The January 1975 issue of Popular Electronics introduced the Altair 8800 and hit the newsstands in December of 1974, so it is only natural that around the New Year people start thinking about the old computer. [Shadowtron] did more than think about it. He ordered some replica PCBs and is building a new one. Even better, he’s posted an amazing number of videos (up to number 56 as I write this) detailing his progress. You can find part 1, below.
The boards are from Trailing Edge Technology. There’s a backplane board (about $100) as well as a few boards to fit it available for about $30 each — unpopulated, of course.
We’ll confess we haven’t watched all 56 videos yet. They cover the construction of the CPU and front panel, the serial I/O board, floppy drives, software, and a homebrew EPROM board (using 1702s, no less).
I have to confess, I’ve always wanted a real Altair — but probably won’t ever get one. But if I were going to build one, this would be the guide to watch as it covers just about everything.
Of course, there are replicas (I have a few and even helped develop one). Plus you can run a great simulation on your PC or even in a browser. But there’s something about knowing you have the real deal — or as close as you can get to it — that makes an intangible but real difference.
after 2 monts kit will shiped on aliexpress ;)
Actually, I don’t miss these old computers at all… I still have my original 2650 (https://en.wikipedia.org/wiki/Signetics_2650) some where and have no intention of every power it up again..
Old computers really don’t age well, and if you want a similar (but better experience) get a current $2 embedded cpu and have a play..
Terrible advice. Starting at this low level will give you a much better understanding of how processors actually work.
It is terrible advice to learn on such a crippled machine. You will learn the wrong way to do things. You should learn the modern way to do things on modern hardware. A neophyte does not need to learn how to burn an eprom or how to enter a program from the front panel switches, these skills are not useful in the modern world.
Modern developers need to be brought up in an environment where secure coding is the paramount concern. This old-school style of learning has produced an entire generation of programmers whose mission in life is to learn bad programming technique and fill our computers with buggy programs. We need a new modern approach to learning about computers.
I find that Arduino, especially on the AVR architecture, to be tight enough to give that feel of the old systems and small enough that one person can understand a significant portion of software running on it. But with the modern conveniences of an IDE, easy to use editor, compiler that gives useful warnings, and quick serial/USB programming.
I’m from the old-school programming generation. Primarily 8086/8088 and Z80 assembler and Pascal coding, which you’d think would have lead me to especially bad habits. But I got through it and am quite able to program properly as a professional C and Go developer.
Coding style, and even structure are sometimes overemphasized. It’s not a habit you build up and can’t shake. It’s a methodology and technique that can learn and choose to apply. In the professional world, your ability to community ideas and document projects is more important than writing tight loops. A mediocre programmer can rise up above a lowly programmer like myself if they have good organizational skills. I’m a better programmer than most of my bosses, but so what. That’s not really important in the big picture.
On the contrary, I think “an entire generation of programmers” have grown up not knowing how computers actually work inside, giving rise to a corresponding generation of software users just accepting bugs as part of life. The scripting kiddies of today don’t know how to test their software before it get deployed, and are too lazy to get weaned off their JavaScript nipple, resulting in adults still suckling on Node.js, writing desktop apps using Electron, and calling themselves “software engineers”. (At least where I work. YMMV!) If a bridge collapses, or a variable-speed (motor) drive product starts blowing up, the engineer/s responsible will probably have legal liability for not doing their job properly. But a software “engineer” can just say sorry, please try rebooting your machine and see if the problem goes away.
If you buy a faulty TV / heater / microwave oven you expect to get a replacement or your money back. Imagine trying to get the purchase price of Windows Server back when an update results in trust being lost on the domain controller or some other design fault. No liability!
I was writing a telephone billing system and having difficulties translating a overmuch. I got some wield code from a Russian expats that worked. When I ask how he came with the code his response was that was how we did it on the Commodore 64 when I programmed in Russia
It looks like they copied the Altair right down to the motherboard made of small sections joined by 100 wires. There were one piece motherboards available back in the day that were a huge improvement.
Having assembled about 30 IMSAIs from kits, I have no desire to build a reproduction Altair, but it does look like an interesting retro project.
I’m the guy building the system. Yes, these are accurate reproductions of the boards. The edge connectors are also the same ones originally used, as I recall the guy who manufactured the boards went to the original manufacturer of the connectors and they were able to do a run form the original molds etc. I have several S-100 motherboards I could have used but as this system will never see anything faster then 2MHz these boards will work Ok.
I remember that issue of PE. It was the beginning of the end for the mag. Not that I dislike computers, but after that issue it seemed that computers were the big focus.
I think the “beginning of the end” for magazines like RE (Radio Electronics) was SMD (surface mount devices)
that made it harder for hobbyists to use and solder.
In addition, firmware; the need to be able to download and install, was beyond the reach of many of their readers.
So, those who really wanted to build a featured product needed to buy a pre-programmed chip from the author.
I might add, that the introduction of new schematic symbols for logic circuits didn’t help either.
But, I could be all wrong, perhaps it was The Internet that made niche tech magazines obsolete.
It was the internet.
One can find lots of projects for free online and even websites that show cool projects like, well….
Hand soldering surface mount stuff surprisingly isn’t that hard.
The Internet is where you can also find all the archived copies of those old PE, RE, etc. magazines too (e.g. https://archive.org/details/RadioElectronics195701/page/n4).
Unless you’re dealing with narrow footprints like TSSOP packages I find surface mount boards faster to solder than through hole. No flipping the board over or having to bend and cut leads.
None of that happened in the 1970’s though.
I built the earlier Mark 8 computer published in Radio Electronic July article (see https://en.wikipedia.org/wiki/Mark-8). I ordered the CPU board and build all the rest including a hex keypad entry system and 2 KB SRAM board myself while in college. I used the computer in my senior EE design project to program a PMOS 1702A UVEPROM that required -46 V programming pulses. My prof was impressed. It took me longer to write up and type my “report”.
As for the S-100 bus, I also have an SD Systems Z80, Ackerman Systems 6809 processor board, a 64 kB SRAM and DRAM boards “sleeping” in my attic. I really liked the 6909’s instruction set back then since everything was done in assembly.
I have mixed feelings about investing time/money into reproductions. I had an S100 system back in the day, and now I have several emulators and “real” retro machines using original chips (8008,8080, PDP/8), that take up much less space, use much less power, are quieter, and so on. Let’s face it, you won’t be breaking any new ground doing this.
The reason that I did the 8008/8080 emulators was to help resurrect old code and also to become somewhat familiar with the ESP8266 and ESP32. I think that I got more enjoyment per buck doing these than I would recreating an older machine perfectly (and it’s still a re-creation).
The piDP8 and the Altair-Duino sort of hit a “sweet spot” of money/size/value IMO.
It’s amazing how sparse those early S-100 boards were, and that you’d need half a dozen of them to get a functional system. Then when the ball really got rolling you got things like 32K static RAM boards crammed in from edge to edge with 1Kx8 24 pin DIPs. Looks like these guys are going more for the early end of the spectrum.
For best effect, be sure to get a 40 lb power transformer and a couple of those 50A bridge rectifiers, and a couple of soda can sized electrolytic caps for your linear unregulated 8V and 16V supplies. Just be sure to use some good heavy wire for that. =P
I’m the guy building the system. Yes, I have several linear power supplies with multi-pound transformers I could use however for now I’m using switching supplies.
Show me a Mark-8 or an EDUC-8 – both of which were true computers which predated the Altair 8800 by several months – and I’ll be impressed. The Mark-8 was 8008-based, and the EDUC-8 was a roll-your-own CPU based on SSI and MSI and inspired by the PDP-8. There is an EDUC-8 in the Computer History Museum in Mountain View.
The EDUC-8 would be especially difficult to replicate, because a couple of the TTL devices used are apparently nearly impossible to source anymore. A redesign would be needed.
Nah, lets torture some board house by ordering re-pros of said TLL chips executed in grain of dust size surface mount components on a PCB in the 20 pin DIL footprint.
I’m the guy building the system. In my personal stock I have a huge supply of original TTL chips going back to the early 70s. In builds like this I use as much original NOS (or pulls) as I can as I want the date codes to be as authentic as I can.
Amazing! Such painstaking work.
I thought I’d provide more detail on the as of 4-Jan-2020. Note that the majority of the system is vintage, not modern clones.
Software – All software is appropriate to the machine and is ran on the machine (no emulation)
ALTMON in 1702A EPROM (Monitor, boot loaders)
CP/M 2.2mits
MBASIC
AZTEC C
Digital Research 8080 Assembler / Linker / Loader
Back-plane
Reproduction of the original 4 slot PCB – using 3 in total
Card guides 3D modeled on the originals and 3D printed (thanks to whoever created the models!)
Floppy Drives:
Two Siemens FDD 100-8 – these are 330K 8-inch floppy drives – From about 1980
Vintage Boards:
CPU
WMC (Wamco) CPU-1 – Intel 8080A – From about 1979
TTY0 (UART)
MITS 88 SIOB – From about 1978
TTY1/2 (UARTs)
SSM (Solid State Memory) SSM IO-4 – From about 1980
RAM (0x0000-0x7FFF)
Compukit EconoRAM IV – 2 boards – from about 1979
RAM (0x8000-0xEFFF)
Memory Merchant MM16K – 2 boards – from about 1981
ROM (0xF800-0xFFFF)
Custom wire-wrapped 1702A EPROM PCB using vintage PCB and components
I currently have no RAM/ROM mapped from 0xF000-0xF7FF) as I’m using this space for development.
Modern Boards:
DRAMP FDC+ (https://deramp.com/)
Floppy Disk Controller – On the Altair bus it’s electrically identical to the original MITS Altair floppy controller. Although it provides additional capabilities I’m using it as a stock floppy controller.
DRAMP Front panel
On the Altair bus it’s electrically identical to the original MITS front panel. Although it provides additional capabilities I’m using it as a stock front panel.
Case (https://altairclone.com/)
Using a case from the Altair Clone. Although it looks much like the original its not identical mechanically.
Power Supplies
I’m currently using two modern switching supplies for the +8 and +/-15 rails. I’m considering using a linear supply as I have the transformers and filter caps that are age appropriate.