Inside the 1977 KIM-1: CPU, Video, RAM, ROM and more!

Dave disassembles the oldest home computer on eBay, the venerable 1977 KIM-1 system, and explains what each of the boards does and how the backplane interconnects them.

For information on my book, "Secrets of the Autistic Millionaire":

My other channel, join now so you're there for episode 01!

Primary Equipment (Amazon Affiliate Links):

[lwilton]

That 276- part number is a Radio Shack part number. They made a number of prototyping boards of various shapes and configurations through the 70s and 80s. I remember hand wiring quite a few boards much like that little IO expander and considerably larger and more complex. I'm sure thousands of other hobbyists did the same in that era.

[timthezombie]

Wonder where we would be without Chuck Peddle. I love the content Dave! I was an unusually recluse, but curious child growing up. Your channel has brought me nostalgia for late middle 90s-early 2000s. I always look forward to your videos!

[johnsimon]

Personal Computing in the 70’s is a lot more of an extremely advanced home electronics project than it is what we’d recognize as computing today or even what’s considered computer use in the 80’s

[gumbykevbo]

On the local regulation power distribution scheme: In those days the chips were drawing a fair amount of current on average, and spikes (glitches) when they switch. (hence all the bypass caps). Now if you try to send 5 V through the backplane bus, you are flowing that current, and the glitches through a bunch of skinny traces, wires, and connector interfaces, all of which add a milliohm here, ten millohms there, and inductive impedance all over the place. (any and all wires have inductance)

Mr. Ohm tells us that when you flow current through all that resistance and impedance you are going to get voltage drop. If this voltage drop becomes excessive, then one or more chips becomes flakey. Maybe it only happens on a read if all the data lines change state at once, or maybe it is temperature dependent...flakey hardware will drive you nuts trying to track down the problem.

Now all those little caps will help a lot with glitches. They limit both how much voltage change a glitchy chip puts on the power rails, as well as softening glitches from the neighbors so the local chip is less likely to be influenced by those glitches.

The voltage drop due to average current consumption can't be addressed by capacitors, because they only deal with transients.

Say the current load on the power bus is enough to cause 1.5V of drop on the 5V rail and 2V drop on a 12V rail. Well, the 5V rail dipping to 3.5V is a problem for a couple of reasons. First off, the chips just might not work at such low voltage, or work at reduced speed. But even if they would work fine on low voltage, you have the issue that the cards nearer the power supply (electrically, if not physically) are seeing far less drop, so maybe they are running at 4.5V. That means that their high logic outputs are sending very close to 4.5 volts out to those distant cards that have only a 3.5V supply. And the chips won't like that AT ALL!. Driving inputs (or tri-state outputs) above the supply rail can cause a host of evils. If the difference is less than .5V it usually works OK, because it takes about 0.6V to overcome the junctions that are trying to protect against this.

So you want the all the cards to see the same supply voltage. The method of achieving this on your boards is local regulation. The cards are supplied with a raw supply of, say, 18-20VDC. The same rules of voltage drop apply, and because the 12 and 5V current both flow through that supply route, the drop might be as high as 4-5V. So you end up with 13-14V at the farthest card. The local regulators drop this to putch-near exactly 5V and 12V on the cards, and life is good. The local regulators compensate for however the raw supply voltage varies. The other thing is that while switching regulators technically existed in those days, they were exotic beasts, and you needed a true guru to make one that worked at all, and a freaking magician to make one that worked well. So all the early computers used linear regulators, which generate a lot of heat. By putting lower power regulators on each board, you spread that warm warm lovin' over lots of area, which lowers die temperatures, which keeps the smoke inside the regulator pass devices.

We had to do the same thing when I designed a board for the DEC Alpha. That sucker needed a ton of current at 3ish volts. So we'd put 5V on the board, and that fed a voltage regulator that lived right next to the processor. Pretty sure all the really fast processors still have to do this.

[nesmaster]

So many bypass/decoupling capacitors :)

[surferdude]

I think it was 1976. My school bought a Wang computer. It had: 12k ROM for the OS and Basic interpreter, 8k RAM, two 128k 8 inch floppy drives, input console CRT monitor and keyboard, mark-sense card reader, dot-matrix printer, letter quality printer (looked like an automated IBM Selectric typewriter). The whole thing cost $25k. We had to sign up for time on the computer after school hours. It was a lot better than shipping off mark sensed cards to be processed overnight on the one IBM computer in the school district. That was state of the art at the time.

[mikekochanek]

I love the Mercedes wheel "table" cool blend of Top Gear, Roadkill and Curiousmarc. Your videos are awesome, btw!

[chswin]

The architecture, with the backplane layout, seems quite sophisticated and forward thinking for the time. Just fascinating!

[kgxxx]

I got a KIM-1 as my first computer back in 77. Added a 4K memory board and a TVT video interface. Then eventually expanded it to 64K with video card (80x24), rom based monitor. Never got around to adding in a floppy disk controller. Still have it, but had to replace keyboard so it is no longer stock. I did end up building my own CPU card to replace the KIM one on my home built machine.

[JonBailey]

Dave, you did a fantastic job of production on this one, crystal clear, and you did an respectable clean up job on the top layer to show off the mystery chips!

[dvogel]

Dave you seriously have the best background music. Your videos are both entertaining and soothing at the same time. Best way to end the day.

[InventionTherapy]

Those white pads become conductive as the rubber begins to age and deteriorate. I have seen them add resistance and even shorts to the PCB. Remove and clean the surface to save yourself headaches.

[rickdearman]

In Byte Computer magazine (Volume 4 Number 9, September 1979) there is an article called "Interface a Chessboard to Your KIM-l". If you can find a copy it is an interesting read and was written by: Jeff Teeters. Mr Teeters made a chessboard with a load of wires which he fed into the Kim I in order to program it to play chess.

[GodmanchesterGoblin]

While the disk controller may have been intended for 8 inch drives, that NEC 765 chip was *very* widely used for many years and that included plenty of use with 5.25 inch drives. The compatibility is going to be down to the software drivers and other hardware considerations, including whether the disks were intended to be soft-sectored or hard-sectored.

Oh, and that hand assembled board - could it have been an early SASI/SCSI host adaptor, or is it too early for that? Maybe for some kind of tape drive? I didn't get to count the number of ways in the ribbon cable. And that's not wire-wrapping - that's pretty standard hand soldering for digital electronics in the late 70s - early 80s. If you're a hardware guy, that's a pretty simple board to wire up. For wire-wrapping, you'd have IC sockets with very long square cross-section pins, and the wire is wrapped very tightly around the base of the pin for 8 - 10 turns for each connection, using a specialised hand tool. The tightness of the wrap ensures a good connection, such that no soldering is then needed. (I still have a hand powered wire-wrap tool in my box of 80's stuff.)

Last thing - you're not the only guy that straightens out decoupling capacitors. It's a "not quite OCD thing" for me too. Just be careful with those older ceramic disk caps - they don't like the leads bending close to the capacitor body since it can fracture the coating and then moisture can get in which doesn't do them much good.

[ChrisP]

FYI "bodge" is an Australian term for fixing something in a very messy or hacky way, i.e. bodge job. We pronounce it boj, like bog but with a j sound on the end instead. Loving the classic content!

[anotherfellasaiditsnunya]

Brilliant visuals and narration! Thanks

[macfixer]

My first guess is that the prototyping card was a parallel interface for a printer? Although it could’ve been for something like a paper tape reader, but that would seem kind of pointless since they apparently had a disk drive connected. I’m pretty sure that same 765 chip is also compatible with 5 1/4 inch Shugart drives. That chip was used on the original IBM PC/XT floppy controller.

[kencreten]

Whenever I see you have a new video, Dave, I usually say to myself, over dramatically, WHAT? WHAT?!" The next 12:48 of my life is set in stone.

[PeranMe]

Love that wheel table! :-) Thanks for continuing the KIM videos, this is so interesting!

[ejharrop]

Awesome video. Much like looking at the Wright brothers equipment in their early days of flight. Very cool thank you. Take care and peace out!