I am in TTL chip heaven (India) where I can get these things for pennies on the dollar, so I bought batches of chips to tinker with. Among them a 16 AT28C16 EEPROMs, looking like new, and then 2 very old AT26C64 (or AT2664 without the C). Not finding any ready-made EEPROM programmer, I quickly (more or less) rebuilt on a breadboard the one I learned from Ben Eater, with an Arduino Nano, and two 74HC595 shift registers.
I revamped my software so I could control the programmer from the terminal, and test individual pins to debug the wiring. I first did it for the 26C64 chips because they looked so old and worn, and I wanted to make sure they actually worked. Turns out they did. Perfectly.
Then I looked for the easiest way to plug in the smaller 64C16 chip without having to do lots of re-wiring. I was happy how nicely the pins line up between these models. Here I showi the two pin-outs from their datasheets overlaid to see how everything aligns on the GND pin, so well indeed that they had the kindness to leave the old Vcc pin 24 as 26 NC in the larger model. Pin 1 of the '64 isn't needed. The only variance between the two is whether A11 or /WE is routed into the matching pin 21/23. A single jumper wire allows me to make that choice in my setup.
I know it works for the '64 and I tested the wiring and functions are correct for the '16 configuration. Yet, the '16s don't work. None of them do. I tested 4 of them so far. They don't even read correctly. Every time I read, the values come out slightly different.
Could it be that this whole batch is bad? That they are just fakes?
One observation I had, because I don't know what I did wrong once, after my test one of the '16s got very hot. I unplugged power. I could not see how I had it inserted wrong. I took it out. Later I could not reproduce the problem again. I don't know what I did wrong. Definitely not reversed inserted. Perhaps offset by one, leaving GND unconnected that one time. Not sure.
One of the '64s I had inserted reversed once, that is usually what fries them, but there was no problem and it continued to work afterwards, but none of the AT26C16s work so far.
What could I possibly have done wrong? It all seemed so easy. I tested all pins with and without chip inserted, I cannot find any mistake.
The ever helpful Bruce Abbott suggested I post pictures of the possibly fakes. Here goes:
Ha! I found one in 8 that actually worked! And here I made photos of front and back side, and lo and behold, the working one has something printed on the back, and the notch is also different, non-standard square and bigger rather than round and normal size.
So now I will tell the store and see if I can find more in their stock with the printing on the back.
Electrically I also checked that the non-working chips have all data line at some kind of high-impedance state, no matter if the /OE is high or low. I tested by trying to light an LED sourcing from the pins as well as sinking into the pins, and nothing worked correctly.
Maybe I should give them a 12V shock on /OE (the quick erase protocol)?
