ST M2716-1F1 21V FAST UVEPROM Behaviour

[NivagSwerdna]

I'm trying to use two 2716 UVEPROMs in an old arcade game and getting very strange behaviour...



On the board pin 21 is wired to a 1k Pullup to 5V but when the ROM is fully inserted it doesn't seem to operate correctly... no data returned.  If I pull the pin low then it seems to operate (although some bits go missing) but if I leave it floating it seems to work.



The ROMs were fully erased and then programmed in my TL8266 with VPP set to 21V

There is something spooky going on.

Thanks in advance

[fourtytwo42]

I wonder if it has not programmed properly, the spec seems to say it should be 25V on vpp, not 21V.

[retiredfeline]

I don't think the TL866 can program 25V EPROMs. Someone once reported that it didn't even reach 21V for 21V EPROMs. You could use a 2816 instead.

[CaptDon]

I know it seems stupid, but maybe they are 'too fast' for the rest of the data/address bus. It makes no sense that 'floating' gives the best results. You could try a stiffer pull-up like 330 ohms and bypass it with a .1uf disk ceramic cap. Also, was the mother board strapped for an Intel 2716 or something else? There were different pinout versions and I think there was the strapping option for (2316?) ROM's. I have noted that many of the old gamers are having troubles similar to yours with mis-matched access times on 2016/6116 rams for instance.

[NivagSwerdna]

I UV erased them and dug out my Data I/O 29B from the loft and programmed them as 2716 (Fam 19 Pin23) which I assume blasted them with the 25V from the 2716 spec...

... and they work with all Pins attached.

So I guess the theories above are correct... they were incompletely programmed by the TL866A and the 21V on the chip is probably a bit optimistic.

At least I can put these to good use rather binning them as 'fake', 'counterfeit' etc. even if there labelling is a bit bogus.

Thanks all.

[Kjelt]

they were incompletely programmed by the TL866A and the 21V on the chip is probably a bit optimistic.

Does your programmer or you, not verify the contents after programming ?
If the eprom verifies correctly it should work.

[Ramakers]

Does your programmer or you, not verify the contents after programming ?
If the eprom verifies correctly it should work.

I have the same problems here.
The TL8266 (I have a TL8266CS) does verify the eprom, but I noticed the chip looses data after a short while. I dumped the fresh rom into a file, right after burning and compared it to the known good dump (by notepad++) to find out they are the same. After 24 hours I compared the rom with TL8266 compare tool and most of the data got corrupted. I'm trying to find a way to get around this.

The datasheet told me there are two problems:
1: the  write time is to short, only 1 ms instaed of 45 minimum. This could be overcome by writing several times?
2: vpp should be 25v not 21V. Therefore I'm planning on making an adapter so I can put 25v to the vpp pin.

I'll keep jou informed with my progress.

[wraper]

Relabeled counterfeit. There is no chance 2716 produced in 2005 would have such large die. Likely some 25V EPROM made in early 80's pulled from e-waste.

[NivagSwerdna]

Does your programmer or you, not verify the contents after programming ?
If the eprom verifies correctly it should work.

Both myself and the programmer have a verify function.   I would imagine that the timing of these verifies may differ from operation though.

Interestingly the earlier burns with the TL866 (which verify correctly in the TL866) do not verify in the Data I/O 29B.

I did think the labelling was contrived but they do appear to work as 2716s with 25V programming so my surplus can stay in the parts drawer for now.

[Kjelt]

I have high doubts about these chips as wraper does.
I don't see why ST would make these antique chips in 2005 when eeprom variants like 2816 are abundant available.
Also the 21V FAST addition?
Can anyone post an official ST datasheet with these markings explained ?
IMO It would suggest a 21V Vpp but in 2005 already 12V was available if you reboot some old chips it would be way easier and cheaper to use the 27256 or 27512 with a much smaller die and higher yield and just put the extra datalines to gnd.   
I am not sure if the old 2716 already had a chip Id where you can verify?

[Kjelt]

Even if ST did reboot the 2716 line it would not do it in different die shapes.
Look at this example totally different die size, much smaller as expected and even with an extra chip (21V to 12V regulator Your chip with the large die makes no sense.

[NivagSwerdna]

For the ROFLs... here is my ragbag collection...

A variety of dies and constructions even within identical labelling... I thought the AMD ones at the top were genuine but now I'm no longer sure!

[Ramakers]

In my opinion this tread is going in the wrong direction. Fake or real is no issue, the problem is can they be burned?
The only thing that is important is, can we use them to keep old equipment running for next generations to enjoy them.

[NivagSwerdna]

YES!  They can be used (see above); you just need to program them with the traditional 2716 programming algorithm... and a Data I/O 29B is proven to work... as probably do other programmers that use 25V and the correct programming timing.

[Kjelt]

In my opinion this tread is going in the wrong direction. Fake or real is no issue, the problem is can they be burned?
The only thing that is important is, can we use them to keep old equipment running for next generations to enjoy them.

Yes but the average charge lifetime is about thirty years maximum with these older dies and that is not including possible layer isolation degradation.
So generations no, one generation yes.
Personally if it was an important piece of electronic equipment I would look for an modern eeprom or smd flash chip equivalent with over 130 yrs of data retention.

[Ramakers]

The datasheet told me there are two problems:
1: the  write time is to short, only 1 ms instaed of 45 minimum. This could be overcome by writing several times?
2: vpp should be 25v not 21V. Therefore I'm planning on making an adapter so I can put 25v to the vpp pin.

For future reference:

My tests gave me this results: (using my cheap TL866CS)
1:writing several times?
There was no reliable result in writing several times, even when I programed the EPROM 50 times in a row some chips lost some BIT's the next day? Some kept working.
There where even chips I could not program at all, they keep throwing errors.

2: I used a machinepin socket and removed pin 21. I provided 25v from my PSU to the vpp pin of the chip while programming. All chips worked fine after one cycle, even the next day.
The chips that could not be programed in the first test, worded fine with the 25v vpp. (after erasing)

For now I did two cycles on the chips that I use in the pinball I made them for (Just in case) and they keep working for one week now, I think they will keep working for a long time. The one's I programed only ones also test fine till now.