• Please review our updated Terms and Rules here

Help debugging Altair 8800

R

rbgrn

Experienced Member
Joined
Apr 5, 2013
Messages
66
Hi All,

I'm not a super EE guy but I learn what I need to and try hard to do things right and understand what I'm doing for the most part. I picked up an Altair 8800 about 6 or 7 months ago that had been gutted and rewired for a different bus (non-s100) and had a regulated PSU (not one that could actually work with S100) and the front panel had been reconfigured in about a dozen and a half spots. I spent maybe 50 hours or so restoring the front panel to match the schematic (AFAIK) and re-attached the original S100 bus to it. I also rebuilt the original PSU as I found the original PSU board and transformers.

So now I have what should be a working Altair 8800, however I'm at the part where I can't figure out what to debug next.

With just 2 cards: MITS 8080 rev0 and MITS 4k dynamic ram rev0

What works:
Examining an address does move the address LEDs.
Examine/next does increment address LEDs correctly.
Run/Stop/Step/Reset all seem to function correctly.
All LEDs are functional in that they do turn on so I don't have any burned out or misconnected.

What does not work:
D0-D7 are constant on.

It looked like the base address of this dram card is at 4096 so I tried starting there with deposits but still nothing. I'm assuming sold D LEDs indicates that RAM is not working.

Could someone please tell me where to look next or what debugging I could do to help get more info? I'm just not sure what the next steps are. I have not checked to see if the capacitor values on the front panel have been changed to match the spec of the MITS dram card, though that should only cause intermittent deposit error IIRC and would not result in solid D indicators.

Thanks!
 
Find something other than that 4K DRAM card to debug with. The first iteration (RAMs across the top of the board) is an absolute mess and I could never get the second iteration (RAMs down one side of the board) working reliably with a front panel. If you don't care about it being a MITS board, there are plenty of moderately priced static RAM options out there, and for a little more money you can even get one that's fully tested.

Generally I start with either a "Halt ROM" or a board I've built to pull the Data In bus to values specified on a switch. The latter gets address decode problems out of the way.
 
I'll second that on the 4K DRAM board! I had 2 and, while i could get mine to work, contents would often evaporate after a reset.

If you're determined to figure out the why of your failures, start by checking the output of the voltage regulators (+5, +12 and - 5). Some were overworked and a failure wouldn't be surprising.
 
I agree that it's very important to be working with a fully functional RAM card because you're not 100% sure you have the front panel working correctly until you have a good RAM to test with. Or a 8080 ICE unit. Also, simple stuff like, check the voltage regulators on the RAM card and find a schematic for the card to trace the basic stuff.

are all of the power supply caps new or new-ish? Did you test the voltages on the backplane? (I assume yes, but just asking)

May be that the RAM card jumpers are shorted open/closed and may look to be set for 4000 page but are not actually set. Check first 20 or so RAM addresses for every RAM page starting from 0000 through FFFF, just in case.

Try setting the RAM address on the card to 0000 anyway, it'll be easier on you to test and toggle in little programs you find on the web that are often set to be run within the 0000 page.

Maybe you need a terminator card? Backplane may be freaking out for whatever reason. What if you move the RAM around (next to the CPU card, two down, three down, etc...) Do you detect voltage changes on the backplane when RAM is installed? Should go down a little but not too much.

Getting nowhere? Here is a RAM board that will work in your system (Ebay) - 151084227079 (Wameco). Most of the rest that I saw on ebay at the moment are "who knows" condition, not worth the money.

A lot of people check their systems by running the Altair music program by toggling in the code and a short song. This allows you to test your Altair without anything other than the CPU card and a RAM card set to 0000.
Read here:
http://vintagecomputer.net/cisc367/dr dobbs feb 1976 Music With Altair.pdf

Bill
 
Can I state the obvious? Have you tested that the front panel data LEDs actually work? If not, I would suggest removing all of the S100 cards and going along the S100 control, data and address lines connecting each one to first +5V and then to ground (you will need a 5V regulated supply for this - not the unregulated S100 +8V supply. Check the front panel schematic for which S100 pins are which indicator.

I looked on the schematic and noticed that most (all) of the address and control lines should light the associated front panel LED when connected to +5V and should extinguish the LED when connected to 0V (GND). I also noticed that the data LEDs seem to be buffered by 74LS04 logic gates - although they still work the same way (0V = LED OFF, +5V = LED ON). This will give you confidence that the LED indicators are working and connected to the correct S100 pins.

You can then look further afield at the front-panel logic to see if it is actually producing the correct signals for the CPU/RAM (you will need a logic probe or multimeter for this. A logic probe can be something as simple as an LED with a series current limiting resistor if you want - although you can make your own to detect positive and negative going pulses in addition to voltage levels).

Give me a short while to review the schematics for the control section of the font panel.

Dave
 
Hey guys,

I did a ton of debugging yesterday and here are my findings so far:

First I found that one of my two MITS 4k dynamic RAM cards almost works it seems with the 8080 CPU. With A14 high (as the card is jumpered for), I can protect/unprotect and the machine will actually correctly hold address lines for any selected address. When leaving the 4k page, the protection will shut off if it was on.

I took my logic probe to the bus for checking every address line to verify and all 16 work when toggled and examined.

I then checked the +16 and +8 voltages. With MITS CPU and MITS 4k in, they are at +17 and +9 - so that looks good. When I put a Z16 card in, the 16 line drops down to around 1v - so either it is overdrawing or there is a short. I've seen similar behavior with other 16 and 64k cards I have. Is this just a weak PSU? I didn't think the Z16 would draw that much current.

I have a modified ZPU card but it looks like the previous owner was adding TDL lines for A17-A20 and there are some modifications to the ZPU and one modification to the Z16 - so they aren't stock and I'm not sure I can trust them.

All my LEDs work - I've confirmed that. I actually saw activity on the data LEDs when I was messing around with protect on - so it still seems I just don't have a working RAM configuration yet and that's probably the biggest problem.

Any ideas on why my voltage would drop so far? How can I debug that besides sticking the card into a different machine (Which I have a few other S100s but i haven't setup yet).

Thanks!
 
Get out your ohmmeter and measure the resistance of the output of the regulator that takes the +16 down to +12 for the Z16 to ground. If it says there's a short, then suspect the capacitor on the output side of the regulator. It could also be the regulator--remove it to be certain.

Although the MITS 8800 PSU is pretty wimpy, you shouldn't be able to take the +16 down that low with just a single card.
 
Agreed - Voltage regulator or shorted capacitor(s) (or both).

I would also strongly advise checking the regulators and power lines with an Ohm meter before first powering on anyhow. I had a shorted 12V regulator (input to output) on a card once so that would have put the full +16V on the card's +12V rail - not doing the semiconductors any good in the process! I had the sense (?) to use a variable DC power supply with an in-built current limiter. I increased the voltage and monitored the regulator's output voltage. Once I had exceeded +12V and a bit at the input side and I still saw the output climbing I suspected something was wrong. I de-soldered the regulator and tested it on a bit of breadboard with a simple load and sure enough it did 'nada' in terms of regulation" That saved a blown disk drive as a result.

The recommendation, be very careful if you have boards of unknown origin as simple power supply regulator faults could cause much more damage to the semiconductor devices on the card. Be on the safe side and do as much checking as you can without power applied - then power up in a controlled manner.

You could try putting a 12V and 6V car lamp in series (18V) (or even two 12V car lamps) and running them from your +16V line. Chose a bulb wattage suitable for the current capability of your power supply line (Wattage = Voltage * Current). If the lamps light OK and the power supply is not in distress - then your card is at fault not the power supply.

Dave
 
Hey guys,

It was a shorted capacitor. Pretty obvious once I got in there and checked - near 0 ohms across it. So I removed that and also put on a fresh 7812 for good measure. I don't have any replacement caps in spec (It wasn't electrolytic AFAIK but still had polarity markings on the board?). 4.7uf 35v I think is what I need so I'll have to get a stock of those as I think that's a pretty common one for a lot of these boards. I plugged it in without the cap and voltages were good. Not doing any more testing than that until I get that cap on.

So there is a little rewire on the back of my Z16. Pin 60 was moved to Pin 67... Not sure how that would affect anything with the 8080 CPU since all those are in the unspeced range.

There is also one other jumper on my Z16 on the back that isn't factory.. U7 pin 1 (SMEMR) to U3 pin 5 (PSYNC). What function does that modify do you think and how does that impact the 8080? That wasn't required to run on the ZPU was it? I always though the Z16 worked stock with a stock MITS 8080 card as well as the ZPU and IMS CPU cards.

Thanks
 
Last edited:
rbgrn said:
What function does that modify do you think and how does that impact the 8080? That wasn't required to run on the ZPU was it? I always though the Z16 worked stock with a stock MITS 8080 card as well as the ZPU and IMS CPU cards.
Click to expand...

Yes, it works as a standard RAM board with most CPU boards. I've tested mine with a Byt-8 MPU, Altair rev A CPU card, Cromemco ZPU, Vector Graphic 8080 board, Vector Graphic ZCB, IMS Z80 board, and probably more. Don't remember if there are patches on the back of mine, but I don't think there are.
 
If it is a polarised capacitor it is most probably either an electrolytic or a tantalum bead. Short, stubby blue, brown or red capacitors with a coloured dot on them are probably tantalum.

Dave
 
It is indeed tantalum. I replaced it with an electrolytic. Is that going to be an issue do you think? I don't know really know the behavioral differences in capacitor materials.. Only rated values and polarization, so I thought it would work as my replacement is 4.7uf / 35v, where the tantalum original was 4.7uf / 25v.
 
Drumroll please.... I just got back to testing and I found a cut trace I had missed causing MWRITE not to ever pulse. Well, there you go. Right now with a Z80 and Z16 in my Altair I can officially read and write to memory. Almost there!
 
One more update... It works! I put in an adding program and played with a few input values, everything worked perfectly! I'm so glad this is finally running. Time to have some fun now!
 
You must log in or register to reply here.
Back
Top