Thread by @WysWyg_Protogen

I never thought I'd see the day, but we have a new friend, and a VERY important one at that!

Welcome to 1977, and welcome to the Altair 8800B-DM. This computer has been off for decades, so let's see if we can change that and bring back a legendary piece of computing history 🧵

Thankfully, the internals look a lot nicer than the externals do. The computer is pretty loaded up too, though not with anything super interesting. We have 2 ram boards, seen at the bottom, the stock Altair turnkey board above that, a 3rd ram board, 2x floppy, and CPU above that

But wait, isn't an Altair 8800 supposed to look like this? Kinda, but there existed many versions of the Altair. The switch panels were only the initial versions, with later models changing the front design, and even later ones still removing the panel entirely

These later machines without the panel were called the 8800B turnkey, because they just were flipped on and could immediately be told to load from floppy, whereas the older switch panel machine required much more input to do the same thing. Mine though has 2 floppy drives?!

That's because this is the final version. The SM (single drive) and DM (double drive) version integrated floppy drives into the case, removing 6 of the internal slots to make room. These machines were made for Foley's department store, and are referred to as the "Foley Edition"

A few of these -SM versions have appeared, but AFAIK, this is the first Foley DM online! So while this may be the least collectible version of the Altair, it represents the final version. The pinnacle of the Altair's development as a computing platform! How neat is that?

ANYWAYS, enough with the history lesson. This is a repair thread 👨‍🔧

Looking around back reveals this is serial # 15, which is... wow. I'm not sure how many Foleys were made, and I'm sure the number is low, but #15 is really, really cool

I'm so giddy, but taking this slowly. This is a different beast from what I normally worth with. Removing the S100 cards reveals a couple of missing screws annoyingly, but also that the planar is in good shape. I don't see any corrosion on the pins there, which is good

Notice how I said planar and not motherboard. Yeah, this doesn't have a "motherboard" per se. The planar has no smarts on it at all. It's just a dummy board to connect different boards together. I'll get photos of those boards in a little

Anyways, let's look at these cards. We have 2 of these MITS 16K ram cards up first. Nothing special about these. I believe the computer came stock with at least one, maybe 2 of these preinstalled

Then we also have another 16K ram board from MCS, this time in pretty ceramic form. Functionally though it's identical to the MITS branded ones, meaning this computer is currently specked for 48K of ram, which is really solid for an 8800

We have a stock MITS 8800B CPU board as well with the Intel 8080 CPU in it. Again, really typical for this machine. This is the same CPU board used in the switch panel 8800Bs as well

We have not one but TWO floppy controllers for the internal floppy drives. Both are required for the drives to function properly from these controllers. There are some that are single board, but we're not gonna worry about that right now

We also have the Altair turnkey module, which is specific to the 8800b turnkey as well as the Foley spec 8800 units like this one. This also connects to the front panel for those 2 switches and a couple of lights

Examining the case leads us to an interesting discovery. There isn't an ounce of plastic on this thing at all. The chassis, rear plate, bottom pan, and even the front panel and bezel are all powder coated steel, which may contribute to the HEFTY 47 pound weight. Build quality 💯

Well, I guess that's not entirely true. The 2 Tandon floppy drives have PVC fronts on them, and it's actually the only spot of any kind of damage. One of the white plastic pins is missing, and it looks like the mount on the FDD is broken. The drive no longer closes as a result

So let's do a little cleaning. Wow... This thing is dirty AF

Also dammit the 2 switches on the front are broken off, though thankfully there's no damage to the paint. Hey, @systems_glitch, would you by chance have replacements for these? I know you have a lotta parts for S100 machines, so maybe you'd have some spares?

Clean now, but this side of the case has really seen some action. Lots of dents and bends in the case that not only don't look good, but also hinder the ability for the case to close fully. Let's see if we can fix that

Oh yeah that's much better! Still not perfect, but much improved over what it was. This should close just fine now

Took me a while to figure out how to take this thing apart more, but I figured it out. 2 screws on each side lets the blue bit slide away from the rest of it. Cool! Build quality!

In the blue pan, I found not only part of a broken off switch, but also the little plastic piece I'm missing from that floppy drive. The switch is still irreparably broken, but the plastic piece can return to its home inside the drive with the help of a little bit of glue

Another nasty dent here. Got it out somewhat , and you can see that it's better in the second photo, but it's still not perfect. I think it's bad enough that the powder coat would flake if I went much further, so I'm gonna leave it. It IS 45 years old after all. Cleaning time!

You can see the construction of the Foley here at the bottom. The original rails on each side of the case are still there, with a big pan installed to hold the drives, controllers, planar, and other bits. The drives mount like normal to the plate and rest on these big washers

Floppy drives are here. They are both Tandon TM100 single sided drives. The machine uses 80KB hard sectored disks. Really pathetically small, even for the era, but these are the only drives that would fit!

Front panel! Cut up from the standard 8800B turnkey and then silkscreened differently for the SM and DM version. The SM does not have the second cutout, so if you have one of those, good luck adding a second drive. Dunno why they didn't just make a bay cover 🤷‍♂️

Computer is now almost completely torn down. All that remains is the power supply parts

15V @ 95000uF. That's a spicy can of soda there

Finally got everything removed. Getting the rest of these parts out took longer than everything else combined! Flat blade screws and varnish/red loctite is not a fun combo

Bottom pan is clean now! We can start reinstalling some parts as we clean them

Power supply is reassembled now. I'll need to research what the voltages are supposed to be unloaded so that I can see if the unit works tomorrow. I don't see why it shouldn't. It's a very, very simple power supply and one that isn't very prone to failures

Jk smoke test fan go VRRRRRRR

Getting the front panel put back together temporarily. You can see how the keylock switch works by looking here on these 2 switches. All the lock does is move that cam to close those 2 switches. The top one controls the 120V AC input, and the other switch controls the power LED!

Previous owner had soldered in this switch box for controlling either the RS232 serial terminal or a serial printer. The wires have gotten brittle and snapped on the logic board, so let's rewire it so that it can be used as it was originally intended!

and rewired! I didn't run the wires through the case like he did, so I can remove this card should I need to fix it when the time comes to power it on!

Which speaking of, the computer is put back together as much as I can right now. I need thermal paste to install the drives, and the legendary @systems_glitch was able to sell me a couple of replacement switches, so once those come in, we can see if this thing works!

In the meantime, I decided to put the computer back "together" without any screws or whatnot, just to get it in a single piece for easy storage. This thing already looks great! It'll be even better once the switchesare installed, and the floppy drives are cleaned

Oh hey new switches came in sweet! Thank you once more to @systems_glitch for selling these to me! Let's get these soldered in and see what the puter does

Now THAT is pretty

Powering on anD OOP SNAP CRACKLE AND POP OH NO

Yeah I'm a dumbass here, but it's also partially MITS' fault. The 2 floppy controller boards have unkeyed connectors and are hard to see, and I had accidentally plugged 1 in with 1 pin offset off of the connector. Ran those voltage lines right to ground and nuked the fuse 💥

After fixing that though, it SEEMS like it's working, though I can't tell. I think this thing needs a straight through serial cable, not a null modem like I currently have on it. Hmmmmm

I say it "seems" like it's working because I know it's doing *something* because I see activity on the I/O line here on the front panel. @ChartreuseK thinks that this could be the serial line looping as it outputs to the display, and indeed it is supposed to look like this

Following advice from @systems_glitch, I'm recapping the machine before working further. Should've done this from the beginning. Ordered a full set of capacitors for the machine including the power supply and all of the logic boards. Let's start to install these before bed!

One board down, 7 to go + the power supply! Thankfully it seems the leads on the 33uF radial caps (axials were unavailable) are just long enough to fit and look good, so yay for that

I'm getting good at this now! These radial replacements do still look quite good when I bend the leads up nice and pretty. Adding a big piece of heatshrink to caps that now touch other components due to their increased size just for safety. Looking good!

Yayyyy all the S100 boards have now been recapped. There's one it seems I missed on one of the floppy controller boards, but hopefully that won't matter. Everything else went really smoothly

Floppy power boards have also been done now. 4 caps each, but this was not fun to do, especially with all that white thermal goop getting all over the damn place

5 caps whoops can't count. Anyways there's only 4 remaining caps, but they are in an IMPOSSIBLY hard to get to spot, and removing the board they're attached to requires almost complete disassembly of the machine to do so. *sigh*

After an intense match of swearing , unscrewing, photographing, and wondering why I thought this was a good idea, we have the last piece of the puzzle out once more. Let's get those last 4 caps replaced

Yayyyy replaced and it only looks kinda silly. These new caps are a lot smaller than the old ones are due to technological innovation. Still looks ok though

Time to get it reinstalled into this disaster ugh. Do not like the design of this

Ok ok nothing went kablooey so that's a good start!

We have halt and reset! Didn't have that before actually! Let's get a terminal on this thing and see if we got lucky!

WE HAVE LIFE!!! It's in a VERY barebones state right now, but we're able to go through and jump through ROM addresses. Writing to RAM isn't working though, and attempting to load the bootloader seems to crash the computer with the FDD cards installed, BUT STILL IT IS ALIVE

Doing some more investigation, it turns out that the computer had no RAM, hence why nothing was writing or reading. 2 of my 4 RAM cards are bad. One of them gets super hot (short on 5V somewhere), and the other just doesn't seem to do anything (no clue)

The other 2 though seem to work! When moving to address 0000 (M0000), we're able to save the hex value C3 to that location, then recall it. Same thing happens at address 4000, which is where the second ram card resides. We have 32K working now, at least mostly

One thing that isn't working though is the boot rom command. Typing JFF00 (rom address) should jump and execute the boot rom, but it just hangs the system. It's supposed to hang it, but it's also supposed to turn on the floppy drive and look for a disk. It is not doing that

Flipping the data cable to the floppy drive though does show that the LED does work, and the spindle motor does move slightly when power is cut, so no issues there with the motor spinning. Something just isn't quite working right

The computer isn't halting. I/O is still on, meaning the CPU IS working and looking for information. Somewhere between the CPU and the floppy drive though the information is being lost

This is essentially nerd speak for "It's mostly working but there's something small that's just not quite right and preventing the signal from getting all the way through", which is good. I'm gonna keep going through fundamentals and make sure everything looks proper

Bad connector resolved and oh my god it's looking for a disk now! This is working!!! I need to clean a drive now and see if one of the boot disks still works after all these years

Floppy drive time. Taking the board off, we can see the spindle there at the bottom left as well as the head (yes singular) at the top. The spindle already moves freely, but the head seems to be stuck

Ugh that is completely stuck. The bearings in the stepper motor seem to be seized up from sitting

After a lot of lubrication and wiggling back and forwards, it seems the actuator is moving again. Attempting to load the bootloader moves the head to track 0, and you can see it move here!

OH

IT JUST WORKS

PLAYING STAR TREK ON AN ALTAIR YES

Second floppy drive comes to life! We've got dual drives now in the machine! This one has that broken door though remember, so we'll need to fix that! Electrically though it's perfect!

Glued the latch on the secondary floppy drive back together and reinstalled the pair! Still working, and the machine is starting to take shape again!

All together, finally! Apart from the hotwired power switch and the 2 broken ram cards, this thing is fully functional again! Isn't it gorgeous?

Here's a full boot of the machine initializing, running a couple of basic memory checks, booting basic, then running a simple "This is an Altair 8800B-DM!" line of text over and over again! Works absolutely perfectly I cannot be more thrilled!

Glue has set up on the damaged TM100 and I'm glad to say it seems to be holding perfectly now! I can now use the secondary drive like normal once again. May be getting the key along with an additional ram board tomorrow, so I'm excited for that. Can finally undo my hotwiring

Good time to remind y'all that the machine is on permanent loan to the museum, rather than a donation, at least for the time being. The owner of the machine is in his 70s now, but still remembers EVERYTHING about the machine. He's been an awesome resource for this project

I was actually unaware of this 4K limitation in regards to the turnkey monitor. I will very much have to investigate how to disable that part of the memory so that we can run all 4 boards in the machine. It shouldn't be too hard

Key acquired and it works perfectly with the replacement microswitch installed! Turning the key 90 degrees turns the computer on. Turning 180 disables the toggles so people can't mess with your computer in a public setting. Can see the start switch behavior change!

So 64K is apparently kinda silly, and really annoying to make work on this machine, so it may remain a 48K machine, provided I can get this other SRAM board working. Thankfully, I do have a working one that I can probe alongside it to compare behavior, so hopefully this is easy

Looking at the chip select pin, which should in theory trip whenever the machine writes to memory, and we already see some interesting behavior. The bad board seems to have a much shorter and later pulse than the good board on power up, but that shouldn't matter on startup

Annoyingly though, neither board seems to do anything when memory is written though. I suspect my probe is maybe not sensitive enough to catch it, but I dunno. One board does write correctly, and the other does not. Bad board doesn't even detect memory at all

Forgot I also had a logic probe. Poking the /CS line when accessing DOES show a pulse, which is good. I did also swap the single socketed IC there in the middle, and it too is good. We need to look deeper for the fault

Poking on pin 15 of that chip is a data output pin. As I toggle through addresses with the space bar, you can see it pulses sometimes, but not others. That means there IS data in SRAM, but the computer isn't able to see it

This is because SRAM powers on in an unstable way. The values in each address will be random unless cleared, so that pin intermittently pulsing means it's seeing that garbage there. Something is a pulse, nothing (FF) is not a pulse, so it picks it up sometimes, as it should

Moving downwards to these 2 buffers, we should see pulses on the control pins for each buffer, but we don't. I don't see it on the input side OR the output side, which means these chips aren't bad, but whatever is driving the chips isn't... well driving the chips

What drives that chip is this 74L04N inverter named "L" in my schematics. While it did initially test good, poking all the signals did show that it was not inverting the signal properly, so it must be replaced

Thankfully, I have several battery bombed boards that are beyond saving, so I can look through those to scavenge parts. 7404 series inverters are very common chips, and sure enough there's a suitable chip I can desolder to replace the bad one. Hopefully this one works

HEY! We've got life now! The memory board is finally reading its memory properly through all addresses! However, when trying to write a block of memory (See 0004 00 changing to EF), the value does change sometimes, but not to the right value. Something is wrong still

Checking the write signal did show that the write pulse is reaching the chips, which means that write IS working, but it's just not writing the value it should be. It's writing garbage or nothing at all. Examining schematics more shows that it runs through inverters U and H

BOTH U and H failed my tests, having intermittent or non-existent signals, so we need to steal 2 more 7404 series chips to replace those too. Dang this thing was messed up!

It's closer still! We're now able to write SOME addresses. You can see at the bottom though that we're still failing on address 22, and there's several others that seem to fail too in the same relative location, so I think we've got 1 or 2 more faults somewhere to sniff out

It is worth noting that the card currently is piggybacking these 2 inverters with most of the ram chips removed, so let's actually replace them and work from there. Maybe it's just an artifact of the pieced together nature of the board at the moment

Nope. Still being weird. Let's look for patterns though. Writing EC does not work at all until we get to address 6, at which point it starts to take the write all the way down through address 20, but look at this. Notice how at address 18 it ALREADY HAS what I'm trying to input?

This isn't coincidence. It doesn't take writes until address 26, works fine through address 40, but then copies the value, this time BA, into the rest of the memory at 38, exactly where it did last time, relative to the first block. This is a consistent behavior, if bizarre

Something appears to maybe be "bleeding" the signal to other memory locations, which may imply that whatever chip determines what address goes where may also be defective, along with the continuous issue of it not writing properly, though just to low address ranges it seems

While we take a break, @SuigintouLain had a Televideo 912C kicking around that they were willing to sell for a really nice price! It works perfectly too! Just needs some cleaning and minor tweaks hah. This thing is really coming together

I just need to get that damn ram card working again. I think I might take the nuclear approach and just replace ALL the logic chips. There's only like 13 of them, and I can't imagine the cost of parts would be particularly high

SPEAKING OF, THAT RAM CARD IS WORKING AGAIN! 4 inverters were bad, 3 on input and 1 on output! We've got our full 48K of ram now, though for some reason BASIC isn't seeing all of it, or it's somehow using 22K of ram. Either way, it's working enough for software debugging now

After much fighting, FLOP2PC is running, which is a program that images Altair format diskettes over serial. Slow as dirt but it works! We're now imaging 45 year old software so it can be recreated perfectly in the future

The process is quite slow, but it does work. I've got about 100 or so disks to scan in, but they will all be documented and uploaded to archive.org with permission from the owner of the computer and software. Some software he wrote himself!

I'm doing 2 reads per disk in the left and right drive since I can't test the disk image files. I don't know if these disks are good, but I'll know that they're at least being imaged properly if the disks are identical on both reads 🤷‍♂️

I was mistaken! There is an Altair 8800 Emulator hosted on altair32.classiccmp.org/Altair32code.h… that works exactly how I want it to. I'm able to boot the disk image file I wrote of the Pertec Basic disk, and it should allow me to test the other disk images for proper functionality!

And yes, star trek does indeed run on the emulator too, just as it did on the actual machine. This disk is 100% good to go, so let's get back to imaging the disks

Here's where stuff gets super cool. The owner of the machine was an independent software developer back in the late 70s and early 80s. He wrote a game called "Hunt the Wumpus" back in 1982; a game he says was derived from Zork, where the object is to locate and kill the Wumpus!

This is quite possibly the most nerdy I've felt ever, but I'm super happy these disks are able to be read! I'll work with the owner to see if he's ok with this game being published online for others to play with!