Talk:EPROM

	This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing articles
???	This article has not yet received a rating on the project's importance scale.
	This article is supported by Computer hardware task force.

Is the table in this article up to date? I can't believe the largest EPROM is only 512kb. sjorford →•← 10:02, 21 Mar 2005 (UTC)

The table isn't really inclusive. It only lists one family of devices. There are larger ones, but I've never heard of one > 20 megs. Certainly, the common ones are all pretty small. - Bryce —Preceding unsigned comment added by 66.81.40.136 (talk • contribs)

First EPROM

The first EPROM was the 1702 (256x8 = 2k memory) used to store the firmware or boostrap of the first microcomputers or programmable calculators. —Preceding unsigned comment added by Cl barbot (talk • contribs)

how does it work

How does an EPROM work? I think that there is missing that part. Explanation of EPROM memory cell work.--User:Vanished user 8ij3r8jwefi 16:28, 2 February 2006 (UTC)[reply]

I don't know the details but i belive there is a "floating gate" on the transistors which can be charged using a higher voltage than that used for normal operation and discharged using UV light. Plugwash 01:26, 8 July 2006 (UTC)[reply]

Erasure

It says it can be programmed electrically, but can only be erased by UV light. Does that mean that once the chip has been programmed, you are unable to overwrite the settings with electrical signals? Does the chip need to be erased via UV light to change any programming?

Yes, EEPROMS are the only ones that can be erased electrically —Preceding unsigned comment added by 67.177.254.247 (talk • contribs)

Not quite, when erased an EPROM will go into a blank state (all zeros or all ones depending on its design), you can then electrically flip bits out of the blank state but you can't flip them back again. Plugwash 01:28, 8 July 2006 (UTC)[reply]

"The recommended erasure procedure is exposure to UV light at 253.7 nm of at least 15 W-sec/cm² for 20 to 30 minutes, with the lamp at a distance of about 2.5 cm"
W.sec = J is a unit of energy. Shouldn't that be a unit of power instead? (i.e. W instead of Ws or J - BTW I assume sec stands for seconds... the official abbreviation of second is s, not sec)Sclaes (talk) 20:26, 23 June 2015 (UTC)[reply]

Electrically programmable rom?

Please note this: I read in some books that EPROM means Electrically Programmable ROM, and not Eraseable Programmable ROM. If other users confirm this, please correct the article. —Preceding unsigned comment added by 151.56.233.30 (talk • contribs)

which books? i've certainly never seen this before and proms are also programmed electrically. Plugwash 01:17, 8 July 2006 (UTC)[reply]

Plugwash is correct for microchip based PROMs. Before the advent of the microchip, PROMs existed that had to be manually programmed with a stylus or even the insertion of a punched Hollerith card. And there are (or, were) microchip PROMs programmed by means of laser light. —Preceding unsigned comment added by 71.195.85.16 (talk) 23:56, 6 June 2010 (UTC)[reply]

What does the UV actually do?

I understand the basics of the programming side, the floating gate, but how exactly does the UV erase the data? Maury 12:16, 1 August 2006 (UTC)[reply]

Einstein worked out that light can drive electrons off metal surfaces depending on factors like "work function" and wavelength. In the case of the crystalline lattice of a microchip, ultraviolet light (or x-rays) can easily drive electrons from a metal surface to a lower potential point, thus erasing the cell by UV induced photoelectric current. The UV light makes the crystalline structure conduct electricity by ionizing it as well as causing electron emission from the floating gate. —Preceding unsigned comment added by 71.195.85.16 (talk) 00:01, 7 June 2010 (UTC)[reply]

Re-programmable EPROM?

Can I reprogram the EPROM after erasing it or how many time could I reprogram it?

UV erasable chips tend to store data for a minumum of 10 years with estimates that the typical cell will last 50 years (holding a charge). In the case of Floating Gate cells, whether UV EPROM or EEPROM, the insulator under the floating gates is subject to (relatively) ENORMOUS electrical forces. This dmages the insulator, which eventually fails. In the case of EEPROMS, write cycles started at 1000 min and have been expanded to 1 million or even 10 million, in the case of chips used in Solid State Drives. For EPROMs, this limit was rarely approached, but we did some tests on chips from 2701 (1K low voltage programmed version of the 1701, which needed 55VDC) to 2732 memories for NASA (See Data Control Systems, where I was the microchip maven). They would fail at between 1000 uses to 5000 uses. Failure modes were, predictably, insulator punch through. —Preceding unsigned comment added by 71.195.85.16 (talk) 00:10, 7 June 2010 (UTC)[reply]

data table submission from anon editor

the following was tacked on to the end of the article outside of the rest..

programming.
klu
EPROM Type Size — bits Size — bytes Length (hex) Last address (hex) 
1702, 1702A 2 Kbit 256 100 000FF 
2704 4 Kbit 512 200 001FF 
2708 8 Kbit 1 KB 400 003FF 
2716, 27C16 16 Kbit 2 KB 800 007FF KKESIT 
2732, 27C32 32 Kbit 4 KB 1000 00FFF 
2764, 27C64 64 Kbit 8 KB 2000 01FFF 
27128, 27C128 128 Kbit 16 KB 4000 03FFF 
27256, 27C256 256 Kbit 32 KB 8000 07FFF 
27512, 27C512 512 Kbit 64 KB 10000 0FFFF 
27C010, 27C100 1 Mbit 128 KB 20000 1FFFF 
27C020 2 Mbit 256 KB 40000 3FFFF 
27C040 4 Mbit 512 KB 80000 7FFFF 
27C080 8 Mbit 1 MB 100000 FFFFF

If I stick it into a wikitable it looks like this:

EPROM Type	Size — bits	Size — bytes	Length (hex)	Last address (hex)
1702, 1702A	2 Kbit	256	100	000FF
2704	4 Kbit	512	200	001FF
2708	8 Kbit	1 KB	400	003FF
2716, 27C16	16 Kbit	2 KB	800	007FF
2732, 27C32	32 Kbit	4 KB	1000	00FFF
2764, 27C64	64 Kbit	8 KB	2000	01FFF
27128, 27C128	128 Kbit	16 KB	4000	03FFF
27256, 27C256	256 Kbit	32 KB	8000	07FFF
27512, 27C512	512 Kbit	64 KB	10000	0FFFF
27C010, 27C100	1 Mbit	128 KB	20000	1FFFF
27C020	2 Mbit	256 KB	40000	3FFFF
27C040	4 Mbit	512 KB	80000	7FFFF
27C080	8 Mbit	1 MB	100000	FFFFF

I have no idea where this should go in the article. DMahalko (talk) 01:34, 31 March 2009 (UTC)[reply]

How does it work

I've added some theory of the devices here, but it still needs work (a schematic of a memory array would be good, and the floating-gate transistor picture doesn't show the channel or the connections to the control gate). There's always an exception - I just wrote that EPROMs weren't usually erased in-circuit, but many years ago I knew of an Automate 35 programmable controller made by Reliance Electric that came with a custom UV lamp to hang on the side of the card-cage, for erasing program memory. If I can find a reference, I'll list that as an exception -but it was a pretty clumsy way to change programs. --Wtshymanski (talk) 15:55, 22 October 2009 (UTC)[reply]

We have a large Stag EPROM eraser that is capable of erasing EPROM chips while they are still mounted on boards. It hold 2 dual VME sized boards at a go. It has 8 UV tubes in it. 20.133.0.13 (talk) 15:05, 25 November 2009 (UTC)[reply]

Must be a monster, though enormously convenient and easy on EPROM chips. Not the sort of thing you can drop in your field service toolbox, though. --Wtshymanski (talk) 19:49, 25 November 2009 (UTC)[reply]

Paper labels on UV windows?

In years past I saw many an EPROM with a paper label over the erasure window. Is this simply a matter of convenience for labeling of custom EPROMs or does the paper actually have some UV protection property for the chip? I would assume adhesive foil would be more appropriate as "erasure protection", but perhaps a simple paper label is sufficient to protect it? -- DMahalko (talk) 10:31, 14 December 2014 (UTC)[reply]

I don't know how reliable the information was, but if full erasure of a chip occurs in 7 minutes from a X strength UV lamp at 3 cm, it stands to reason that exposure to florescent light fixture 200 cm away for hours to months would affect a few bits. For this concern we always religiously covered the windows. —EncMstr (talk) 17:36, 14 December 2014 (UTC)[reply]

It served both purposes -- to show that it was no longer a blank EPROM, to show the version number of the firmware that had been programmed into it, and to protect the silicon from UV exposure. Depending on the type of label, it either blocked UV well or didn't do much at all. Cheaper, thinner labels obviously let more light through. However, sunlight exposure would still take a long time to damage the data, especially with a label on it, and inside a case as most ICs are the possibility of damage was essentially non-existent. The label, I assume, was mostly for protection during shipping and testing. Most EPROMs were used for development and debugging. After a firmware was complete, they would then use the much cheaper PROM or windowless EPROMs to permanently program it (much of the cost of an EPROM is from the quartz glass and the side-brazed ceramic DIP that they often are packaged in). Air♠Combat_Talk! 22:18, 7 October 2018 (UTC)[reply]

Confusing sentence

"The EPROM was invented by Dov Frohman of Intel in 1971, who was awarded US patent 3660819[1] in 1968."

I just edited this to show the patent was granted in 1972, which the Google Patent reference shows. However there is no citation that it was invented in 1971; in fact, the existence of commercially available EPROMs in 1971 like the one pictured makes me very dubious that it was invented in that year. Perhaps the dates were switched, and it was *invented* in 1968 but given a patent in 1971? This doesn't make much sense either, since it seems the patent was submitted in 1970 and granted in 1972. Anyone with more info or a better source is encouraged to add more info! Air♠Combat_Talk! 22:23, 7 October 2018 (UTC)[reply]