Since some readers appear to enjoy identifying prehistoric graphics cards, I thought I’d post photos of two more TIGA boards from my junk pile. I know what these are, but do you? One of them ought to be fairly easy to identify and photos of it exist on the Internet. The other one might be tougher and to my knowledge no photos of it are available.
Let’s start with Exhibit A: This is an ISA-based graphics card by a well known manufacturer whose name is clearly legible on the PCB. It’s a combined VGA + TIGA card, a somewhat unusual yet very logical combination.
There’s a 60MHz TMS34010 GPU and a BrookTree Bt473 DAC. The DAC is a 66MHz part, so no fancy high-res modes, just basic 1024×768. Somewhat unusually, the board uses standard 30-pin SIMM slots for GPU memory, which means it was probably user upgradable. There is separate VRAM in typical ZIP modules.
Common for boards of this vintage, there’s a large array of oscillators, made even more impressive by the fact that there’s one for the GPU, two for the VGA chip, and four for the TIGA part. No wonder designers switched to PLL clock synthesizers.
The most unusual feature of this graphics card is its VGA chip… an Intel 82706. Yes, Intel made VGA chips back in the 1980s! Clearly not very successfully, and sadly, a data sheet for this chip is not available, only the document’s part number is known. The board was manufactured around mid-1990.
And now for the second board. This is a Microchannel graphics card made in USA by a large Japanese company, not a common combination in the world of PCs. The board is equipped with a memory expansion daughter card, shown separately in the photo:
Once more there’s a TMS34010 chip, though running at only 50MHz. The DAC is a Brooktree Bt471, again a 66MHz variant. And there’s a Chips & Technologies 82C611 MCA bus interface module. Here’s a close-up of the major chips:
The reverse side of this board is populated with a fair number of chips (latches?).
There is no VGA function, as this board was again intended to be used in PS/2 machines with on-board VGA graphics. This specimen was manufactured roughly in mid-1989.
Can you identify these TIGAs? No cheating, please…
Is using Google cheating? 🙂
No, of course not 🙂
First off, there’s an 82076 datasheet contained here:
pg. 1136 – 1153. (I’ll separate it out for you when I get a chance.)
The first card is a Hercules Graphics Station GB1024+2, I believe.
The second card is a NEC MGE-MCA 16 -Multisync Graphics Engine for Micro Channel PS/2.
(make that 82706)
Parted out the datasheet, it’s currently living here:
Thanks! I wonder why Google didn’t find it for me (inside the bitsavers databook). Anyway, that datasheet is disappointing, I was hoping they’d have actual programming reference in there (and then I’d have yet another conflicting source of VGA information). Oh well.
I expect the reason is probably because the OCR performed by bitsavers is incomplete or added to the file in such a way that it is impossible for it to be indexed by Google’s “search-within-PDF” feature. Perhaps Google has a size limitation? It is a 109MB PDF after all.
However, any PDF uploaded to archive.org (like its mirror of bitsavers) will receive a second OCR or at the very least a text extraction so that they can offer the document in various formats, like ASCII.
and it’s there in the third result.
When I saw that Ingo had it in his collection at http://ic−ts−histo.de, I knew at the very minimum I could perhaps request it from him (I’ve done this before), and instinct told me to search bitsavers more thoroughly.
I’ve also spotted a document number (240194-001) for a manual in classiccmp.org’s Manx, Intel’s 1988 Literature Guide, etc. I’ve requested it from the Intel Corporate Archives but I expect it will basically be the same as bitsaver’s preliminary datasheet (240194-002).
I’m fairly certain Google does their own OCR. I’ve been able to find documents via Google that weren’t OCR’ed. Whatever, the datasheet isn’t all that interesting 🙂
I did find the document number before, both on classiccmp and on bitsavers.
If your Hercules photo was a little larger and clearer, I would have been able to make out the FCC ID, which may have got it in one. However, it’s not quite detailed enough for me to make it out completely, it looks like maybe it is EW65T5HERCULESHGS. Is that correct? FCC IDs can be very important in properly identifying a card.
(Not my writing but as an example)
Sure, if the photo was larger and clearer, you’d also be able to read the label on the BIOS chip, but where would be the fun in that? 😀
And yes, that’s exactly what the FCC ID is, and I’m well aware how important it can be when identifying unknown hardware. If only every board had a FCC ID on it!
FWIW… yes, Google definitely has some sort of size limitation. While downloading a 200+ MB ZIP file via Google Drive I noticed a message to the effect that the file is too large to be scanned (and therefore might be dangerous etc.). A 73MB file did not produce such message. So that’s the most likely explanation why the datasheet embedded within a giant PDF doesn’t show up in the search results.
I received the 82706 datasheet from Intel Corporate Archives.
I’m not allowed to upload or share it (without permission) but I can confirm that it is 240194-001 82706 Intel Video Graphics Array (Preliminary) from May 1988 and contains really no more or less info than what is currently hosted at http://www.mess.org
Too bad there’s not much there, but thanks for letting us know! BTW according to contemporary reports, the 82706 wasn’t much of a performer. Could be why almost no one used it.
If you’re looking for old(er) drivers, you might want to mirror this BBS I just found:
under /hardware/display/… there are lots of old drivers, some with timestamps from the early 90’s
Zenith Data Systems had a TIGA card in the early 1990’s — my first 24 bit colour system! Cant remember if it was ISA or EISA based. But probably didn’t sell many