1

One of my customers brought in several old EPROMs that he wishes to be re-programmed with the current firmware their system is using.

My usual way of dealing with this is to place the EPROMs in the eraser - this has a 10-minute timer. When the first time cycle has elapsed, blank-check the chips.

If they test blank after one 10- minute erase cycle, put them back in the eraser for another 10 minutes.

If it takes more than one erase-cycle for them to become blank, count the number of 10-minute cycles for them to become blank, then double that erase time.

These 5 chips came up as blank after only one 10-minute erase cycle, so they all got an additional 10 minutes of erasing.

I'd like comments on the above. This has worked well for me in the past but I'd like the guidance of people smarter than myself.

My real question is in regards to the actual chip programming. I'm worried that the chips may not hold their data as well as new chips.

What I'm thinking of doing is double-programming them. Program each chip as normal and ensure they verify okay. Wait a day, then verify again. If they verify properly, then program them again (and verify).

These EPROMs are NOT in mission-critical equipment. And although I used to regularly do this for customers oh, so many years ago, I want to give the customer the best chance these chips will work correctly for the indefinite future.

Yes: the obvious answer is to purchase brand-new chips. Problem is these are mostly obsolete and what I'm finding is New Old Stock. That doesn't make the new chips much better than what the customer brought me to use.

Programmer is a Xeltek 560U and the chips are 27C020 and M27C2001 (256k x8).

Again: guidance very much appreciated.

[Edit]

I did peruse the archives and found Older Tread about multiple programming cycles

Dwayne Reid
  • 23,390
  • 2
  • 35
  • 67
  • When I used to have EPROM based products, I ended up replacing the UV-erasable EPROM with plastic OTP (One Time Programmable) versions. The high cost of EPROM is due to the quartz window and ceramic package. The OTP is the same exact die, but in plastic package. Cost is low enough that 3-4 OTP chips used to cost as much as one UV-EEPROM. For development I would upgrade to EEPROM, for production I would cost-reduce to OTP. So I would replace M27C2001(256kx8) with a modern OTP such as Atmel/Microchip AT27C020-90PU. – MarkU Feb 19 '21 at 01:47
  • _"What I'm thinking of doing is double-programming them."_ - Which programmer are you using? – Bruce Abbott Feb 19 '21 at 01:56

2 Answers2

1

I don't see any reason not to double program them (I don't think it can hurt much, and it could well help), unless your programmer is "smart" and refuses to start if the 'blank check' fails.

Pro programmers used variation in the Vdd to ensure full programming, if memory (ha) serves. The programmer programs at one Vdd and subsequently verifies at another to ensure there is a certain amount of margin.

Unless the chips have been used in a (horrific flashbacks) development environment where the poor engineers didn't have an emulator they will likely have been erased once at most, even if they are 20 or 30 years old. It's likely that erasing (high intensity UV) and programming (high voltages to cause tunneling) that ages them.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
0

Just from my memory from way back when: EPROMs have a limited but large number of erase cycles; the silicon dioxide around the gates accumulates damage from each cycle, making the chip unreliable after several thousand cycles. EPROM programming is slow compared to other forms of memory. Because higher-density parts have little exposed oxide between the layers of interconnects and gate, ultraviolet erasing becomes less practical for very large memories. Even dust inside the package can prevent some cells from being erased. Programing with the same data should not affect the life. It is the UV that kills them.

Gil
  • 4,951
  • 2
  • 12
  • 14