Improvising An EPROM Eraser

Back in the old days, when we were still twiddling bits with magnetized needles, changing the data on an EPROM wasn’t as simple as shoving it in a programmer. These memory chips were erased with UV light shining through a quartz window onto a silicon die. At the time, there were neat little blacklights in a box sold to erase these chips. There’s little need for these chip erasers now, so how do you erase and program a chip these days? Build your own chip eraser using components that would have blown minds back in the 70s.

[Charles] got his hands on an old 2764 EPROM for a project, but this chip had a problem — there was still data on it. Fortunately, old electronics are highly resistant to abuse, so he pulled out the obvious equipment to erase this chip, a 300 watt tanning lamp. This almost burnt down the house, and after a second round of erasing of six hours under the lamp, there were still unerased bits.

Our ability to generate UV light has improved dramatically over the last fifty years, and [Charles] remembered he had an assortment of LEDs, including a few tiny 5mW UV LEDs. Can five milliwatts do what three hundred watts couldn’t? Yes; the LED had the right frequency to flip a bit, and erasing an EPROM is a function of intensity and time. All you really need to do is shine a LED onto a chip for a few hours.

With this vintage chip erased, [Charles] slapped together an EPROM programmer — with a programming voltage of 21V — out of an ATMega and a bench power supply. It eventually worked, allowing [Charles]’ project, a vintage liquid crystal display, to have the right data using vintage-correct parts.

78 thoughts on “Improvising An EPROM Eraser

  1. Lol I had dug my EPROM eraser out only a couple of days ago and dusted it off to try and cure some UV glue but the wavelength of the eraser was too short for the glue :(

    That’s neat a single led consuming sweet FA power will do what my big metal box does!

      1. Shorter does not always mean better in photochemistry, as light must be absorbed to produce an effect, and if optical absorbance of your particular substance drops at shorter wavelengths, you’re out of luck.

  2. I bought a box off the bay for hardly anything and it only takes a few min to erase a whole batch of chips, but if you only ever need to do this once in a blue moon why not give it a try

  3. These erased with the 253.7nm mercury line. As fluorescent lamps go out of style, the nuisance erasure that we used to deal with is pretty much no risk. Now, how to erase it is the question, and germicidal lamps are the answer. I still use the UV erasable devices on occasion, and a germicide cycle in a UV sterilizer does the job. Nice to know that the UV LED’s can do it as well.

  4. I’ve been told numerous times that the wavelength of UV LEDs was too long to erase EEROMs. I’ll have to try it, I have a couple I need erasing and the worst that could happen is nothing at all.

  5. Those 3W UV flashlights are pretty cheap on eBay, they’re very good for curing UV glue, I guess they would work fine on EPROMs too. But if you don’t have it laying around and you have to wait for 20 days for it to arrive, I guess I’d try some other options. For example asking manicure salon staff to expose chip in UV nail-polish dryer. Or focusing Sun on the chip’s window while trying not to melt it.

  6. An atmega32 and a UV LED blowing minds in the 70s? Hardly. The Motorola 68000, probably the most popular chip to come out of the 70s, has a shitload more performance than an atmega32. While there weren’t UV LEDs in the 70s but there were blue-violet ones and the jump to UV isn’t exactly mind-blowing despite how long it took.

    The price, though, is what would really impress them.

    1. Might also be a bit of exaggeration to say the 68000 was “the most popular chip to come out of the 70s”.

      According to Wikipedia, the 68000 was “introduced” in 1979, but initial samples shipped in February 1980 and first production began in November.

    2. There were no blue or violet LED’s prior to the modern LED revolution created by Nakamura, Akasaki, and Amano in the early 21st century. It was such a significant achievement that they got the Nobel Prize for it in 2014. Before their work the only LED colors were red, green, and the combinatorial yellow, and they were much dimmer and less efficient than the LED’s we have now.

        1. Blue LED’s were never commercially available during the 1980’s, much les the 1970’s. Practical blue LED’s were the holy grail of LED research because with a blue LED you could make white or full-color RGB displays, and that didn’t happen until the modern stuff came about — which it does seem was in the 1990’s, a bit earlier than I realized. But if there were ever blue LED’s before Nakamura they were impractical curiosities that normal people never saw.

          There is a REASON those three guys got the Nobel Prize.

          1. I mentioned in another comment that I paid $10 for my first sample Blue LED in the early 90’s, when I was working as an engineering technician fresh out of college. Based on the employer it was late 1991 or early 1992. So they were available but also very expensive.

        2. In 1972, not much was available. I think the first LEDs I bought were floor sweepings, the dies at an angle, or in reverse, casing that didn’t match the dies wavelength, uncertain output level. I can’t remember when I got those. I know for certain I got generic 7 segment LED readouts surplus in 1974, and by then they were “cheap”. They were still working on making LEDs good, especially in terms of output. You could also get orange but we only dreamed of blue or white.

          Each color was a different process, it took “forever” before blue and white (I recall in that order) arrived.

          We did have fun, about 1975 in high school, asking the physic’s teacher how they worked. He didn’t know.


        3. When I saw “blue” LEDs in the drawer of the electronics shop in the 80ies, they were really IR in a blue casing. The first blue LEDs in the 90ies were horribly expensive and inefficient. They were based on SiC, which is unfortunately an “indirect” Semiconductor which is very poorly suitable for light emission.
          In the 70ies you had red, green and yellow LEDs.

    3. I found a NY Times article from June 1984 (introduction of the 68020 at $487 each!), that says the 68000 was going for around $15 at the time ($35 inflation-adjusted to 2017). An ATMEGA32A-PU is $3, so the order-of-magnitude lower price would have impressed them. The on-board 32KB of flash, 1KB of EEPROM, 2KB of SRAM, 8-channel 10-bit ADC, 3 counter/timers, and 3 serial interfaces would have blown their minds; a CPU and whole board full of memory and peripherals crammed into a 10 mm square package.

  7. Back in the early 1990s, I tried erasing EEPROMs with UV florescent tubes, which I normally used to expose photoresist for PCB etching. I put about two dozen EEPROMs with known code under those lights. None erased after the normal 20 minutes from a regular EEPROM eraser. In fact, all of them were a perfect verify against the original HEX file. So I left them in overnight. Again, no changes. Then I tried leaving them under the UV lights for a whole week. After all the chips still verified, not a single bit in any of those chips changed, I decided to give up on erasing by the UV florescent tubes (which put out plenty of light for PCB photoresist, usually needing less than 60 seconds exposure).

    Maybe today’s UV LEDs put out deeper (shorter) UV wavelengths than those old tubes?

    1. Depends on the tube. The ones that the eproms wre designed for are the 253.7nm mercury line emmitting tubes (long end of UV-C), which are quartz and have no phosphor. Many of the UV fluorescent tubes suitable for curing resist are essentially blacklight tubes. They are UV-A and long UV-B emitting. They will not, in reasonable time, erase an eprom.

    2. I did the opposite: I used a 75W high pressure mercury quartz lamp, which was a converted tanning light, for PCB exposure and occasionally for EPROM erasure.
      The strong smell of ozone and nitrogen oxides the lamp produced made it difficult to understand how anybody would have exposed his skin voluntarily to the radiation of this lamp. The device was from the 1970ies.
      The conversion consisted in removing the IR radiator, which was used as resistive ballast for the lamp and replacing it with two parallel fluorescent tube chokes.

      1. These old Mercury-Lamp + Heater Combos drove the mercury quite hard which increased the temperature and pressure inside to a point where the mercury itself started to absorb a large portion of it’s shorter-wavelength emissions. By driving it with a ballast you dropped the power considerably thereby enabling it to emit harder UV.

  8. The HAD write-up is misleading (presently). In the source article the author hypothesized that his 5mW LED would only take a few hours. But in reality even after 20 hours it had not reset all the bits. Fortunately his application was tolerant of garbage data and he used it anyway. It wasn’t until he bought a 3W LED that he was able to erase everything quickly. Given enough time the small LED should work but this is a detail worth mentioning since it greatly extends the test cycle. I’d say that this qualifies as a Fail Of The Week since things didn’t work as planned but turned out to be educational in the process.

    1. So then that “secret pen” that issues invisible ink at one end and a UV light t the other end isn’t good enough? The description had my hopes up.

      Leaving eproms out in the sun long enough was said to work. Bulbs for killing germs seemed to be a common source of erasing bulbs forty years ago.


    2. I am the author of the article. The 3W UV Led (390nm) I bought off Aliexpress was a total disappointment and didn’t do a good job at all! I abandoned it and eventually erased the EEPROM using two UV Leds, 5mW each in 48 hours.
      I don’t have a good explanation. According to the datsheet of the EEPROM, the wavelength should be under 4000Å (400nm), which it was for both types of Leds.

      1. Aliexpress? Dear Friend:Thanks for your supports and trusts on us. We are always trying to offer customer our best service and high quality products. Your satisfaction is our first priority! Length:390nm ±2mm
        Exquisite workmanship 12 months Warranty. Buy with confidence! If you are not satisfied when you receive your item, please return it within 14 days for a replacement or money back. Please contact me before you return it. If item is defective in 3 months, We will send you a replacement without extra charger, or offer refund after we receive the defective item. If item is defective after 3 months, you can still send it back to us. We will send you a new one after receiving the defective item. But you have to pay the extra shipping fee. Long service life Material:Steel/plastic Color:As the pictures shown Note: Due to the difference between different monitors, the picture may not reflect the actual color of the item. Thank you!

  9. The current commercial BK Precision 851 UV eraser uses a 10W bulb ( I have never encountered any 300W erasers. If they existed, they must have been used to erase chips in seconds, not minutes. The BK manual states ( “When the eraser is working, the power lamp (LED) will light up, the wavelength of the UV lamp operates at 253.7nm (2537 Angstroms). Intel recommends this value for the erasure of their
    EPROM’s.” The shortest wavelength UV LED seems to be 365nm, but 395 or 405nm are more typical.

    1. Probably the 300W bulb was a combined mercury vapor tube and a tungsten filament used as ballast for the discharge tube. So only a fraction of the power went to the UV tube.
      I also can not see how an UVA LED should be able to erase an EPROM. But that’s obsolete for me anway.

  10. You can still buy 254 nm Mercury tube lights cheaply — they are in those surface sanitizers (which you can get for a few bucks) and work great on EPROMS. In my experience, inexpensive UV LEDs are too long-wavelength to do the job. A 280-nm UV LED on Digikey goes for about $150 now.

    1. You can also get a nice UV-C source from old streetlights (the kind that doesn’t use sodium), they have an outer shell coated with phosphor (you gently break that off) and the inside is a mercury arc lamp made from quarts…

      Just watch out for the exposed leads with line voltage ;-)

  11. Back in the ealry 80s I used to use a Manaslu(?) brand geological prospecting / mining light to erase my 2708s and it worked very well. It was a portable unit about the size of a house brick.

  12. For a more traditional solution, I got an eraser for about $15 at

    I recall an article in the mid-70s about rejuvenating eproms. After enough cycles the metal in the junctions would crystalize and it wouldn’t all erase with UV. But if put on a cookie sheet at baked in an oven (ceramic case only obviously) for like 20min it would redo the crystallization and make it usable again. This also erased the eprom.

        1. All the ones with windows that I’ve seen have been in ceramic packages. The only plastic packaged ones that I’ve seen have been windowless (being low-cost OTP versions of the more expensive originals).

        1. I’ve never seen a plastic package EPROM with a quartz window. That doesn’t mean they don’t exist, but as many times as I’ve heard this story, no one has ever posted a picture or a datasheet.

          Have you a link to either?

    1. I have one of these erasers. They are super common, cheap, and has a mains cable that looks like it’s rated for about 12v. Still, it works for my uses but I would not leave it unattended! The fact that it has a clockwork timer in 2018 says a few things about it’s age :-)

  13. I had a PAR UV light (75-150watt) with a heavy ballast, I’ve seen bouncers with their magic invisible markers using said item. Put the chip right under it for hours and no erasure. My boss knows some EE prof at Purdue and I get the real thing, one of the little 4 watt tubular bulbs with a timer in thirty minute intervals. done. Having an instrument of the same model it was easy with a bay sourced programmer and new chips to copy and finish repair. Near lightning strikes tend to erase-short them.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s