Character generator ROM woes, need troubleshooting advice

[Bob-O-Rama]

I have TWO AT&T 4425 terminals circa 1983. One of them had minor character generation issues, with a few letters which displayed improperly. On pulling and dumping the ROM I found it was damaged:



And for the most part the characters being displayed incorrectly were those from the 80 colum font ( see right section with black vertical bands which are 1 nibble wide. So I pulled the ROM from the "good unit" and found it was, well good:


I took the "bad" EPROM, erased it, and programmed it with the data from the "good" ROM. Then confirmed its contents were identical and that the character matrix looked fine. So now I have a supposedly good ROM to put in the suspect system. When I did, all hell broke loose. Now every character is displayed as garbage and some of the garbage flickers. So its possible I borked the main board popping the EPROM. Need to look into it. The wierd thing is popping it and reading it still shows good data.

Is there a way the EPROM can appear good to the reader, but not to the main board - something triggered by overwriting it? Whatever I did created a far worse issue. so I have to think its not the EPROM.

If there is some good guide / process for troubleshooting these issues, I'm all ears.

-- Bob

 

[resman]

This may have nothing to do with your issue, but when I program a character ROM for my Apple ][, I have to write the image a number of times for it to stop "sparkling". I have one of those cheap USB programmers, so maybe it doesn't have the properly spec'ed voltage for the EPROM, or something. I dunno, but the more times I write the image to the EPROM, the better it looks. I think I usually write it four times over, then it looks fine on-screen with no sparkling.

 

[Bob-O-Rama]

I looked at the problem child EPROM and it is a HN482764G which has a programing voltage of 21v. The TL866II+ programmer maxes out at 18v. So while it squeaks by, it probably leaves the cells incompletely programmed. I hadn't noticed it had a different pgm voltage than the other ones. I have a 21v fixture already, if I can find it. Basically it is a couple 28 pin sockets stacked up rigged to isolate the programming pin and place a 3v button cell and a current limiting resistor ( like 330 ohms ) in series. This gets the pin to 21 volts.

Great suggestion, totally makes sense. Thanks!

 

[Bob-O-Rama]

This may have nothing to do with your issue, but when I program a character ROM for my Apple ][, I have to write the image a number of times for it to stop "sparkling". I have one of those cheap USB programmers, so maybe it doesn't have the properly spec'ed voltage for the EPROM, or something. I dunno, but the more times I write the image to the EPROM, the better it looks. I think I usually write it four times over, then it looks fine on-screen with no sparkling.

That was it! I just re-burned it about 10x in a row and its working again. There are still an occasional missing bit here and there if you really pixel peep, but I'm OK with it, as I thought I had totally destroyed it. Thanks!



-- Bob

 

[resman]

Glad to hear this worked for you, too. That's a nice clear screen.