There are numerous experiments to erase UV EPROM, but I have not read about experiments to upset UV EPROM contents with light. Has anyone done it in a quantitative way? I’m thinking of using a 6502 to drive a WS2812B LED which has 3 colors and 256 levels of intensity to find out how much light is needed to upset UV EPROM. Curious to know if that has been done already.
Bill

I remember Jeff doing some impromptu experiments and intermittent bug hunting with uncovered EPROMs and ambient lighting, but I'm link challenged at the moment.

_________________
Got a kilobyte lying fallow in your 65xx's memory map? Sprinkle some VTL02C on it and see how it grows on you!

Mike B. (about me) (learning how to github)

@revaldinho looked into sunlight erasing (a particularly sunny June, southern England) - he found 3 eproms, different capacity and different manufacture, and found erase times varied between a few days and a few weeks. He also found this link:
https://goughlui.com/2016/09/11/experim ... -lamp-sun/

I feel this is an area where we can roll out the old adage "all electronics is analogue electronics" and perhaps add "especially memories". The charge on a bit cell is comprised of some large number of trapped electrons, and there's a sense amp to apply a threshold as to whether it is or is not enough charge to count. That threshold will be variable and noisy, and the true state of the cell could vary between fully discharged, just barely under the threshold, just barely over the threshold, all the way to fully charged.

So, one might well learn some interesting things, but the usual tactics apply: charge at least two or three times longer than it takes to cross the threshold, and erase two or three times longer than it takes to get an apparently empty ROM.

And also, there's the idea that you can't prove a negative: you can show that some amount of light has had an effect in some particular experiment, but you can't show that some smaller amount of light will never cause anyone to suffer a head-scratching debugging puzzle.

Edit: an overview of the development of the technology here (may need a proxy to read it.)

Found it - it wasn't erasure in this case, but photoelectric disturbance... Here, here and here.

Yes, and all three of these posts refer to the same incident (involving the EPROM-based 68HC705 microcontroller).

There was one other EPROM incident, but I'm presently unable to find a post about it. Maybe no such post exists, so...

Back in the early or mid 80s I was troubleshooting my expanded KIM-1; I think it might've been after a certain accident that sent 110VAC into the 5V logic (ouch!!). These and many other details are foggy after all this time. But as a "logic probe" of sorts, I had a couple of 4029 counters whose divided-down output fed into a small loudspeaker. (The human ear is astonishingly sensitive, and I'd successfully used similar tricks in the past.)

I don't recall which KIM-1 signal I was probing, but the speaker was producing a steady, audible tone... Steady, that is, until I happened to move my hand near the KIM, and that's when the pitch (ie, frequency) suddenly skyrocketed.

Through some fluke, what I seemingly had in front of me on my workbench was a Theremin!

At first I thought the frequency change was due to the capacitance of my hand approaching the circuitry, but that theory was quickly disproven. In fact the change was due to the shadow of my hand falling on the (unprotected) window of a 2716 EPROM. To this day I can't imagine how light got translated into frequency, but I promise I'm not making this up. And it was a very, very sick computer, with many damaged chips.

-- Jeff

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

I prototype a UV EPROM board to do the upset-UV-EPROM-with-light experiment. The CPLD isolates the UV EPROM from the 6502 bus and provide necessary wait state for EPROM. The CPLD will also provide the WS2812B timing. CRC65 65C02 SBC is the CPU testing the UV EPROM board. I'll program UV EPROM with $FF, $00, $55, $AA, and other patterns. I will drive WS2812B with different color and intensity shining into UV EPROM's quartz window while reading EPROM's values with the SBC. I'm starting with Intel's 2764A-3 but I also have few other makes/types of UV EPROM on hand. My speculation is UV EPROM's sensitivity to light do vary by makes and types; it may also varies by how deeply erased or programmed the part was. It seems logical that as the part aged and more electrons escaped from the floating gate that less light is needed to upset the programmed values. This may provide a qualitative way to assess how close an existing UV EPROM is to 'bit rot'.
Bill