Thread by @TubeTimeUS

new arrival at the Bench of Investigation: an Exatron Stringy Floppy drive!

complete with a controller board!

despite being called a floppy, the media is actually a very thin tape. Foone has discussed them before, and has some of them!

https://twitter.com/Foone/status/837905521210355712

off with the cool wood-grain panels, and we can see the circuit boards inside.

wow here it is!

hmm, this upper board is an "up date," maybe some sort of decision? also the designer seems to have signed all the boards (lower left)

looks a lot like a tape drive, lol

it's also missing the belt.

and the drive head is filthy

looks like the mechanism is made by Microvox.

so the controller card has some modifications.

the wires had popped out but I'm pretty sure they go in this way

i'd like to dump the EPROMs but they're 2708 and 2704 devices and I don't have any programmers that can read them. I suppose I could breadboard something.

this'll have to do.

oof, board has those terrible TI sockets

well, the dump looks good. ROM says "Howdy!"

now i just need to figure out the architecture it was designed for.

maybe 8080? the z80 also has pop instructions with the same op code.

unplugged one of the mods. the chips are a 7492, 7493, 7493, 74LS02. the '92 and '93 don't have VCC and GND pins in their normal corner spots!

here's the mod. it takes a 2MHz clock and divides it down. there's also a reset/sync feature.

got the bus interface figured out. it's S-100! and it only uses a minimal set of pins.

this version of S-100 annoys me, it has 8 data inputs and 8 data outputs. they're not bidirectional. apparently that came later, so instead, this card uses these special transceivers.

control and communications to the tape drive is done entirely with this 8251. but they don't seem to be using it as a shift register, but basically as a GPIO port.

there are actually 4 ROM sockets on the board, but only 3 are used. the one on the far left is occupied by the weird bodge i showed earlier. i think it's SRAM! the bodge wire is the CPU's write control line, since that normally isn't pinned to a ROM socket.

now i'm wondering what U2 and U3 used to be before the clock bodge was plugged into U2. (U3 has no chip in it).

so the 74LS86 and the 74C86 have different pinouts. ridiculous

sadly this is real, it's not photoshoppery. 😔

anyway, here's the schematic so far with the 74LS74 installed, which i think is one of the mystery chips. the RxC net goes into the 8251 receiver as the baud rate clock. "rx data" gets inverted and goes to the RxD data input of the 8251. based on that, i think i can figure out U2

looks like it is part of an older data separator design, at least for the receive gate. the "clock bodge" also has a data separator, and it samples the signal using 64us bit cells. flux transitions closer than 64us are latched as a 1, and longer than 64us are a 0.

hmm, something isn't right. to make the transmit section work, the gates circled in red need to be NAND gates--anything else gives you no output. but in order for the transmit clock to work (tied to RxC), the gate circled in green needs to be OR or XOR.

the only clue we get is that the drive itself may allow write pulses to get through the read amplifier and come back out. this *could* wiggle the other input of the logic gate and get it to reset the one-shot that generates the transmit clock. so it *might* be a 74ls00.

all right, my guess is that this circuit never quite worked right, which explains why the bodge looks factory-made (wires are glued down with epoxy, and other bodge wires on the back are exactly the same way, including ones that fix obvious fab issues like crossed traces)

in other news, @Foone has loaned me a stringy floppy. it's 5 feet long. well, the tape loop inside is.

what's that you say? c'mon Mr Stringy Floppy Drive, don't talk with your mouth full

so yeah the Stringy Floppy is like Fruit by the Foot but for data.