• Please review our updated Terms and Rules here

Compaq Portable II, Recreating the 512/1536 Kbyte System Memory Board (104176-001)

Are either of you on Mastodon?
bitbang.social

Brad Bell (@Btb@bitbang.social)

Attached: 1 image I’m missing the Seattle #icf because I caught a nasty cold, but working from bed myself and some friendly folks on the VCF forum were able to completely reverse engineer the rare memory expansion board for the Compaq Portable II...
bitbang.social bitbang.social
 
HoJoPo said:
For the SIMM version, are you going to do 2x1MB parity 30 pin SIMMs and waste 1/4 of the RAM, or use 6x256k parity SIMMs?

Edit: Thinking this over, to work with the three RAS signals without additional logic, you'd need to do the 6x256k SIMM design. Of course, turning those signals into the additional addressing wouldn't be too difficult.
Click to expand...

Yeah, 6x256k w/parity 30-pin simms. However, on the motherboard, there is a signal attached to pin B3, which looks like it should be a tenth address line.
So it's possible that all the logic is already in place to address as much as 6 megabytes, which would be wild.
 
There's another mystery signal from the motherboard on B4, like most of the memory-related logic, it originates on the big LSI chip at U54.

silvervest said:
My eyes are going a bit square, but I've tidied up a bunch of footprints and arranged labels and drawings based on the photos. Looks pretty good!
Click to expand...

It sure does!
That AMP connector also needs some mounting holes. I added some to the file in the repo, but I just realized those aren't just big holes, there's like a rivet that gets soldered in place?. I'll update it again. Speaking of which, this connector might be the hardest part to get a hold of in order to make a faithful reproduction.
 
btb said:
Yeah, 6x256k w/parity 30-pin simms. However, on the motherboard, there is a signal attached to pin B3, which looks like it should be a tenth address line.
So it's possible that all the logic is already in place to address as much as 6 megabytes, which would be wild.
Click to expand...
Stock you can get to just over 4 meg with this 1536k card and a 2048k card in a 16 bit ISA slot. That's probably enough for a 286.

I did notice that the setup program has an option for a 638 meg hard drive for Xenix, so that was definitely a Compaq supported option. I wonder if that was an external drive, hard to think that would be internal half height IDE back then.
 
btb said:
That AMP connector also needs some mounting holes. I added some to the file in the repo, but I just realized those aren't just big holes, there's like a rivet that gets soldered in place?. I'll update it again. Speaking of which, this connector might be the hardest part to get a hold of in order to make a faithful reproduction.
Click to expand...
Yeah, as far as faithful reproductions go, I think I'm ready to let that particular part slide. I definitely could not find one for an acceptable price, so probably just a similar 52p .100 header will have to suffice.

I've dropped you a PR with my design changes to incorporate
 
  • Like
Reactions: btb
merged. I updated the board to use my new connector footprint and tweaked one trace (D11) which was coming off the connector at an odd angle (part of my original sloppy layout) but everything else is the same - do you think you're going to try to build one soon?
 
Yep I have a couple other designs I want to put the finishing touches on before sending off a big batch to JLCPCB. This one's gonna be pricey, it's a big board!
I think I'll need to stock up no 41256's as well, I have some but not enough
 
  • Like
Reactions: btb
I noticed Jameco also has the Am2966 in stock, which is equivalent to the 74S734. https://www.jameco.com/z/AM2966PC-A...-with-Three-State-Outputs-DIP-20_2303095.html

I was never clear on whether it’s always ok to use much faster DRAM than your hardware is specified for. Jameco also has some 120ns. Unicorn has the 150ns chips, US $1.28 each if you were getting >100 (I.e. enough to build two boards). https://unicornelectronics.com/IC/DYNAMIC.html

Of course, utsource has them even cheaper, and while not foolproof, they're better than aliexpress or eBay and have better customer support.
 
I actually have a parts CP2 mainboard, so I've got all the 74 logic and at least one bank of 41256's ready to be harvested. I'm going to have to check my tubes to see what other DRAM I have available.

You can use faster DRAM ICs, they'll just operate at the lower speed. Anything equal to or faster than -15 or 150ns is perfect.

I've had a pretty decent success rate with sourcing ICs from AliEx, only around a 5% failure rate overall I'd say and sometimes the price is just right!
 
For those standoffs, they look like these Keystone ones: https://www.jameco.com/Jameco/Products/ProdDS/28337.pdf
Specifically the "Sturdy Lock-in Release" style.
I was thinking the .875" height version before, but the male side of the AMP connector is supposed to be .785" tall, and it seems like the female side adds no height, or not much? So it is probably the .750" and there's more than enough give in it to make up the difference. Does that seem right HoJoPo?

If you get a chance to measure it with your calipers, just measure from the surface of the motherboard to the top of the shoulder, or if you can somehow measure when the memory board is attached, the inside distance between the two boards.
 
Last edited:
The best affordable alternative for the male AMP connector, is probably what they call "board stacking headers"
bergstik-stack-dr-th.jpg

but the exact height configuration you need will depend on the standoffs used and might take a little figuring.
 
Can't measure with the board attached unless I remove the mainboard from my system, which I'm not going to do.

From the surface of the board to the shoulder of the connector is... 1cm (at the C69 end). From the shoulder to the top lip of the connector is... 1cm. So, 2 cm overall height. The pins are a bit recessed from the top lip, about 2mm, and the base of the pins is at the shoulder height. So a header that sticks up about 1cm and has 8mm long pins should work.

There are no rivets. The connector is not quite flat on mine, some varience in manufacturing. The end near C69 is about 1mm above the board, the end near the outer edge is almost touching.
 
I meant the shoulder of one of the standoffs, if you get a chance.
 
The stand offs have two spring tabs that hold them tight to the motherboard on the side with the memory card. With that spring taken into account, the stand off is about 2cm from the shoulder that holds the memory card above the motherboard.

Edit: Added picture, this one has one broken of spring ear, but was in the best spot to pull.
 

Attachments

  • Memory Card Stand-off.jpg
    Memory Card Stand-off.jpg
    710.4 KB · Views: 5
Last edited:
I've re-installed the memory card, 2176k is working fine after running the setup program. No need to change the jumper.

I have the Compaq EGA card in this as well, along with a 1GB Sandisk CF card acting as a type 42 528MB IDE hard drive in an adapter (card slot type so I can change the operating system). Maybe I'll install Xenix on it at some point....
 
  • Like
Reactions: btb
If one doesn’t want to go looking for a 74S734/Am2966, I bet a regular 74F244 or maybe 74S244 would work fine. I finally took a look at the schematic for the Portable III memory expansion board and it’s really very similar, and uses a 74F244 for address line buffering.
 
I guess that must probably depend on the DRAMs being used really. Populating Z1, R1 and/or R2 might come into play.
 
You must log in or register to reply here.
Back
Top