OT: Electronic question, eproms and such.

JL Sargent · Oct 23, 2006

I have a data collector that collects spectrum and other vibration data. It has been in a drawer for approx. 10yrs. before I got it. It has a problem. When I install a new 7.2v battery screen turns on for a sec and looks good but then flashes off, back on, then off repeatedly until I remove the battery. Ocassionally when I can install new battery without the unit turning on the enter button will turn on unit and that isn't supposed to happen. I suspect the two eproms are bad, but I dunno. Now I do have another one of these units in good working condition.

My questions are: How long does an eprom hold its memory? Whats the chance of me reading the eproms from the good unit and rewriting to the not working units eproms? Also I notice the working unit has a newer release of the eproms. Any suggestions, guesses or comments?

azsort · Oct 23, 2006

Eproms are nonvolatile and supposed to last for ever however they are also very dense circuitry with small geometries and are a little prone to failure - especially static damage if you handled it without being grounded. This can happen even if it never left the circuit board if it had data lines going off the PCB. Also, if the EPROM window wasn't covered it will become erased over periods of time in just ambient light levels. All in all, I doubt the Eprom is your problem. Get out your eagle-eye and look for bad solder joints.

Evan · Oct 23, 2006

Actually, eproms most definitely don't last forever. Depending on the type and the technology used they can be good for as short as 10 to 15 years or more than 25. The older they are in general the shorter the data retention.

If the data in the old eproms has faded with age it doesn't mean the eprom is bad. The trapped charge in the gate structure leaks off with time and if it is rewritten it should be good again for another equal length of time.

lenny baledge · Oct 23, 2006

Like Azsort said, the windowed versions will lose a few bits over time if the window is exposed. If your eprom were bad, you most likely would not have an "occasional" problem. The code would crash every time it booted up. If you want, I can read the eproms & program you a new set (or verify the suspect ones for that matter).

It sounds like more of a power supply issue to me. Most processor circuits have a power supervisor chip that pulls reset on the processor when the voltage level dips below a pre-set value. Also, electrolytic caps get crappier and crappier over time. (these are smoothing ripple in your power supply circuit - the same ripple that might be tripping the supervisor) I'm not sure of your tools available, but if you have an oscope, you can look at the power pins and tell if it is clean. If you don't have a scope, swapping in new caps is pretty easy & they probably need it anyway depending on well it was designed. A good design will specify a capacitor with double the working voltage on your circuit. For instance, if the circuit is 5V, you would specify a 10V or 16V cap. This isn't for voltage spikes, but to combat capacitor aging. I have seen designs that don't follow this though & they will not last as long in the field. 10years would be pushing it if the design was mediocre to begin with.

Let me know if you need any more help or parts, etc. lenny *AT* bitworks-inc.com

Lenny

3t3d · Oct 23, 2006

I beleieve that even with the windows covered, the eprom does have a limited data retention life span.
Someone should be able to read the good eproms, and make new copies for you.
If you are brave, you could also swap the good eproms into the other unit, and see if it works.

at that point you will know if the eproms will fix the bad unit. After swapping them back you will know of the bad unit destroys eproms.

Maybe best to check the power supplies first on the bad unit. Then try the swap.

Pete

JL Sargent · Oct 23, 2006

Well, I have a scope so I'll give looking at the power board test pins a try in the morning. Best I can tell this unit has a board dedicated to power. Ill also post a picture of the board in the morning. Thanks for all the help.

lenny baledge · Oct 23, 2006

"turns on for a sec and looks good but then flashes off, back on, then off repeatedly until I remove the battery."

This sounds suspiciously like the processor is resetting over and over again. Could be:
1. power supply issues like above
2. Bad code such as the eprom flakey. (Try the good set to see if it changes anything)
3. Poor eprom connection in socket (try removing/reseating)
4. Any other numerous connections flakey. (Inspect for cold solder joints.

Lenny

Evan · Oct 23, 2006

What type of eproms does it have and what are the date codes? Note: The date code will be a 4 digit number on the chip such as 2387 or 1492 etc. The first two numbers are the week of manufacture and the second two are the year.

JST · Oct 24, 2006

I second the capacitor possibility....

If it has ANY "solid tantalum" types, and it is 15+ years old, they very well may be bad.... typically shorted.

They are the ones that look like a large seed, often about 1/8" to 0.2" diameter, dipped in epoxy.

If bad, they might be causing any sort of troubles. It's a known problem.

If not that, and it really IS an EPROM ssue, there are two types of failure.

1) a "hard" failure.... the data is gone in some spot, and whenever it hits that piece of code, it "heads for the weeds". Repeatable failure.

2) A "soft" failure.... it is GOING bad, but isn't quite there.... doesn't always read right, depending on PS voltage at that instant, etc. In that case, it may work and then flip "into the weeds" unpredictably.

Or, it could be some other thing, like a bad solder joint.... could be temp sensitive, or not, might be vibration sensitive. Might be very predictable and imitate a solidly bad EPROM.

I don't think you can be sure, although it is possible that the EPROMs are losing it.

I have 10+ year old stuff that has yet to lose its memorized data.

Actually I have never seen a real EPROM leak-off failure, even though I have built a goodly amount of eprom based stuff. But I am perfectly certain that they do occur.

dennh · Oct 24, 2006

One more bit of potential bad news. While you're inside, check for any Dallas Semiconductor non-volatile memory products. They will look like a top-heavy DIP IC package and could be 40 pin (processor), 24 - 28 pin (SRAM or clock).

These devices have internal lithium cells, rated for 10 years but frequently lasting less due to storage temperature and other factors. They can hold everything from actual operating code to calibration constants without which some products become mediocre night lights.

Den

Evan · Oct 24, 2006

It's mostly the really old eproms that have fairly short data retention such as the original 27XX series. Those were only guaranteed for 10 years. They have improved dramatically since then with many types now rated for 100 years.

When the floating gate loses it's charge they begin to go flaky at first. The gate charge is just on the verge of switching/not-switching the memory cell and it may read ok one time and not another.

If it is that type (or same vintage, 70s and early 80s) they can be reprogrammed and to ensure that the programming "holds" it's a good idea to program 2 or 3 times with the same data. That makes sure the gate is fully charged.

Note also that a lot of eproms were made with no erase window. They were intended for short run production where it wasn't cost effective to use a mask rom. The windowless types were a lot cheaper. If it has no window you can still reprogram with the same data, as long as it really is exactly the same.

abarnsley · Oct 24, 2006

Don't forget to check for a stuck button or collapsed pcb film switch

wippin' boy · Oct 24, 2006

don't forget to spit in the back
and give it a big womp

gar · Oct 24, 2006

061024-0501 EST USA

JL Sargent:

Evan has given you good information on the eproms. I have units programmed in the mid 1970's that still have valid data. As soon as these devices became available I started using them. Some were not very good others have lasted a long time.

Visible light has virtually no capability to discharge the stored charge, however leakage resistance does.

You should copy the old roms several times to disk memory at each of the following voltages for Vcc --- 5 V, then at 4.8 V, 4.6 V, and last at 5.2 V. compare the files. Any differences would imply the stored data in the eproms was marginal. I suspect you do not have multiple voltage eproms.

If all the results are identical then there is a hard failure or no failure.

From the saved data you can program new eproms.

From your other unit make copies of the data in its roms, then program new roms and try these in your old device. May or may not work because the hardware may be different between the two units.

.

JL Sargent · Oct 24, 2006

Here is what Im working with

Anybody recognise this type? Test points on powerboard all indicate correct clean voltage except one thats labels -vldc in the upper right hand corner. It stays flat at ground.

JL Sargent · Oct 24, 2006

Update: Spent the last hour or so swapping out the proms with others I found from long ago updates. With new proms installed, exact same activity persists. So Im thinking thats not the problem. Power management my main suspect now. Any suggestions appreciated.

Using an o-scope with the two units laying side by side, maybe a test point to test point comparision should be my next step. HELP! Now if I kill the good box tring to revive the old box, that wouldn't be too good. :eek:

JST · Oct 24, 2006

If you look at your pic of what appears to be the power supply.....

C11 and C12, C14, and at least 2 others appear to be the solid tantalum parts I mentioned.

You may need to replace them with same type, due to that looking like an SMPS, which are dependent on some capacitor specs for proper operation.

A check may find one shorted, especially on the supply line you mentioned, although from this side it looks like none of them are ON that line. Possibly the other side shows a connection.

There are some other parts at that TP area, which may indicate it is really a signal line of some sort.

Or, it may be created by a "charge pump" composed of those parts, since you have a battery, and that is presumably a negative voltage..... in which case the circuitry may be bad, or being told not to work by some other failure.

The pic just barely fails to allow reading the chip numbers..... which would help.

JL Sargent · Oct 24, 2006

Lets see if this is readable:

Not really, but:
U1 and U2 are MAXIM 8936 MAX634CPA
U3 is MAXIM 8915 MAX632ACPA
U4 is P89526 MM74HCO4N MC7HCO4N
U5 is 8827H 74HC164AP

JST · Oct 24, 2006

Looks like "charge pump" may be the winner.....

http://www.maxim-ic.com/appnotes.cfm/appnote_number/90

In which case, it may not BE pumping, for various reasons....

U1 has 2 of those *wonderful* solid tanatalums on lines coming from it, C2 and C16.

C11 and C12, with CR 3 & CR4, and supporting circuitry, look very much like they could be as per the app note linked above. At least there are about the right number of parts.

On the other hand, I don't know why the -V, which may be for a display, would cause it not to work at all. Unless of course, it IS working, but the display is not telling you so

{edit.... Not telling you so because it hasn't the right voltages to work due to -V being dead}.

And, the "8926" etc are date codes on teh parts..... format YYWW W=week number. That tells you when it could have been made, at the earliest. Could obviously be later.

JL Sargent · Oct 24, 2006

I have a spare screen/keypad. Ill swap those and see what I get. Thanks so much for the help.

On edit: I have tried the spare keypad and to no avail. I called the factory about it and they said the fact that its not totally dead means it could be an easy fix. He suggested replacing Q9 on the power board and that board does not even have a Q9

Several revisions over the years make over the phone recommendations tough I guess.

J. I appreciate your looking those ICs up for me and all the help. Im gonna be out of pocket til the first of the week and I plan to continue working on it then. Thanks again to everybody.

OT: Electronic question, eproms and such.

Diamond

Aluminum

Titanium

Cast Iron

Diamond

Diamond

Cast Iron

Titanium

Diamond

Stainless

Titanium

Titanium

Diamond

Stainless

Diamond

Diamond

Diamond

Diamond

Diamond

Diamond

Similar threads