Altair 8800c S/N 13256

[ef1j]

Hello All!

I've been lurking from time to time and benefitting from the discussions. I finally registered.

I recently grabbed a case from Mike after playing with Altair BASIC VER 1.0 on an IMSAI and teletype. It felt like perfect timing to work with an Altair, and one thing I haven't done with S-100 systems is build a computer from scratch.

I got the machine up and running after a few weeks of off-and-on building. What a great project! I want to thank Mike for doing the work to create a case and front panel / interface card. For me, it's a perfect combination of old and new. My IMSAI is usually running CP/M 1.4 with two 8" disk drives and an ADM-3A terminal. I'd like to keep the Altair more bare bones running with the teletype. Sort of like what Paul Allen might have encountered when he flew to Albuquerque. (Minus the flakey DRAM.)

There are still a few kinks to work out. My machine is starting up with a low voltage on the CPU board regulator. The 12V output also starts low and slowly increases. As it "warms up" the clocks start intermittently, at about 500kHz. After a few minutes, everything "locks in" and it becomes steady. Bad cap? Wrong resistor somewhere? The only boards I'm running right now are the CPU, interface board, and a Processor Tech 4K RAM. Other than startup, everything seems solid so far. Once it's finally running, I can deposit / examine, etc. Still have some work to do to figure this out!

Eric

[Wayne Parham]

Sweet!

I love your system!

Welcome to the forum!

[ef1j]

Thank you! I was in no hurry. It felt great to get it as tight as I could.

[AltairClone]

Looks like good progress so far!

Mike

[ef1j]

Thanks, Mike!

So here's something interesting that I've seen: my machine initially took some time to start up. On power-up, there was no clock, then the clock would start, but intermittently. After about four or five minutes, the clock would lock in and everything would work fine. I also noticed that the voltage on the CPU card's 5V regulator would start off low (~3.8V) and it rarely got above 4.86V.

The voltage on the interface board was fine, so I swapped out the regulator on the CPU card. I had installed an "original" Motorola MC7805CP (date code 7702, it looks so good with the gold pins). (I wonder if the 7.5V on the 8V line is too low for this device?) I swapped in a newer (1984!) LM340T5. Now the 5V line comes up right away on power (it's 5.1V). However, the clock is giving me trouble now -- it doesn't stabilize. I'm sure I'll have to adjust R41 and R42 as has frequently been discussed.

This makes me wonder about the sensitivity of the clock to the power supply and regulator (and how the heck hobbyists in 1975 figured any of this out). My guess right now was the steady voltage was low enough that the phi1 and ph2 clocks met their timing requirements, but when the voltage increases, the cycles are a little bit closer together. (I'll have to make some measurements to test that idea.) Maybe the original Altair's unregulated supply with the linear voltage regulators were just a little low -- enough that the R41 and R42 values worked?

I'd like to understand the clock circuit a little better and why it would become unstable. How does the original 8800 start? Why do the R41 and R42 values work there?

Eric

[AltairClone]

Once oscillating, the clock (the oscillator itself) is very stable independent of reasonable variations in the power supply. R41 and R42 have nothing to do with the startup or stability of the oscillator. Instead, they simply determine the duty cycle of phi1 and phi2 which are important for the 8080 itself.

If you’re using a 74LS04 in the oscillator for IC-P, you might try a 7404 instead. If you have multiple 7404s, you may find one that seems to startup more reliably than the others.

Mike D

[ef1j]

I see! Now I have a better appreciation of the clock circuit. It's strange, though, that the clock was turning on and off at about 500kHz. I assumed that one of the one-shots triggered the other and there might be some sort of aliasing occurring. The MITS CPU card is a finicky thing!

Anyway, last night I adjusted the R41 and R42 values by bodging on two resistors in parallel to get the overall resistances in line with the values cited in other threads. Everything seems to work fine now (40 minute burn in!) so we're off to the races! It's a slow race toggling in those programs, though. Looking forward to getting it talking to the teletype.

I haven't seen an original MITS CPU with different values of R41 and R42 from those in instructions. (Sample size N = 2.) I'm using mostly 74L parts (not LS). What is different about a modern build versus those from the 70s?

Eric

[ef1j]

Quick update: after replacing the 7805 and bodging R41 and R42 to bring the clock into spec, this 8800c has been rock solid. Earlier, I built a serial card that mimics the MITS 88-SIO REV 0 (and REV 1) based on the Solid State Music IO-2. The SSM IO-2 is a prototyping board that provides addressing circuitry. The manual contains the directions to build the serial interface using an AY-5-1013 UART. So far, I've tested simple loopback programs that are toggled in to the front panel. The Altair 8800c is connected to an ASR-33 teletype.

[Wayne Parham]

Super cool stuff!

As an aside, I loved that AY-5-1013 UART. I built an XON/XOFF to hardware (DTR/DSR or CTS/CTS) handshaking converter with them. Didn't need a processor or microcode 'cause the AY-5-1013 was designed to be used with discrete logic rather than a buss. I guess that was both blessing and curse, since a UART for a microprocessor is naturally easier to use if it has data buss and addressing logic.

[ef1j]

Wrote up some build notes here...

https://ef1j.org/wiki/index.php?n=Main.Altair8800c

One thing that still puzzles me (besides the clock timing) is the change in the serial specs (specifically, the status word byte) between the 88-SIO revisions and the 88-SIO2.

E