Can UV LEDs erase EPROMs?

UV LEDs produce nearly enough nothing at wavelengths much below 350 nm.

There is a 350 nm model with extremely low output power, and otherwise UV LEDs have peak wavelength 365 nm and up. All easily available ones at any reasonable price have a peak wavelength 370-375 nm or more. Spectral content falls rapidly at wavelengths shorter than the peak wavelength.

The usual lamps used to erase EPROMs and kill bacteria do that with 254 nm. I have heard of sunlight erasing EPROMS over several days or a couple weeks in Arizona, and I would guess wavelengths near the shortest getting through the atmosphere in any amount (around or possibly slightly over 300 nm) are responsible. I doubt Arizona sunlight sterilizes toothbrushes, although a car parked in the sun on a bad day in Phoenix can get hot enough to kill a lot of bacteria.

- Don Klipstein ( snipped-for-privacy@misty.com)

An old Intel appnote suggested that the UV dose required for reliable erasure is five times the dose required to erase all the bits.

John

410 is certainly not UV. I've never seen a "UV" led that's wasn't brightly visible; these are actually violet LEDs. I think true (invisible) UV LEDs are being produced now, but they're still very expensive.

John

I tried it. I made am EPROM filled with checkerboard pattern and ran a continuos "test the EPROM" loop running from another, shielded EPROM. Took 4 days before the test failed. After a week it still didn't test on my burner as being fully erased.

They do not produce a single wavelength, but a sort of bell curve (but more output on the long wave end than the short wave end) of wavelengths. They still have lots of output longer than 400 nm where most people perceive a violet color. Color digital cameras capture a quite different shade because their 3 color response curves do not match those of the human eye, perfectly.

--
John Popelish

That is a pretty good correlation to 'several days or a couple of weeks'. In terms of useability, the first failure was what mattered. In the UK (a lot less Sun...), I have known a couple of test chips fail after about six months, in equipment with no attempt to block light. Even when erasing the EEPROM, in a proper eraser, it is very noticeable that some cells, seem to take vastly longer than others. Unless the unit was 'tracking' the Sun, presumably some cells were getting a lot more exposure than others near the shaded edge, and these would take a long time to clear. I doubt if a UV LED, could erase an EPROM in any reasonable time. The light from these LED's, generally has more in common wth the longer wavelengths given by UV flouresescent tubes, than the lamps used in erasers.

Best Wishes

Nichia makes two incredible ultraviolet LED's, one is 365nm @ 100mw and the other is 380nm @ 85mw. Yes thats right, 85 and 100 milliwatts of OPTICAL output power! Both are single die (not arrays). You can place an order and view datasheets at:

Asa Cannell

Hmm, I guess the answer is no then.

There seems to be long UV for apps like making shirts glow in discos, counterfeit note spotting, reading UV pens etc, and short UV for apps like sterilizing stuff or wiping EPROMs.

I notice Maplin in the UK sell some UV LEDs for about £3 (N74AJ, 410nm peak,

500 mcd) and others (N75AJ)for £15.

I assume the fivefold difference is due to power output.

Strangely, the product images show them glowing violet. Surely they should be invisible?

Over a weekend lost two bits in a test 2716 (Summer, 1980, Leeds). Daylight UV up here seems to be getting worse. Maybe time for another test. regards john

410 nm is visible.

Many LEDs with peak wavelength 405 +/- 5 nm are a little common and marketed as "UV", although 405 nm has been considered visible before these LEDs appeared on the market.

LEDs with peak wavelength 395 +/- 5 nm are common, and that wavelength is visible enough to give these LEDs a violet glow. The longer wavelength portion of the spectral band of these is in the "officially" visible range.

LEDs with peak wavelength 375 nm have the longwave "tail" of their emission band with enough content at wavelengths visible enough to make them glow visibly violet. But with those, most of the visible spectral content is broadband and spurious emissions, so these appear violetish white and dimmer than the 395 nm ones.

- Don Klipstein ( snipped-for-privacy@misty.com)

Down Under would be an interesting place to do this test, what with NZ being under the hole in the ozone layer. In summer burn time is 5-10 minutes. Roll on the SPF3000....

I'd do it myself but I no longer have eproms, or programmers. I do have an old eraser though.

Cheers Terry

Cameras can see things people can't. A digital camera makes a great near-IR viewer.

John

Witness the embarrassing "I can see through your clothes" debacle recently - I forget whose camera it was, one of the big Japanese firms, but in IR mode it could see through clothes.

Cheers Terry

I saw it on the news with pictures of the effect. I concluded that it was a lot of imagination plus reporters looking for something sensational - I just saw the clothes.

This was several years ago, and the cameras (or were they camcorders?) were Sony Nightshot if I remember correctly.

Some fabrics are less opaque to infrared than to visible for two reasons:

1. Most dyes are transparent to infrared. Most clothes will look much lighter to an infrared-seeing camera than to human vision in visible light.
2. Some fibers have some fine texture that is shorter than the infrared wavelengths but longer than visible wavelengths. Some fibers are less opaque and more translucent at some IR wavelengths than at visible wavelengths.

As for how much of a difference this second factor makes? I imagine sometimes nothing, sometimes about as much as wetting the fabric or similar "weight" fabric does, rarely much more.

Thermal radiation from the human body has nothing to do with these cameras, since they only detect infrared wavelengths out to 1.2 micrometers or so - whatever the limit is for silicon photojunctions.

When this became news, I web searched for photos showing the effect. I saw one somewhere showing clothing translucent enough to allow underwear to show, and nothing showed through the underwear. This was lightweight clothing that appeared barely able to hide underwear in visible light (a visible light photo of the same person wearing the same clothes was also shown). And I believe the photo was of a more extreme case rather than of what usually happened.

They said that because of this, Sony added an infrared-blocking filter inside these cameras that automatically got in the way when the camera was not specifically in an "infrared night vision mode".

- Don Klipstein ( snipped-for-privacy@misty.com)

The RL5-UV2030 UV LED

starts emitting miniscule power at 375nm, peak at 405nm, with a whopping 17.802mW absolute irradiance, so it should take (15w/17.802mW)= 843 of thse UV LED's to equal 15W output... that's using (20mA*843=16.86a*3.5v)= 59W of power. And we thought UV lamps got hot!

According to

, 405nm is too low a frequency for photoresist imaging. BUT, according to old hand Harry Lythall (SM0VPO, "light from about

390nm to 450nm is ideal for making PCBs." Who knows?

So... I suppose it might be possible to expose boards with UV LED's, but unless there's a whole lot of them, it's gonna take a *h e l l u v a* long time. Boards that would be ready in 5 minutes using 15w of lamp power would take a hundred UV LED's 42.13 minutes to accomplish.

100 UV LED's would cost $120 from superbrightleds.com, where a decent UV lamp on ebay can go for what, about $30?

Maybe one could find a bargain at

, but still the idea sounds iffy. I happen to have a bunch of UV LED's laying around and want to try photoprinting a few PCB's, so maybe I'll build a small lightbox and post the results.

-M

Don't tell me..You watched the mythbusters episode where they grow fecal coliform on toothbrushes kept in the bathroom?