Most recent revision dated Mar 23 2011. This is PRELIMINARY. copyright (c) 2011 Herbert Johnson all rights reserved.
This page discusses S-100 power supplies, particularly the oldest designs which consider of a transformer, diodes and capacitors. Early S-100 cards used on-board voltage regulators fed by a relatively unregulated AC powered supply of no less than +8V, +18V and -18V. Later S-100 and IEEE-696 systems used various forms of regulated AC supplies. If the supply provided regulated 5V, 12V and -12V on the bus, the individual board regulators could be shorted across or replaced with jumpers. S-100 boards with bypassed regulators CANNOT be used with unregulated bus supplies!.
Some of this activitiy involves dangerous voltages. Please note my comments and disclaimers below. . - Herb Johnson
My S-100 home page
S-100 bus termination
S-100 and IEEE-696 bus descriptions
Here's a 2011 discussion I had with S-100 owner Nick Papadonis who is restoring a Northstar Horizon. - Herb
While I'm waiting for parts, I decided to get back to the N* Horizon. I replaced the tantalum caps that I believed exploded. I measured voltage levels coming out of the power supply section and off a bus extender card. I'm seeing +10V and +20V! I'm assuming these voltage levels are not safe for components. My voltmeter isn't broken because I measured the IMSAI and it had +8V and +16V.
[Another person] tested his N* Horizon and it had: +8 = +10.3; +16 = +21.7; -16 = -21.8. So either our systems have both deviated with age, or that is what N* intended.
Your meter is not broken and the voltages are likely what they are, based on circumstances. The S-100 common standard and also the IEEE-696 standard, says the unregulated bus power voltages are +8V, -18V, +18V. (IEEE-696 also permitted regulated +5V, -12V and +12V.)
Consider the power supply designs on the Horizon. It's an unregulated capacitor-based full wave or half-wave design most likely. It was designed to run a bunch of high power TTL S-100 cards, just like the IMSAI, Altair, and other early S-100 computers. Lightly loaded, the voltages will rise to the peak AC value of the transformer secondary. Heavily loaded, the voltage will DROP to the "average" value of the rectified AC voltage minus the .5V or .6V drop across each power diode.
Some machines like the Compupro used self-regulating AC transformers and so have more stable DC voltages over a range of loads.
To lower the voltage, you have some choices. One choice is to us a Variac or isolation transformer (of the proper POWER rating) to reduce the primary voltage. Or, look at the power supply transformer to see if there is a HIGHER voltage tap on the primary, and use that to REDUCE the secondary voltage (by lowering the turns ratio). Or modify the DC capacitor circuit with a power choke (AC inductor) right after the diodes or between the major capacitors. Or, add more boards and load to the supply (not my favorite choice).
Some people make up an AC outlet box with a SERIES wired incandescent bulb, to drop the AC voltage. (The higher the wattage, the lower the resistance of the bulb, so don't use a tiny 15-watt bulb.) The bulb will get hot of course so consider that. ONe other virtue of this "box" is that the cold resistance of the bulb is higher than the "hot" resistance. So, it acts as a kind of "surge" protector.
(please note my comments and disclaimers below.)
In any event, I think it's a good idea to get the voltage levels down. It will reduce dissipation on the board regulators which have to "drop" more voltage and so produce more heat.
I think this is a problem in the power supply section because that is where the higher voltage originates. Have you seen anything like this? [He did not have clear schematics for the power supply - Herb].
It's not that hard to trace the power supply circuit out. There ain't much there. But simple inspection will likely tell you there's no choke, just big caps and probably a full-wave rectifier. Look for any taps on the primary. ALso, you can use an oscilloscope to measure the secondary voltages. But lightly loaded, that capacitor regulator design has a DC voltage close to the PEAK AC value off the diodes (less the diode voltage drop). A scope will show you the peaks. (An AC meter won't help, it's not "proper" AC with diodes and caps in circuit.)
[Some S-100 owners told me] the values of 10V and 20V are not too far off. They indicated that load (board in the chassis) should cause a voltage drop. Additionally, that the boards should be able to handle up to 30V as the regulators are usually spec'd that high. All and all, I don't think there is an issue here... yet.
Interesting "they" say that. I think I already said the output voltage depends on load.
While I agree than a lightly loaded S-100 power supply like the Horizon's will go above 10V and 20V, I don't think the extra voltage is a good idea. Do the math on power dissipation - the 5V regulators, which handle most of the power, will have to "drop" the difference in voltage between the bus and 5V. Multiply that by the current they pass (an amp or more). Volts times amps equals watts.
And they say 30 Volts is OK on the -18? Let's see...30 minus 12 is 18 volts....if you draw 100mA, that's 1.8 watts....if that -12 regulator is a tiny transistor package, it will cook, dude! I"ve seen that happen.
Do you find that some of the older regulators stop regulating correctly? I.e. the voltage level deviates from target more? Or do they just simply fail open? I'm wondering if I should replace some of the regulators on my CPU boards. I supposed I should measure the output voltage of it first and test it while plugged into an extender board.
Here's the questions I would ask. Are those regulators heat sinked or not? How hot do they get? How OLD are they? All that goes to failure or degrading over time. I"ve found a number of S-100 cards that work better when the OLD regulators are replaced, such as Z80 CPU boards. Most of the time, regulators fail open, so no harm (usually done). If they SHORT....you'll have a bad day about that board.
I've not done a study on 25, 30, 35 year old voltage regulators. (Your Horizon probably dates from 1978 or so, look for date codes on the components.) But consider how OLD that is, for a component that runs hot to begin with. I"ve measured some 5V regulators and seen "creep" of output to fractions above 5V. Some of course are failed and produce no output. If I have a marginal board, it's not hard to replace the 5V regulators to see if it works "better".
I run some of my S-100 systems through a Variac for just this reason. And I monitor all the regulators for excessive heat. And, I check the regulators for the correct voltage.
- Herb Johnson
Extender cards are boards that run all the bus lines to the top of the card, with a bus connector on it. Boards under test can be "extended" from the bus for access. But extender cards will create problems. Think of them as inductors added to the bus in series, and capacitors added between the lines in parallel. Look at the extender card, and think about the physics of it, that's what it "theoretically" looks like. -- Herb
A "Variac" is a trade name for a variable autoformer or autotransformer. It is NOT isolated from the power line. By all means, use every caution when using these. That reminds me:
DISCLAIMER: I AM NOT RESPONSIBLE FOR ANY LOSS, INJURY OR DAMAGE TO YOU OR YOUR PROPERTY. USE THIS INFORMATION ENTIRELY AND COMPLETELY AT YOUR OWN RISK. I MAKE NO GUARANTEES WHATSOEVER THAT ANY INFORMATION I PROVIDE IS COMPLETE, ACCURATE OR CORRECT.
Copyright © 2011 Herb Johnson