• Please review our updated Terms and Rules here

VT100 with RetroGraphics Enhancement

While I was at it, I checked the Clinton CRT with my Müter BMR "rejuvenator". This tube is basically on its last legs, but I didn't bother rejuving (and possibly toasting) it. Also, it's starting to delaminate around the edges (see piccie). This isn't noticeable (yet) when the bezel is mounted, but's only a matter of time. :(
I would say that this 500uA is way enough to give a clear picture on that little screen. But it is too bad the front glass is coming off. I have had these kind of tubes too, that metal rim makes it impossible to remove the front glass. I have replaced some of them with a CRT from a black and white TV. But you have to find a suitable tube off course. They are not all the same and not green as your tube is here...
Hi everyone, hope you're enjoying your Easter break.

Quick update: the VT100 suddenly turned comatose; just garbage on screen, and no beep. According to the docs, the keyboard LEDS L1..L4 *may* indicate an error code from POST. L2 is consistently lit, which would indicate a checksum error in ROM4 on the terminal controller PCB. But that socket is empty, at least using the Retrographics ROMset, which consists of 3x 2532s. I checked dumps of these against those I made in 2022, and they are identical.

So now I suspect there's a RAM fault, although this should be indicated by a lit L2 + L4. The +5V, -5V and 12V supplies check out, so I will try swapping banks to see if that has any effect. The terminal controller PCB is fitted with 4x5 MM5290J-3's, which I understand are pin compatible with 4116s. I may have some spares of the latter, but not enuff to replace all. I guess it's time to invest in a RAM tester... 🤔

Probably should start a new thread before getting OT, but just wanted to plonk this here. Hints of course appreciated.

More to come (I hope), Happy Easter and best wishes,



  • 20240331_003110.jpg
    6.3 MB · Views: 13
Last edited:
Even though you were able to read out the EPROMs and verify they match does not mean the VT100 can read them at speed. One of the problems is that as time passes the access time gets longer and longer. You can present an address and eventually the correct pattern will show up on the data outputs. An erase and reburn could give you another 10 to 20 years of use. The burner hardware generally does not read the part at full speed so it can't tell the programming has gone soft. Of course that might not be your problem. But refreshing the EPROMs is an easy thing to try since it sounds like you have the capability already.

Best Wishes!
Even though you were able to read out the EPROMs and verify they match does not mean the VT100 can read them at speed. One of the problems is that as time passes the access time gets longer and longer. You can present an address and eventually the correct pattern will show up on the data outputs. An erase and reburn could give you another 10 to 20 years of use.

Thx, Doug, will try that. I'm pretty sure I have spare 2532s around.

I also discovered I *do* have a batch of 4116s that I had sitting around for another project that's now stalled indefinitely (ahem), so I could infact replace the whole set and take it from there.

I now realised that the VT100 schematics and tech docs are pretty much useless, as the terminal controller -- the VT640 board -- is very much specific to the Retrographics enhancement. I haven't found any schematics for the VT640 board, and I expect the VT640's POST does not necessarily indicate errors via L1..L4 like the basic VT100 does. It also means this discussion still belongs in this thread.

Will keep you posted,

BTW: is this issue specific only to EPROMs?

As opposed to other types of ROM? Pretty much. There is a floating gate in the EPROM which gets a charge applied when programmed. As time passes this charge diminishes or bleeds off. The UV eraser light causes an acceleration of this. I think the some of the 2732 parts suffered much more from this because they were an older tech pushed to its limits. The 2764 parts tended to be better because they were the next generation. This is what I remember reading. I got interested when I ran into parts that were so slow the access time could be measured in tenths of seconds. EPROMs are among the most interesting of parts because you can see the guts through the window.

The only reason I brought this up is because if it is a part that has gone slow then you could be looking for problems that aren't there. And if you have a burner then it is a pretty easy fix.
Thx for the explanation, Doug. I checked on the web and saw similar reports elsewhere, tho I never had issues with EPROMs myself. Then again, most of my stuff doesn't use the older types. The VT100 and maybe the AIM65 are an exception with their 25xx ROMs. And yeah, EPROMs are cool; I remember how fascinated I was with my first 2764s I got for an EPROM burner project for my Amstrad back in '86. Those cute little windows revealing the chip were a blast. The burner never worked, but I still have it somewhere. That was my first foray into photoetching PCBs. Defo looks homemade. :^)

Quick update: Burning new 2532s changed nothing. Infact, I now realise the "garbage" pattern is actually identical every time it powers up. That seems to rule out an issue with missing reset.

I now also realised I was probably barking up the wrong tree -- I was looking at the VT640 board with custom Retrographics ROMs, but I should be looking at the original VT100 terminal controller board, which contains the original 2K ROMs, for which dumps can be found on Jeff Parsons' website (https://www.pcjs.org/machines/dec/vt100/rom/).

BUT: I have no way to read these on my Data I/O 29B and 288. At the very least, I would have to bodge (pull up or down) the 8316's three custom programmed chip select lines which decode the A11 and A12 address lines mentioned on that website. In my experience, the Data I/Os are very picky about this and will report a faulty device / socket / programming pak if you start buggering around with pins. And assuming these ROMs are faulty, I wouldn't know what to replace them with. If anyone has any ideas, I'd love to hear them.

I might also consider simply powering the board on my bench and do basic measurements (reset clock, bus activity).

Again, suggestions welcome.

Last edited:
Yet another update, this time with good news.

I finally managed to read the ROMs as an MCM68764/68766 with pulled-up CS1 in my 29B (the 288 still wasn't happy). These EPROMs are the only ones I'm aware of that cram 64kb into a 24pin package, which makes them convenient drop-in replacements for mask programmed ROMs, with the CS2 and CS3 pins neatly lining up with A11 and A12. The catch is that they are very rare these days, and much slower than most ROMs. Thankfully, the VT100's ROMs are an exception, as they are also rated at 450nsec.

I compared the dumps I got with those from Jess Parsons' website, and sure enough, the upper ROM in the address space (0x1800-0x2000) is bad, interestingly only in its lower half. This ROM is indeed ROM4 in the VT100's memory map, so the POST error code was spot on after all!

After checking the options for a viable replacement for all four ROMs, I decided to relinquish one of my precious few 68764s, and burned it with a consolidated 8k image. This EPROM of course expects an active low CS on pin 20. Fortunately, the VT100 has an option jumper W2/W3 on the terminal controller board to negate the CS sense. Swapping the jumper link and dropping the EPROM into one of the four slots did the trick -- another VT100 saved!

TBH, I got lucky here; the only reason this worked with such a slow EPROM is because the VT100 isn't picky about access times. In most other situations, it would be way too slow.

Best regards,