PDA

View Full Version : Altair 880 REV 0 PS Mod?



8008guy
May 1st, 2017, 04:53 AM
In the "Altair Introductory Manual for Technicians" document there is a modification to the 10 volt rail to add a pass transistor and zener to drop the rail to 9 volts. Have any of you guys added this pre-regulator to your Altair's?

len

Marty
May 1st, 2017, 05:24 AM
Hi All;

8008Guy, Yes, mine came with that modification..
Why do You ask, I think it was put in to keep the Board Regulators from getting to Hot..
BUT, when I run my Altair, I use the modern Switchers that are like what Grant used..
That way instead of pumping out all kinds of current, the Power Supply, just shuts down, if there is a problem..
Of course, I left in the old Origional power Supply, I just dis-connected it's input and its output, but it's all there in the box..
And that way I have the best of both worlds, it has it's origional supply for show, but when I am running or using it, I have a protected system..

THANK YOU Marty

KC9UDX
May 1st, 2017, 05:31 AM
We used to use fuses for that sort of thing. :)

glitch
May 1st, 2017, 05:49 AM
It's a similar circuit to that used on the B style supplies on the +/- 16V rails. I've had one Altair come through with the modification present, it worked fine and everything ran a little cooler for it. You could also use a large adjustable linear regulator and set the voltage even lower -- 7.5V is usually sufficient.

8008guy
May 1st, 2017, 09:01 AM
Hi All;

8008Guy, Yes, mine came with that modification..
Why do You ask, I think it was put in to keep the Board Regulators from getting to Hot..
BUT, when I run my Altair, I use the modern Switchers that are like what Grant used..
That way instead of pumping out all kinds of current, the Power Supply, just shuts down, if there is a problem..
Of course, I left in the old Origional power Supply, I just dis-connected it's input and its output, but it's all there in the box..
And that way I have the best of both worlds, it has it's origional supply for show, but when I am running or using it, I have a protected system..

THANK YOU Marty

No, I get the reason behind keeping the voltage lower to the on board regulators using the pre-reg outside.

I'm just restoring a REV0 Altair and was curious how many others went this way. I'm just in the process of rebuilding the P.S. I was intending to make the mod.

len

Chuck(G)
May 1st, 2017, 09:15 AM
I think the mod was pretty rare. My own 8800 doesn't have it--and I never received notification of the suggested mod.

The 8800 Rev 0 PSU is a piece of garbage in any case.

glitch
May 1st, 2017, 10:06 AM
The 8800 Rev 0 PSU is a piece of garbage in any case.

Very true. I'd probably follow Marty's lead and use modern switchers, just disconnect the original supply and leave it in place. I've also seen other hobbyists power just the first 4-slot backplane with the original supply, and use an add-in supply for any expansion backplanes. Perhaps bolt a switcher to a S-100 protoboard?

Marty
May 1st, 2017, 11:47 AM
Hi All;

Back in the Old days, I had an 8800b, I bought it as a kit, without sockets, which was a big mistake !!
Anyway, it had Fuses, but it didn't protect it well enough, one small short, I am not completely sure as to exactly where, but it took out the Whole System, every Board..
I tried to repair the Damage, and taking off the Ic's, but I ended with totally Unrepairable set of Boards, which I trashed, later I sold the Cover to someone who was a friend of my Boss..
I actually Don't remember what happened to the Power Supply and the rest of the Case, Only I don't have it any longer..
What I have now is an 8800 with two 8" Floppy Hard Sector Drives..
So, even though it may look terrible, I am much happier using modern Switchers and now all of my Boards (If possible) are Socketed..

THANK YOU Marty

8008guy
May 1st, 2017, 12:13 PM
I like sockets as well. Anything I build or repair gets them. I typically forgo originality when it comes to repairability.

There are a few sockets on the ttl of the CPU board. I assume that they were repairs.

Chuck(G)
May 1st, 2017, 01:20 PM
MITS furnished IIRC, the really cheap TI DIP sockets for their "build it yourself" boards.

Horrible things--almost right up there with the cheap white stranded wire used to connect the front panel to the backplane.

IMS did a much better job of things. Just sorry I never bought one--at one time, they offered a generous trade-in for the MITS boxes.

8008guy
May 2nd, 2017, 04:43 AM
As i tear into the harnes Im suprised at the lack of good engineering practices. This appears common in all the pictures i see. They appear to use the same color of wire for both ac and dc power. The wire size is also a bit questionable in the harnes. The ac fuse in really close proximity to dc connections. I would have done a few things differntly.

glitch
May 2nd, 2017, 05:07 AM
Many MITS designs are flat out bad. Thing is, they didn't seem to really learn from the design process throughout their production of S-100 stuff. Take a look at the Turnkey board for some bad design decisions. Out of all of the ways to do a power-on jump to ROM, they decided "intentional bus conflict" was the way to go! In early versions, there's a transistor that bashes MREAD down to stop other memory boards from responding -- in later revisions, it's a bunch of paralleled TTL gates.

The 16K Static RAM board, best of the MITS RAM boards, makes more questionable decisions, like decoding 8080 specific state ahead of the actual memory cycle so they can shut down the RAM array, presumably to take it easy on the underspecced original Altair power supplies (the SEMI4200s it uses are already pretty low power!). You can't use the board with anything but 8080 CPUs (and maybe 8085s if internal state is latched and brought to the bus). I guess that didn't matter for MITS since they never released anything other than 8080 boards. It's *very* similar in appearance to a Seals 16K SEMI4200 based RAM board, one wonders if one copied the other, or licensed the design, or what!

On the other hand, their 8800B front panel is actually a pretty good design, with separate S-100 interface board, and a state machine implemented with ROM for the actual front panel control. Makes you wonder if they had several subcontractors who didn't talk to each other working on stuff, or if they just bought hobbyist designs.

KC9UDX
May 2nd, 2017, 05:28 AM
They probably didn't think anyone would be picking apart their design in 1987 let alone 2017.

deramp5113
May 2nd, 2017, 06:29 AM
In the "Altair Introductory Manual for Technicians" document there is a modification to the 10 volt rail to add a pass transistor and zener to drop the rail to 9 volts. Have any of you guys added this pre-regulator to your Altair's?

len

Interestingly, this mod was introduced, then phased out as the early computers rapidly evolved. In the earliest day of the Altair, the 8v supply ran high with only two or three boards in a typical system drawing power. This mod lowered the 8v supply closer to the intended 8v level to reduce the heat the 5v regulators on individual boards were dissipating. Then, as systems grew and more boards were present in a typical system, the mod had to be removed to keep the 8v supply above the drop-out voltage of the 5v regulators.

In the 8800b, MITS used a transformer with three different taps for the 8v supply to allow the end user to adjust for system load by using the appropriate transformer tap. I've fixed several "random" problems on 8800b computers by moving to the next higher voltage tap. Each of these systems were on the lowest voltage tap to begin with and had a sizable set of boards on the bus, so the 8v supply was closer to 7v and regulator drop out on boards was causing glitches.

Mike

glitch
May 2nd, 2017, 06:38 AM
They probably didn't think anyone would be picking apart their design in 1987 let alone 2017.

It wasn't good for 1977, though. There are many contemporary boards with *much* better designs.

Chuck(G)
May 2nd, 2017, 06:51 AM
My gotcha on the 8800 was the live AC traces on the PCB to the power switch. Got zapped a couple of times before I mounted a switch on the rear of the unit. IMSAI took the basic Altair design and more-or-less did it right.

When I decided to use my S100 box for real work, I sprang for an Integrand box. No front panel--just a big reset button, but once you've got a decent monitor in ROM, the panel doesn't really matter. Built like a battleship.

KC9UDX
May 2nd, 2017, 07:29 AM
It wasn't good for 1977, though. There are many contemporary boards with *much* better designs.

And there's no reason they couldn't have bit the bullet early on and switched to a common regulated supply.

glitch
May 2nd, 2017, 07:31 AM
One of the parts buckets I picked up over the years had a huge double-pole breaker cut into the back panel, I guess someone else got tired of the occasional shock too :) Keep meaning to design a remote cut-in relay for S-100 boxes that put power on the front panel. Other than the zap hazard, a loose screw can end up putting 120 VAC into logic!

I use my IMSAI's front panel extensively in debugging faulty hardware, IMO that's really the place where a full front panel beats a ROM monitor. My first S-100 debugging experience was with a homebrew debug board with a ROM socket and character display; it got the job done, but the front panel is quicker. Definitely prefer a turnkey system for programming though, you don't end up second-guessing your hardware as much.

glitch
May 2nd, 2017, 07:34 AM
And there's no reason they couldn't have bit the bullet early on and switched to a common regulated supply.

I would argue that per-board regulators probably made a lot of sense in the mid 70's, switchers were still complicated and expensive compared to a linear supply, and spreading out the heat load let them get away with less thought to airflow. In any case, choice of power supply and regulation had little to do with poor digital design choices like achieving power-on jump with an intentional bus conflict!

8008guy
May 2nd, 2017, 07:37 AM
They probably didn't think anyone would be picking apart their design in 1987 let alone 2017.

Back in the 70s I learned what good was by taking apart other comercial and millatary surplus electronics. It wasnt the dark ages of electronics, there were plenty of examples of good engineering. They just were cleary inexperianced, which is fine. If you disect an apple one it becomes clear that it lived on the edge as well.

KC9UDX
May 2nd, 2017, 07:59 AM
Back in the 70s I learned what good was by taking apart other comercial and millatary surplus electronics. It wasnt the dark ages of electronics, there were plenty of examples of good engineering. They just were cleary inexperianced, which is fine. If you disect an apple one it becomes clear that it lived on the edge as well.

My point though is that they got a product to market, probably with no intention of it lasting long enough for anyone to care if it didn't.

Chuck(G)
May 2nd, 2017, 08:22 AM
I would argue that per-board regulators probably made a lot of sense in the mid 70's, switchers were still complicated and expensive compared to a linear supply, and spreading out the heat load let them get away with less thought to airflow. In any case, choice of power supply and regulation had little to do with poor digital design choices like achieving power-on jump with an intentional bus conflict!

Switchers may have been expensive (I still have a few of Bob Boschert's early hand-built ones), but open-frame heavy duty PSUs were as common as dirt. Integrand wound their own transformers--getting good-sized custom transformers wasn't nearly the hassle then that it is today. I suppose that the MITS developers weren't concerned about matters of wasted power and heat dissipation either. They seriously miscalculated on some of the boards, however. The 4K DRAM boards would easily smoke a 7805 regulator in time. MITS' solution was to issue an ECO to bridge the regulator with a power resistor.

8008guy
May 2nd, 2017, 02:03 PM
My point though is that they got a product to market, probably with no intention of it lasting long enough for anyone to care if it didn't.

Tottaly agree. In addition early adopters care less about robustness than just having it. Being an early adopter comes with a completly different attitude.

Chuck(G)
May 2nd, 2017, 02:58 PM
I'm not certain I agree. Perhaps in the days of the early microprocessor PCs that may have been true. But look at the other minicomputer stuff that was out there at exactly the same time--and tell me with a straight face that the early microprocessor systems weren't built to withstand years of use. IMS certainly thought the contrary.

Sure, there were "made to play around with" MPU systems, such as the KIM-1, but that was scarcely more than a PCB with some stuff on it.

KC9UDX
May 2nd, 2017, 06:59 PM
I would argue that per-board regulators probably made a lot of sense in the mid 70's, switchers were still complicated and expensive compared to a linear supply, and spreading out the heat load let them get away with less thought to airflow. In any case, choice of power supply and regulation had little to do with poor digital design choices like achieving power-on jump with an intentional bus conflict!

A switcher wouldn't be necessary. It would be simple enough to have the power supply in a separate box, or maybe in a segregated compartment. Plenty of other designs do this.

glitch
May 3rd, 2017, 06:06 AM
My point was that they didn't seem to *learn* from their earlier digital design mistakes, while competitors and 3rd party add-on manufacturers seemed to be able to get it right:


Many MITS designs are flat out bad. Thing is, they didn't seem to really learn from the design process throughout their production of S-100 stuff.

FWIW, they did have the power supply thing figured out by the 8800B, and I've seen 8800As with the B style supply installed. Selectable taps or a CVT were IMO the right way to go; incidentally, it's what you see in a number of later manufacturers' cabinets. But things like the Turnkey module's power-on jump circuit make you wonder if they tasked someone who didn't really know what they were doing with getting a design to go on too short of a time budget (and the point here is that the Turnkey module was well into the timeframe in which MITS should've known better -- they did, after all, create the bus spec, which included even at its inception the status disable lines necessary to do it right!).