PDA

View Full Version : Anything special about programming 27512 eproms ?



mc68010
May 7th, 2013, 09:50 AM
I bought a Willem burner off ebay a long time ago and it works fine for everything I've tried but, 27C512 and 27512 eproms. It can read them fine and it thinks it is programming them but, it never actually writes anything. I can program 27C256 and 27C010 just fine. I've tried about ten different eproms and it's no go.

I don't know if it's flaky hardware or the fairly buggy software these use. I have tried several versions of the software and nothing changes.

If I can't get this figured out anyone care to burn a quick eprom for me ? I erased the one I was replacing sure it would be a quick job but, I am never so lucky. I should have left the original alone.

Chuck(G)
May 7th, 2013, 09:59 AM
What manufacturer and what exact part number for the 27C512s? This size hails from a time when programming algorithms were not uniform across manufacturers.

MikeS
May 7th, 2013, 10:50 AM
...If I can't get this figured out anyone care to burn a quick eprom for me ?Always helps to put your location in your profile...

mc68010
May 7th, 2013, 11:36 AM
What manufacturer and what exact part number for the 27C512s? This size hails from a time when programming algorithms were not uniform across manufacturers.

I have a few different ones but, they are all AMD. AM27512DC, AM27512-30DC, AM27C512-150DC, AM27C512-155DC

mc68010
May 7th, 2013, 11:37 AM
Always helps to put your location in your profile...
I'm in California.

SpidersWeb
May 7th, 2013, 12:26 PM
If you haven't yet, check the programming voltage.

Some chips cause quite a drain, and at least mine from ebay, doesn't regulate it properly - so when programming the 12.5V can easily drop to 10 or 11. I have to adjust mine all the time, quite a pain. Too much voltage and the chips overshoot, not enough and they undershoot or don't write at all.

Chuck(G)
May 7th, 2013, 01:14 PM
Hmmm, nothing special. It is different in that Vpp is applied to pin 22 instead of Pin 1 as on the smaller AMD (e.g. 27C256) EPROMs. I wonder if it's a bug in the Willem software. Time to check some support forums, methinks.

mc68010
May 7th, 2013, 02:57 PM
Hmmm, nothing special. It is different in that Vpp is applied to pin 22 instead of Pin 1 as on the smaller AMD (e.g. 27C256) EPROMs. I wonder if it's a bug in the Willem software. Time to check some support forums, methinks.

I'm not sure there is a support forum. I get the feeling these are just some opensource hardware project from a long time ago that the Chinese are just mass producing under 100 different names now. The software has never even reached version 1.0 and doesn't appear to be actively developed.

mc68010
May 7th, 2013, 02:58 PM
If you haven't yet, check the programming voltage.

Some chips cause quite a drain, and at least mine from ebay, doesn't regulate it properly - so when programming the 12.5V can easily drop to 10 or 11. I have to adjust mine all the time, quite a pain. Too much voltage and the chips overshoot, not enough and they undershoot or don't write at all.

How are you adjusting your voltage ?

SpidersWeb
May 7th, 2013, 08:16 PM
Also check you've set the switches correctly for the EPROM type - in the program I use it draws them on screen after you select the chip you want. I thought of that when Chuck mentioned about the programming pin moving.

On mine, not sure if its the same, there was a very very tiny trimpot that is adjusted with a jewelers flathead screwdriver. Instructions for setting Vpp were in the software, but I usually just increased it a bit and tried again until the chip was writing correctly. For something like a 27C512 you could just follow the instructions in the software - it should be fine.

With mine set to 12.5 (using their instructions) I could program 27C256, 27C512 no problems. But slow or older ROMs I had to set it a volt or so higher to compensate for droop and increase the timings (I can't remember the names, it's been a long time, but there were two sliders to adjust timings). When it arrived it was set so badly it wouldn't even program the 27C512 they sent with it! Also rather annoying, the chip they sent was zero-filled, so I had to erase it first :S

Of course I'm basing this on your controller/program being very similar to my ebay-special. So if bits don't make sense, that could be why. Also apologies for being so vauge, it's because I haven't used it in almost a year, just hoping I can point you in a direction for things to try.

Chuck(G)
May 8th, 2013, 08:30 AM
I believe that the datasheet for the AMD 512Kb EPROM spec Vpp at 12.75V nominal +/- 0.25V. So 12.5V would be grazing the bottom edge of the range.

mc68010
May 8th, 2013, 12:01 PM
Also check you've set the switches correctly for the EPROM type - in the program I use it draws them on screen after you select the chip you want. I thought of that when Chuck mentioned about the programming pin moving.

On mine, not sure if its the same, there was a very very tiny trimpot that is adjusted with a jewelers flathead screwdriver. Instructions for setting Vpp were in the software, but I usually just increased it a bit and tried again until the chip was writing correctly. For something like a 27C512 you could just follow the instructions in the software - it should be fine..

Mine doesn't have a trimpot. Just a couple jumpers for selecting VPP. Sounds like you got a higher quality one. They do seem to vary a lot. I think I have something wrong with mine. If I check the voltages in the hardware test of the application they are fine, or close to fine. I get 12.5 on Vpp and 5.6 on Vcc. The test readings are made with the eprom out though. When I pop in a 512 and read Vcc goes down slightly to about 5 but, it does read. When I try and program it all goes to hell though. Vcc drops to about 3 and Vpp nose dives down to about 2. It does this on USB power or a 12vdc@1a ac adapter. Vpp is showing up on the right pin though. Just not enough of it.

I am erasing some 256 right now to see what voltages do on those. I imagine they are probably ok since they program fine but, I am curious.

mc68010
May 8th, 2013, 12:03 PM
I believe that the datasheet for the AMD 512Kb EPROM spec Vpp at 12.75V nominal +/- 0.25V. So 12.5V would be grazing the bottom edge of the range.

Doesn't appear like I am getting even that. Think something is seriously wrong with my burner.

dave_m
May 8th, 2013, 05:49 PM
I have a spare 27C512 and a trusty Data I/O. Tell me where I can download the file, and tell me the file format (binary, Intel Hex, etc.). I don't think you can attach a file via private message.
-Dave

dave_m
May 9th, 2013, 02:40 PM
My Data I/O 29B has let the team down. The EPROMs I have are AMD 27C512 and there must be something different about them as they require a V23 rev on the Unipak2B. The other 27C512's from Intel, etc only require rev V19 or less. And since my Unipak is at rev 21, I will need you to send me a device I can handle such as anyones 27512, or a 27C512 from Intel, Atmel, National, NEC, Signetics, Mitsubishi, or TI.

Chuck(G)
May 9th, 2013, 03:42 PM
I can check my Xeltek programmer to see if it can do them, if anyone cares.

mc68010
May 9th, 2013, 04:44 PM
I can check my Xeltek programmer to see if it can do them, if anyone cares.

I would care ;) I really don't want to buy a new programmer. I may never need to burn a 512 again and so, far this guy does everything else fine.

dave_m
May 9th, 2013, 05:07 PM
Thanks, Chuck. The OP has indicated he has some 27512's, but I was trying to save time by using one of mine. But it turned out to be an AMD27C512 which my burner can not handle. :oops:

If you have a couple of spare 27512 type devices you can burn for him, that would be great. Or he can send me some blank devices that I can burn.

The file he needs programmed is a Motorola S record file containing 64KB of data. That doesn't seem like much but it took almost 5 minutes for me to transfer the data via RS-232 to the burner. I'd better check the baud rate I have set in my Data I/O. You gotta love 1980 equipment.

Chuck(G)
May 9th, 2013, 05:20 PM
Well, I checked my Xeltek Supepro and even an old M2L EZEP programmer--both handle AM27C512, no caveats. I don't have the AMD parts (I do have an AM27C010 and both handle that just fine), just the TI parts in 512Kb, so I wasn't able to check it.

Malc
May 10th, 2013, 02:14 AM
Just checked my device list for the Unipak 2B and it requires V27 firmware and adapter 351B086 for the AMD 27C512, It's the Unipak 2 that requires V23 firmware to program that chip.


My Data I/O 29B has let the team down...

dave_m
May 10th, 2013, 06:20 AM
Just checked my device list for the Unipak 2B and it requires V27 firmware and adapter 351B086 for the AMD 27C512, It's the Unipak 2 that requires V23 firmware to program that chip.

My information that V23 was OK for the Unipak2B came from the Data I/O 1991 Wall Chart and the AMD 27C512 was listed in bold type which indicated a recent algorithm change. If your information came from a later source then it implies that the algorithm kept changing. There must be something 'special' about the AMD part...

mc68010
May 10th, 2013, 10:35 AM
My information that V23 was OK for the Unipak2B came from the Data I/O 1991 Wall Chart and the AMD 27C512 was listed in bold type which indicated a recent algorithm change. If your information came from a later source then it implies that the algorithm kept changing. There must be something 'special' about the AMD part...

Maybe I should try and find a non-AMD 512 and try it on my burner. I am not really sure why all I have is AMD parts. I pulled them off all sorts of different gear and AMD is all I've found so far. They must have been pretty common.

Malc
May 10th, 2013, 01:42 PM
In the early 1990's AMD did a die shrink on the 27C512 eproms and there was an algorithm change on the Unipak 2B to reflect the changes, also the FPO code was changed. There's a post on this on the Yahoo Data I/O Programmer's group.

If you can find a very early 1990 AMD 27c512 chip it should work with V23, What are the date codes of the chips you have ?.


My information that V23 was OK for the Unipak2B came from the Data I/O 1991 Wall Chart and the AMD 27C512 was listed in bold type which indicated a recent algorithm change. If your information came from a later source then it implies that the algorithm kept changing. There must be something 'special' about the AMD part...

dave_m
May 10th, 2013, 03:34 PM
In the early 1990's AMD did a die shrink on the 27C512 eproms and there was an algorithm change on the Unipak 2B to reflect the changes, also the FPO code was changed. There's a post on this on the Yahoo Data I/O Programmer's group.

If you can find a very early 1990 AMD 27c512 chip it should work with V23, What are the date codes of the chips you have ?.

OK, that makes sense. The date code on the device seems a little cryptic: 918LA6Z. Probably 1991 part? Wow, Looking into the quartz window, the geometry of the matrix seems a lot finer than a 2764. I wish I had a microscope.
-Dave

Update: The date code for this EPROM is on the underside of the part: 8915 so it is a 1989 part and V23 would have worked (however I have V21). In the Data I/O forum, they mentioned trying code 0DA/0A4 and disabling the electronic ID check. The algorithm is probably a low yield one for this part, but might work. I'll give it a shot.

Malc
May 10th, 2013, 11:17 PM
Worth a shot if you have a non AMD 512, I'm not familiar with the willem's but how old is it, try a AMD 512 a bit older than the willem, There was a change on the AMD 27C512 in the early 90's.


Maybe I should try and find a non-AMD 512 and try it on my burner.

mc68010
May 11th, 2013, 10:45 AM
Worth a shot if you have a non AMD 512, I'm not familiar with the willem's but how old is it, try a AMD 512 a bit older than the willem, There was a change on the AMD 27C512 in the early 90's.

Here's an example of the Willhem burner. http://www.sivava.com/EPROM_Programmer_5.html As far as I can tell it was designed a few years ago by someone named Willhem and put out as opensource hardware for anyone to make or change. The Chinese produce a 100 different versions of the same thing under 100 different names. There are tons of different ones on ebay alone. The run about $30 - $50 and most get shipped directly from China. The quality varies greatly and often resistors and caps of the wrong value are installed from what I have read. There is no central source of software. I think there was only ever a beta version that the original people made that has been modified over the years for improvements made to the design. The latest version is still only version 0.98 .

Its a popular programmer for more modern devices like those found in cable boxes, consoles, and car computers. I don't think most owners care about the 27xxx stuff. If there was a bug in that portion it might go unnoticed and unfixed forever and nobody would care.

Malc
May 12th, 2013, 01:38 AM
I remember looking at similar types a long time ago, Don't know if you have seen the forums at mcumall . com but they have a section for the "open source project willem programmer hardware" might be worth a look, hope you get it sorted soon.


Here's an example of the Willhem burner...