Erasing and programming 2532 EPROMs

[d_a_parker]

Hi All,

I'm working on a Commodore PET 8032 and wanted to use the PETTESTER ROM to trace a possible memory issue. I have two used TI TMS2532JDL-45 chips, and ten (supposedly) NOS Hitachi HN462532G chips.

I was able to erase one of the TI chips and program it with the EDIT ROM for the PET, which then worked fine. But I have subsequently been unable to erase or reprogram ANY of the other chips, and I'm at my wit's end. After 15-20 minutes in the eraser, they contain garbage with a smattering of FF sprinkled in. When I try to program one using my T48 and 2532/2732 adapter (which did work with that first TMS2532), I get a garbled version of the source image for the first 16-32 bytes, then just total garbage. No matter what I do, I can't seem to erase or program these 2532s.

I've tried using the XGecu software on Windows with various combinations of options, but no luck.
I've tried using minipro on my Mac with explicit voltages and a long pulse delay (e.g., -o vcc=5 -o vpp=25 -o pulse=10000) but still no luck.
I verified that Vcc and Vpp are correct at the chip during the programming cycle. Vpp generally holds at 25.2V but sometimes varies anywhere between 24.9V and 26.2V, which seems to be within tolerance, based on the datasheets.

NOTE: I have an old but working industrial UV EPROM eraser (not one of those cheap crap ones on the market today), and XGecu T48 EPROM programmer. They have both worked fine with other ROMs.

Any tips, tricks, or other help is greatly appreciated!

 

[Plasma]

I wouldn't bother trying to program anything that doesn't pass the blank check. Make sure the windows are clean and the pins are all making good contact with your programmer.

 

[channelmaniac]

If the chip die is large under the window you'll need to ensure it's directly under the bulb in the eraser to avoid shadows causing incomplete erasure.

 

[d_a_parker]

I wouldn't bother trying to program anything that doesn't pass the blank check. Make sure the windows are clean and the pins are all making good contact with your programmer.

I did clean the windows, and I'm pretty sure all of the pins are making contact. I'll verify that end-to-end for all 24 pins, though.

If the chip die is large under the window you'll need to ensure it's directly under the bulb in the eraser to avoid shadows causing incomplete erasure.

I only put two at a time in the eraser, and tried my best to make sure they were centered in the tray directly under the bulb. I'm wondering if a longer erase time is needed, although that first TI chipped erased in 15 minutes. I just don't want to damage them.

Thanks!

 

[ldkraemer]

The TMS2532 EPROMS REQUIRE a 25 Volt Programming voltage.

I use my old reliable BP Microsystems EP1 Programmer to program them easily after verifying they are erased.

I do this for ZERO Cost, you just fiurnish the EPROMS, File to burn, and pay return postage.

You won't beat this deal anywhere.


Larry

 

[Hugo Holden]

You could try a longer erasure time but likely the problem is that your chips are USED parts and damaged, rendering full erasure impossible. This happens because at the time they were initially programmed, they were over programmed and damaged, but the damage in that case has no effect because they had the correct byte pattern. It is only later, when an attempt to erase them that some of the cells won't go to FF no matter how long you leave them in the UV light that the damage becomes obvious. I learned this lesson the hard way. After that, I only ever buy unused NOS ROMs and they come already erased anyway.And if they are not all FF when you get them, you know the chip has sustained some other damage and don't use it.

This does not exlain your new Hitachi chips, I would wonder if they might be relabelled 21Vpp parts that get damaged by 25V programming.Or chips that have been through a recycling center. If they were all indeed new parts, none should have required erasure and they should have FF in all cells and pass the blank check off the bat. Can you post some photos of these for examination ?

If the Hitachi chips look like genuine unused blanks, then it is plausible, from what you have said, that your programmer has a failure in evolution. Or something has gone awry with your adapter. Adapters a always a curse to be avoided if possible. You can program the TMS2532JL in a GQ 4x4 programmer without an adapter.

Also many programmers, to program these chips (unless you have a line powered programmer like the BP1400), require at the time they are supported by a power adapter, not just acquiring the power from the USB cable alone (I have no idea how your T48 is powered). Or they can give erratic programming. But this won't explain the inability to erase. I'm not familiar with the T48, but I remember the name (because it reminds me of a Terminator model) and I think it has been implicated in other programming issues on threads, which you never seem to hear about with the GQ-4x4 or the BP1400.

And there is also the possibility that the tubes in the UV light source are weak. So you have some detective work to do.

 

[maxtherabbit]

You could try a longer erasure time but likely the problem is that your chips are USED parts and damaged, rendering full erasure impossible. This happens because at the time they were initially programmed, they were over programmed and damaged, but the damage in that case has no effect because they had the correct byte pattern. It is only later, when an attempt to erase them that some of the cells won't go to FF no matter how long you leave them in the UV light that the damage becomes obvious. I learned this lesson the hard way. After that, I only ever buy unused NOS ROMs and they come already erased anyway.And if they are not all FF when you get them, you know the chip has sustained some other damage and don't use it.

This does not exlain your new Hitachi chips, I would wonder if they might be relabelled 21Vpp parts that get damaged by 25V programming.Or chips that have been through a recycling center. If they were all indeed new parts, none should have required erasure and they should have FF in all cells and pass the blank check off the bat. Can you post some photos of these for examination ?

If the Hitachi chips look like genuine unused blanks, then it is plausible, from what you have said, that your programmer has a failure in evolution. Or something has gone awry with your adapter. Adapters a always a curse to be avoided if possible. You can program the TMS2532JL in a GQ 4x4 programmer without an adapter.

Also many programmers, to program these chips (unless you have a line powered programmer like the BP1400), require at the time they are supported by a power adapter, not just acquiring the power from the USB cable alone (I have no idea how your T48 is powered). Or they can give erratic programming. But this won't explain the inability to erase. I'm not familiar with the T48, but I remember the name (because it reminds me of a Terminator model) and I think it has been implicated in other programming issues on threads, which you never seem to hear about with the GQ-4x4 or the BP1400.

And there is also the possibility that the tubes in the UV light source are weak. So you have some detective work to do.

What do you mean by "over programmed"?
Vpp too high?

 

[Hugo Holden]

What do you mean by "over programmed"?
Vpp too high?

There are programming protocols used for various ROMs. It could be the programming voltage was too high, that could be one way it can happen when 21V parts are programmed with 25V as one example, or it could be that the protocol over-wrote the data too many times or for durations that are too long, into the memory cells, making it harder to reverse the process later and shift the charges back the other way.

 

[channelmaniac]

You can try to program the chip to all zeros then erase it again.

 

[GiGaBiTe]

I would recommend buying a GQ 4x4 programmer for programming EPROMs. The T48 is not a good programmer for programming EPROMs because it can't put out enough write current, and will result in failed programming. Some guides tell you to put the T48 on a powered USB hub, this does not help and it will still fail to program.

I've not been able to get my T48 to program even a 27c256 without failing shortly after programming was started.

The only thing the T48 is good for is newer EEPROMs, especially ones that require a lower voltage. It not being powered means you're at the mercy of USB's power regulation, and it's not good.

 

[Hugo Holden]

I would recommend buying a GQ 4x4 programmer for programming EPROMs. The T48 is not a good programmer for programming EPROMs because it can't put out enough write current, and will result in failed programming. Some guides tell you to put the T48 on a powered USB hub, this does not help and it will still fail to program.

I've not been able to get my T48 to program even a 27c256 without failing shortly after programming was started.

The only thing the T48 is good for is newer EEPROMs, especially ones that require a lower voltage. It not being powered means you're at the mercy of USB's power regulation, and it's not good.

That agrees with what I theorised in post #6 and I agree that he needs to get the GQ 4x4, it has an external DC power connector as well as the USB. I know for a fact this programmer will reliably program the TMS2532JL Its software often advises to connect it to the external power. Its software is non-buggy and it programs many vintage chips and its way better than many of the cheaper bare board programmers out there, which seem to do odd things and drive people within an inch of crazy, apart from their bare pcb's just begging to be shorted out from some debris on the bench.

I think some of the cheaper programmers got hardware cloned and the firmware pirated and they have borderline timing incompatabilities. For example one type on occasions fails to program the first byte of the file properly.

And if the power supply arrangements in them are nothing more than USB power sourced, it appears to be asking for trouble with attempting to program some of these vintage chips which must have higher energy requirements during the process.

The makers of the GQ-4x4 are responsible for the whole thing, each unit has a bar code indentity to help eliminate clones and verify it. They are also responsive to suggestions and requests from users. In the past I advised them that this programmer, when set for a DS1225 BBNVRAM, would program a Ramtron FM16w08 FRAM and they added it to their extensive supported device list.

 

[maxtherabbit]

I've programmed countless 27c256 EPROMs with my TL866II. I'd be surprised if the T48 was really that much of a regression...

 

[Hugo Holden]

I've programmed countless 27c256 EPROMs with my TL866II. I'd be surprised if the T48 was really that much of a regression...

Not sure what that means, but if a 16 year old regresses to an 8 year old, it becomes much more difficult to solve a quadratic equation unless they are particularly gifted.

 

[maxtherabbit]

Not sure what that means, but if a 16 year old regresses to an 8 year old, it becomes much more difficult to solve a quadratic equation unless they are particularly gifted.

The T48 is the replacement (newer) product from the same vendor who made the TL866II. What I am saying is that I am skeptical of gigabite's claim that the newer product won't program a 27c256, since that would be such a large step backwards from the same people.

 

[Hugo Holden]

The T48 is the replacement (newer) product from the same vendor who made the TL866II. What I am saying is that I am skeptical of gigabite's claim that the newer product won't program a 27c256, since that would be such a large step backwards from the same people.

They simply may have been unaware that a design change affected it for programming particular parts.

It would be unlikely the new model was re-tested for every IC speciment on the supported devices list again.

Though it would be interesting to see if the power supply arrangements were altered on the new version. Also it would be worth checking if anyone had cloned it, and if the unit could be a cheap copy of the originals.

In any case Gigabite has detected the problem and it cannot be solved for that part it appears because the unit has no other power than USB.

Unless, and it is always possible, the T48 might work when new but a fault in the power supply section could evolve over time, for example an electrolytic cap with an increasing ESR and/or reduced uF value. This would be consistent with the observation that initially the OP could program the 2532 parts and now cannot. Maybe post the schematic of the T48 here and we can see how they generate the the 25V Vpp from the 5V usb power and if there is any experiment that could be done to confirm or deny this theory.

Even with a limited relatively high internal resistance 5V power source, it should be possible to stabilize the unit's voltages over a along enough time frame for the programming to complete, if there was enough charge storage in a large enough value electrolytic capacitor.

 

[Plasma]

In any case Gigabite has detected the problem and it cannot be solved for that part it appears because the unit has no other power than USB.

It doesn't matter how good the programmer is (or isn't) if the EPROMs aren't fully erased.

 

[modem7]

That agrees with what I theorised in post #6 and I agree that he needs to get the GQ 4x4, it has an external DC power connector as well as the USB. I know for a fact this programmer will reliably program the TMS2532JL Its software often advises to connect it to the external power.

GQ-4x4 = GQ-4X V4 = Version 4 of the GQ-4X

Your information is interesting to know because of what is written at [here], concerning the larger chip, the 2564. My early GQ-4X ('GQ-4X' printed on top) has no problem with the TMS2564JL, but some others, who have a later version of the GQ-4X report a problem. But 'cause and effect' has not been researched. Faulty TMS2564, inadequate power, etc. have yet to be ruled out.

 

[DeltaDon]

I believe the old TL866 could provide a higher programming voltage than the newer T48. So it might be possible that the T48 can't reach 25v.

 

[maxtherabbit]

They simply may have been unaware that a design change affected it for programming particular parts.

It would be unlikely the new model was re-tested for every IC speciment on the supported devices list again.

Though it would be interesting to see if the power supply arrangements were altered on the new version. Also it would be worth checking if anyone had cloned it, and if the unit could be a cheap copy of the originals.

In any case Gigabite has detected the problem and it cannot be solved for that part it appears because the unit has no other power than USB.

Unless, and it is always possible, the T48 might work when new but a fault in the power supply section could evolve over time, for example an electrolytic cap with an increasing ESR and/or reduced uF value. This would be consistent with the observation that initially the OP could program the 2532 parts and now cannot. Maybe post the schematic of the T48 here and we can see how they generate the the 25V Vpp from the 5V usb power and if there is any experiment that could be done to confirm or deny this theory.

Even with a limited relatively high internal resistance 5V power source, it should be possible to stabilize the unit's voltages over a along enough time frame for the programming to complete, if there was enough charge storage in a large enough value electrolytic capacitor.

I have no idea where one would get the schematic for the T48, I don't think it's an open source device.

It is worth noting that most 27C256 parts only require about 13Vpp

 

[Hugo Holden]

GQ-4x4 = GQ-4X V4 = Version 4 of the GQ-4X

Your information is interesting to know because of what is written at [here], concerning the larger chip, the 2564. My early GQ-4X ('GQ-4X' printed on top) has no problem with the TMS2564JL, but some others, who have a later version of the GQ-4X report a problem. But 'cause and effect' has not been researched. Faulty TMS2564, inadequate power, etc. have yet to be ruled out.

Interesting , I have the older model GQ-4x which appears to have been replaced by the 4x4 model, and I have no idea what the internal circuit differences are. Though whenever I use it, I always have an external DC power pack connected to it, because without that for some chips the program will fail, so I just use it all the time to avoid any trouble. Trouble is, without it, it might seem to work & verify, but the programming of a cell could be marginal and a bit could flip later.