universe-series EPROM / PAL programmer, shout. I'd be interested to know if you have programming cartridges that are different to mine) cheers Jules

On 14 Sep 2007 at 16:13, Chris M wrote: Not entirely--particularly in the 128Kbit and under categories, the programming algorithm could vary significantly from vendor to vendor. I suppose that if one programmed an EPROM slowly enough, one might be able to do it in an algorithm-independent manner, however. Cheers, Chuck

Many years ago I hacked a driver to program 27xx's using a SD systems S100 programming card. If memory serves me correct the the I was using want a 100ms pulse to each address 100 times to build up the bits with out over heating the gate. It has been over 30 years since did it. There were several voltages and duty cycles that were prefered by different chips, but that info was stored in what now are mostly dead brain cells. Now I will have to go look for the SD card, I am sure it is still in storage in one of the boxes .... later The other Bob On Fri, 14 Sep 2007 16:49:21 -0700 (PDT), Chris M wrote:

It depends on the chip :-). If we ignore the really old devices for the moment, and concentrate on the 2716 and later, there are 2 common algorithms. The old, slow, one goes as follows : 1) Power up the chip 2) Apply Vpp. This is the programming voltage, and was 25V on old chips, 21 on slightly newer ones, and 12.5V on newer ones still. If you use too high a voltage, you'll damage the chip. If you use too low, either it won't program at all (which isn't too bad), or more worryingly, it'll programm, but will loose data soon afterwards. 3) Select he location to be programemd using the address pins (normal TTL levels) 4) Apply the data to be programmed into that location to the data pins, again at TTL levels 5) Give a 'programming pulse' This is a 50ms TTL level pulse applied to a particualr pin o nthe chip. Some devices have a special PGM input. On others you apply it to the OE/ pin or something like that. The data sheet should tell you. and 2732s, it's OK on 2764s and 27128s, but at least some manufactuers don't recoemnd it on larger chips The other algorithm is called the 'intelegent algorithm' and is faster. The first 4 steeps are as above, but instead of applying a 50ms programming pulse, you apply a 1ms one, and then try to read the data back. If it doesn't read back correctly, apply another 1ms pulse, and so on, until either it verifies or you've given 50 pulses. In the later case, the EPROM is bad (!). In the former case, you then apply 5 times as many programming pulses as you already have given (I believe either as sepearate 1ms pulese, or as one long pulse). Some manufactuers suggest doing that verification of the programmed data with the 5V power pin at some other voltage (often 6V). And that you should do a final verification of the programmed data both with a high supply voltage and a low one (4.5v) just to be sure. IIRC, for some later EPROMs there was a way of reading out a manufacture/device ID. It involved applying Vpp, and probably pushing the 5V lione up to 6V, and then doing odd things with the other pins (but at TTL levels). This may be one some programmers refuse to program some manfuacturer's devices, they don't recognise this ID (even though using one of the 'geneirc' algorithms would work. -tony

Hi My experience is that the old 50 ms pulse method works for most any of them. Most of the newer algorithms are related to programming as fast as one can. If you don't mind taking a long time, the slower pulse should work. They also usually have special voltage conditions for manufacture ID and checking programmed values when running fast algorithms. Dwight _________________________________________________________________ Capture your memories in an online journal! http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us