Why do EPROM dies?

[Uzumaki]

Every now and then, an EPROM dies. Usually they read all zero and won't get erased no matter how long under the UV lamp, or are all stuck at FF and fails at programming at the very first byte.

 

But why do they fail? I've had to toss out 2x 2732 EPROMs that wouldn't get erased. I have a pile of about 30 EPROMs that died in the last 2 months including a couple 2 mega and one 4 mega.

[Rybags]

Unsure. Don't they require fairly specific voltages when burning them?

[jsoper]

Because the first time a baby laughs, an eprom is born. But when the baby gets older and says "I don't believe in eproms" it drops dead. JM Barrie explained it all in Peter Pan.

[Zerosquare]

Every now and then, an EPROM dies. Usually they read all zero and won't get erased no matter how long under the UV lamp, or are all stuck at FF and fails at programming at the very first byte.

 

But why do they fail? I've had to toss out 2x 2732 EPROMs that wouldn't get erased. I have a pile of about 30 EPROMs that died in the last 2 months including a couple 2 mega and one 4 mega.

Are you sure you're using the right lamp ? The wavelength of UV rays matter ; the EPROMs will erase slowly or not at all if it's not correct (I think you can even damage them with too long an exposure to the wrong wavelength, it happened to somebody I know). I know different kinds of EPROMs have different sensitivity to UV light, maybe the wavelengths that "work" are also different. Edited November 11, 2006 by Zerosquare

[Uzumaki]

The lamp I'm using provides UV-C and was listed as a germicidal. It has worked for probably close to 500 EPROM I've erased in the past few years.

 

There's also quarts bulb but it's frickin' expensive to replace. And some quartz bulb are filtered, finding the right one can be a challenge itself. Compare to $10 12" unfiltered fluorescent bulb...

[PressureCooker2600]

Because the first time a baby laughs, an eprom is born. But when the baby gets older and says "I don't believe in eproms" it drops dead. JM Barrie explained it all in Peter Pan.

 

lol...that couldn't have been said any better

 

on a serious note....on average, how expensive do ROM burners cost....are they hard to operate

[supercat]

But why do they fail? I've had to toss out 2x 2732 EPROMs that wouldn't get erased. I have a pile of about 30 EPROMs that died in the last 2 months including a couple 2 mega and one 4 mega.

 

Some EPROMs are rather touchy. Different chips require different voltages and programming times. Use of an incorrect voltage can easily fry a chip rendering it completely useless.

[Bruce Tomlin]

Because the first time a baby laughs, an eprom is born. But when the baby gets older and says "I don't believe in eproms" it drops dead. JM Barrie explained it all in Peter Pan.

 

lol...that couldn't have been said any better

 

on a serious note....on average, how expensive do ROM burners cost....are they hard to operate

About $50 on ebay for a good one.

 

http://cq-search.ebay.com/search/search.dl...+Needham+EMP%29

 

Note: the PLD-1128 doesn't burn EPROMs, and the EMP-10 only works in DOS/W95/W98.

[supercat]

Note: the PLD-1128 doesn't burn EPROMs, and the EMP-10 only works in DOS/W95/W98.

 

Note that many parallel-port burners may damage chips if they're used with a multi-tasking operating system. Under W95/W98 they'll "usually" work, but if Windows decides to steal some CPU time in the middle of a programming cycle, a memory cell may get zapped.

[LiQuiD iCe 9]

i'm no expert but i've been experimenting with the 2600 this summer trying to improve the video mod and power useage. want to go portabel. I would check your volt regulator on the system to see if its putting out 5 volts. the carts run on the 5volt line if they get hit with to much above 5volts they'll decay over time and you said more than one has died on you.

 

i also found that if you replace L200 on 2600 motherboard(5v to cart next to cart slot) with a 10 ohm resistor(i used 1/4 watt 1%) the game only uses what it needs to work, i think. I get different voltages and amps with different games instead of the same 5v. Boxing only needs about 3 volts it works with a 100ohm reisitor in place, none of my other games would do that but all 35ish games of mine worked fine with the 10 ohm in place.

i thought of this based on modern memory getting worn out by overclocking the voltage witch blasts away electrons of the pathways killing the memory in a decaying manner. another way to tell if the voltage is to high is brighter picture and louder sound and more heat but i don't know how relyable that is.

 

i dont actually know but i think hitting the carts with less voltage will improve their life cycle.

 

this post is dedicated to my only copy of MS PACKMAN who unwillingly gave her life to further our knowledge and inspired me to find a cure for over volting carts everywhere

[+batari]

Note: the PLD-1128 doesn't burn EPROMs, and the EMP-10 only works in DOS/W95/W98.

 

Note that many parallel-port burners may damage chips if they're used with a multi-tasking operating system. Under W95/W98 they'll "usually" work, but if Windows decides to steal some CPU time in the middle of a programming cycle, a memory cell may get zapped.

Ironically, many of the parallel port burners also don't work properly on faster systems. My old EPROM programmer would not work with anything faster than a Pentium 100.

[Uzumaki]

Ironically, many of the parallel port burners also don't work properly on faster systems. My old EPROM programmer would not work with anything faster than a Pentium 100.

 

I guess it depends on the burner and the software. Willem burner I have are parallel port and I haven't had any misburn while multitasking. I had to disable XP's port polling so XP doesn't interfere with the burning software when it checks for any PnP devices.

 

I haven't had experience with many other burners though. Before the Willem burner, I used to have Needhams Emp-10 and it required an ISA card slot so you're limited to first generation Pentium or older PC. The software is DOS only and while it could run under Windows, I was advised to boot in DOS mode anyway. I doubt newer Windows would work on these PCs anyway.

[supercat]

I guess it depends on the burner and the software. Willem burner I have are parallel port and I haven't had any misburn while multitasking. I had to disable XP's port polling so XP doesn't interfere with the burning software when it checks for any PnP devices.

 

Some burners have circuitry to generate write-pulse timing. The PC outputs a signal that says "feed a pulse" and the burner will feed out a proper-length pulse. Such burners won't have trouble with multitasking PCs.

 

Other burners let the PC generate the timing. The PC will turn on the output, run a delay loop, and then turn off the output. If the PC gets interrupted while the output is on, it will remain on for longer than intended.

[Bruce Tomlin]

Note: the PLD-1128 doesn't burn EPROMs, and the EMP-10 only works in DOS/W95/W98.

 

Note that many parallel-port burners may damage chips if they're used with a multi-tasking operating system. Under W95/W98 they'll "usually" work, but if Windows decides to steal some CPU time in the middle of a programming cycle, a memory cell may get zapped.

"many parallel-port burners" means the super-cheapies like Willem. The BP and Needhams units have their own CPU and don't depend on the parallel port speed for timing.

 

I haven't had experience with many other burners though. Before the Willem burner, I used to have Needhams Emp-10 and it required an ISA card slot so you're limited to first generation Pentium or older PC. The software is DOS only and while it could run under Windows, I was advised to boot in DOS mode anyway. I doubt newer Windows would work on these PCs anyway.

Look at the picture, the EMP-10 is a parallel-port burner, you must have had a PB-10:

http://needhams.com/products-legacy.html

 

Still, it was never supported under NT/2K/XT. The EMP-20 is barely supported with a driver that lets you burn in full-screen DOS mode only. (obligatory rant: whoever decided that DOS-mode access to I/O ports in Windows NT and above should be determined by the freaking SCREEN MODE should be strung up)

 

My first burner was an EE-Tools ChipMax, which was an ISA card. They were still selling them at Fry's as recently as last year.

 

Now if only they can upgrade Parallels Desktop to full USB 2.0 support, then I'll be able to use my recently acquired EMP-31 on my MacBook Pro.

Edited November 19, 2006 by Bruce Tomlin

[dinosaur]

Ironically, many of the parallel port burners also don't work properly on faster systems. My old EPROM programmer would not work with anything faster than a Pentium 100.

 

I guess it depends on the burner and the software. Willem burner I have are parallel port and I haven't had any misburn while multitasking. I had to disable XP's port polling so XP doesn't interfere with the burning software when it checks for any PnP devices.

 

I haven't had experience with many other burners though. Before the Willem burner, I used to have Needhams Emp-10 and it required an ISA card slot so you're limited to first generation Pentium or older PC. The software is DOS only and while it could run under Windows, I was advised to boot in DOS mode anyway. I doubt newer Windows would work on these PCs anyway.

 

Do you still have the EMP-10 and would you like to sell it?

[Shawn]

Eproms die when the legs get snapped off.

[Uzumaki]

Do you still have the EMP-10 and would you like to sell it?

 

Sorry no. It (along with the 386 PC it was installed in) was left behind at one of the previous place of employment as they used it a lot. I don't know if they still used it but last I checked, they still used it to burn PC programs to install on PC/104 for the industrial machines. EPROM based DOS was defiantly faster than hard drive or floppy drive loading.

 

Eproms die when the legs get snapped off.

 

Heh that'd work. I do have 6 or 7 2Mbit EPROM that refused to be erased blank even after it's been under UV light for several hours. It turns out most EPROM have a typical life of about 100 erases so some EPROM do die from being used too much.

[+Allan]

Here's my new baby. It was a Christmas gift to myself. I just got it tonight. Works like a charm. It was a lot but well worth it.

 

Allan

[classics]

Here's my new baby. It was a Christmas gift to myself. I just got it tonight. Works like a charm. It was a lot but well worth it.

 

Allan

 

Nice basic programmer. I used to use the parallel port version, and the pc-card version before that.

 

Steve

[ijor]

Some burners have circuitry to generate write-pulse timing. The PC outputs a signal that says "feed a pulse" and the burner will feed out a proper-length pulse...

Other burners let the PC generate the timing. The PC will turn on the output, run a delay loop, and then turn off the output.

 

Interesting. I can understand the saving in not including a microcontroller, but saving the small circuit that is required for generating a pulse seems crazy.

 

OTOH, some modern DEBUG/OCD cables works exactly like the latters. The PC needs to generate the whole timing for the JTAG port. And they aren't cheap! The only difference is that JTAG ports can usually work as slowly as you want.

[supercat]

Interesting. I can understand the saving in not including a microcontroller, but saving the small circuit that is required for generating a pulse seems crazy.

 

Not unreasonable when you consider that: (1) for awhile, different chips would require a lot of different pulse lengths; (2) a DOS-based PC could generate pulses more accurately than any reasonably-priced hardware circuit.

 

Suppose you built a programmer with a precision circuit for generating 100us pulses. Then someone comes up with a chip that needs 50us pulses. What do you do?

[+batari]

OTOH, some modern DEBUG/OCD cables works exactly like the latters. The PC needs to generate the whole timing for the JTAG port. And they aren't cheap! The only difference is that JTAG ports can usually work as slowly as you want.

There's no good reason why most of the parallel port JTAG cables are so expensive. I cracked one open last year and all I found was two 74244 buffers and a few resistors, caps and maybe a diode or two.

[ijor]

Not unreasonable when you consider that: (1) for awhile, different chips would require a lot of different pulse lengths; (2) a DOS-based PC could generate pulses more accurately than any reasonably-priced hardware circuit.

 

I guess this depends on the period we are talking. Yes, at that time the hardware required could be not so cheap.

[ijor]

There's no good reason why most of the parallel port JTAG cables are so expensive. I cracked one open last year and all I found was two 74244 buffers and a few resistors, caps and maybe a diode or two.

 

And is there any technical reason nowadays for USB ones to be as expensive as they are?

 

The only reason is the market. If there were a massive market for them, even very powerful ones would cost a small fraction that the current price.

 

Try to get a serial PCMCIA card based on a plain 16550AF compatible UART. They cost much more than a wireless card! Again, it is just the different market.

Edited January 6, 2007 by ijor

[+batari]

There's no good reason why most of the parallel port JTAG cables are so expensive. I cracked one open last year and all I found was two 74244 buffers and a few resistors, caps and maybe a diode or two.

 

And is there any technical reason nowadays for USB ones to be as expensive as they are?

 

The only reason is the market. If there were a massive market for them, even very powerful ones would cost a small fraction that the current price.

 

Try to get a serial PCMCIA card based on a plain 16550AF compatible UART. They cost much more than a wireless card! Again, it is just the different market.

I agree the market is smaller, i.e. maybe 10,000 units are sold, but I still balk at the price. I mean, the entry-level USB units I looked at were around $1000 each, and the top end models were running upwards of $4000, and the hardware is still quite simple - consisting of a USB controller chip and some glue logic, and some simple software drivers.

 

Companies will pay the price merely because it might be cheaper to buy even a dozen or so rather than develop their own for in-house use. (Though at those prices, it would be really close!)