• Please review our updated Terms and Rules here

California Computers S100 smurf grenades me

In my high school electronics class, the teacher had one rule regarding lab work: You blow up a cap, accidentally or otherwise, you're out of the class. Fortunately there were no such incidents.

In my first year of tech school, when we were in power supply lab, we were all busily following along with our current lab assignment when the silence was shattered by a loud POP and a large puff of smoke wafting up from two now red-faced students' breadboard. All three instructors wandered over and several second-years peeked around the corner from the TV lab next door. The can remained intact, but it had been uncorked neat as you please, with its shrink tubing pushed partly off the can (not sure how that was possible). The two students' books, parts boxes, and everything was just covered in paper flecks, and the capacitor plates had partly unwound. After the teachers had a good look, they concluded the cap was probably an old one that finally gave up the ghost.

Good to know on those blue tantalums (I don't care much for ANY through-hole tantalum caps, given how unforgiving they are). We're probably not far from a point that we need to start recapping any old electronics coming across our benches just like the vintage radio / TV folk do.
 
The blue Tants appear to be the main ones, but I have had many fail including striped ones of various colors in vintage Tek gear. I won't always replace them all, definitely, just because one shorts , but it depends on the circuit they are in.

When there is current limiting from analog regulators, and there are Tants sprinkled around the board, mostly on the regulator output lines, if they short, its no drama because the analog regulators go into self limiting mode, no pcb tracks get vaporised. So you can replace them if and when they fail. Same with many computer mobos covered in Tants powered by SMPS, that were designed for a computer, they simply go into shut down mode with direct shorts & overload. Still, if its a high current capable supply the capacitor can still get smoked. Usually the collateral thermal damage is not too bad, in many cases it much worse with a Rifa line power filter cap burning up. But a Tant can really stink out a room. Be aware though that generic multi-application SMPS's don't often have the kind of detailed overlad protection seen in many computer SMPS's. This is one of the reasons I'm not that keen on them to replace analog supplies in my vintage computers.

In the case of Tants on a backplane, with a very beefy pre-regulated power rail, it does risk the pcb if the tants short out, so in that case if one did, I would replace them all as I suggested. It is not the caps that you have to worry about, its the pcb.

I tend to look at each case individually and decide. I certainly have not replaced all the Tants in my SOL-20 or on my collection of S-100 boards. But given if all the caps are the same type and age on a board, if more than two of them short on the same board, it is a fair indicator that the others will follow. If just one does it, and the others are in a current limited circuit, mostly I'd wait to see what happens next.
 
falter said:
So let's say you have a Northstar Horizon or Poly 8813 like I do that you've never powered up.. what would be the procedure you would follow to minimize the risk of kaboom?
Click to expand...

A Northstar is a tricky case because it’s backplane isn’t completely passive; it doesn’t have as much built into it as a SOL-20 but it does have about an S-100 card’s worth of circuitry (serial and parallel ports) integrated into it, along with regulators for just those components. I’m not sure how happy that stuff would be if you slowly cranked up a variac.

For my Northstar, which I have not remotely finished restoring, the capacitors that really had me worried were the giant electrolytic ones in the power supply. I read that charging them *very slowly* sometimes worked to reform them, so I disconnected the PSU from the backplane, made some dummy loads out of series wired incandescent Xmas lights, and then rewired the supply so the caps were charged through a *very high* resistance to limit the charging current. (I think I started with something like 50Kohm?) Then I let it run for… gee, I don’t remember, two weeks or so? Occasionally I’d measure the voltage drop to get an idea of how it was proceeding, and as it improved I started lowering the resistance by adding additional resistors in parallel. (I didn’t have a big rheostat or variac for this.) Eventually, when things looked good I took out the limiters, and when there was no drama with the dummy load I hooked it up to the backplane board, took a deep breath, and… hooray, no explosions.

I got as far into the system as checking out the voltage regulation on the cards I have one by one, fell down a rabbit hole when I found they were waaaay out whack on a DRAM card… and haven’t gotten back to it since. But if it helps calm your worries at all absolutely nothing went nuclear so far.
 
tech58761 said:
In my high school electronics class, the teacher had one rule regarding lab work: You blow up a cap, accidentally or otherwise, you're out of the class. Fortunately there were no such incidents.
Click to expand...
There were no instructions like that when I took electronics in high school. In fact, one common prank was for a student to insert the leads of a electrolytic capacitor in one of the bench outlets after power to them had been shut off for the day. The next morning, when the benches were powered back up, *BANG!* Even the instructors got in on such fun; one of them kept a large electrolytic capacitor and a large well-scarred screwdriver in his desk specifically for similar pranks on the freshmen. During their introductory labs, he'd sneak off to the tool crib, charge up said cap using a vintage B+ supply (likely also kept around solely for this purpose), quietly walk behind the freshman benches, and short the cap with said screwdriver. *BANG!* Guessing neither prank would go over too well with the higher-ups nowadays...
tech58761 said:
In my first year of tech school, when we were in power supply lab, we were all busily following along with our current lab assignment when the silence was shattered by a loud POP and a large puff of smoke wafting up from two now red-faced students' breadboard. All three instructors wandered over and several second-years peeked around the corner from the TV lab next door. The can remained intact, but it had been uncorked neat as you please, with its shrink tubing pushed partly off the can (not sure how that was possible). The two students' books, parts boxes, and everything was just covered in paper flecks, and the capacitor plates had partly unwound. After the teachers had a good look, they concluded the cap was probably an old one that finally gave up the ghost.
Click to expand...
That reminds me of an incident during my senior year in the shop. Two of the students were building an AC inverter using plans they'd found online (back when the internet was a relatively new thing to the school), and with parts found around the shop. Something was definitely not right about the inverter, with it making an ominous squealing noise and outputting over 200V rather than the intended 120V; I stood as far away as I could, holding my ears. Sure enough, several minutes later, a loud *BANG* resounded as the remains of a large electrolytic capacitor went flying halfway across the shop! :eek: They eventually got the inverter working properly, but I never entirely trusted the thing after that incident.

A year or two earlier, a couple of students (possibly the same ones) were working on an IR remote tester they'd bought a kit for. It wasn't working, and the instructor was looking over the board. Just as he said "I think that capacitor is wired in backwards," a small electrolytic cap on the board proceeded to blow its top, venting smoke right in his face! :ROFLMAO: Once a replacement cap was wired (correctly) in its place, the IR detector worked properly.

Getting back to the topic at hand, I recently acquired my first S-100 box in the form of an IMSAI 8080, and all this talk about exploding tantalum caps (plus horror stories I've read elsewhere involving them in other vintage computers) has me leery of powering up the beast. I'll probably excise whatever tants I find on the backplane, and possibly the ones on the various installed S-100 cards, before I put much voltage into it. The seller I bought it from said it came from an estate where much of the gear was in working order, but neither a terminal nor disk drive(s) were found within the estate to go with it, so I have my doubts as to whether it's been powered on in a very long time.
-Adam
 
johnx993 said:
Amazing how many people don't recap machines during the restoration process.
In the vintage radio realm, all caps are replaced. Doesn't matter if it works or not.
It was explained to me by analogy:
Not replacing them is like finding a classic car in a dry barn in Kansas.
And then driving it without replacing the oil.
Click to expand...
I would say that that differs a lot depending on the manufacturer of the capacitors. In my anecdotal experience the light blue ones that Philips made in the 70's tend to be good even after all the years.

Meanwhile in particular the FRAKO and ROE capacitors, also most (I think) ones from Japan tend to need replacing.

I would say that not doing a recap of any capacitors that is in the control section of a switch mode PSU is like driving without changing the oil but not changing other capacitors is rather like yard driving a barn find car with dry rotted tires. Unfortunate if a tire bursts, but it won't cause a catastrophe.
 
I'll post how I brought up my Cromemco machine later when I am at home with a decent laptop, rather than my phone.

If you are planning to bring up multiple systems, I have a neat solution for you that is as safe as it can be...

Dave
 
The first thing is "safety first"...

A visual inspection to make sure that everything looks OK. No frayed or bare wires, no burn marks etc. Check that the voltage setting is correct for your mains voltage and that the correct fuses are fitted (as per the documentation for your mains voltage).

I would document what cards are fitted into which slots and which way around. A picture paints a thousand words of course...

Remove the S-100 cards and store away sensibly.

If you want to follow my description below look on PDF page 18 of document: http://www.s100computers.com/Hardware Manuals/Cromemco/Cromemco Z-2 - Instructions.pdf.

At this point I will use my multimeter and trace the mains cabling from the plugtop through the on/off switch, fuse and transformer - just to make sure that things look sensible. Note that energising a transformer using a multimeter can result in a brief 'nip' if you disconnect the multimeter probes, but keep your fingers attached...

I would then use a PAT tester (portable appliance tester) to check for earth continuity (between the plugtop earth connection and the case) and for electrical insulation (between live and neutral and the earth pin with the mains switch ON). I would choose a test voltage that corresponds to your mains voltage. if you don't have a PAT tester, use a multimeter. Check the resistance between the plug earth pin and the metal chassis - and ensure it is less than 1 Ohm (or as low as you can sensibly read using your multimeter). For insulation resistance, check the resistance between the LIVE and EARTH pins on the plug, and also between NEUTRAL and EARTH pins on the plug. Take this measurement with the On/Off switch set ON.

Next, check whether you have any RFI filters in the mains socket on the computer. If you have, make a decision as to whether to replace it or not...

Below I talk about disconnecting cables. Make sure every cable that you disconnect is suitably insulated. I use a small plastic bag and an elastic bag for each cable.

For powering up, I disconnected the transformer from the bridge rectifiers. Fortunately, these were easily removable connectors on the bridge rectifier. This just leaves the transformer and fan in circuit.

I connected the computer to the mains via a 'safety plug' that would trip if there is an imbalance in the live and neutral mains supply. This may protect your life one day - so buy one and use it.

As we are just energising the transformer primary - with no secondaries connected - I decided to change the mains fuse in the Cromemco from 5A to a lower value. This is optional, but would ensure the fuse would blow at a lower fault current than it would at 5A. This is a bit of a judgement, as there will be an inrush current to the transformer.

My initial power-up would be outside - this saves the wrath of the wife if something does 'go'...

If everything powers up OK to start with, I would read the AC voltages present on the secondaries of the transformer. Remember, I disconnected these from the bridge rectifiers earlier - so we are not energising anything on the DC side.

If the voltages are OK, the next thing would be to reconnect the AC wires to the bridge rectifiers but disconnect the DC wires from the bridge rectifiers. I would (personally) check that the bridge rectifier diodes were not open circuit or short circuit first though. When powering up this time, we will be measuring raw, rectified (but unsmoothed) DC on the output from the bridge rectifiers.

If this is OK, I would then disconnect one end of the smoothing capacitors. These (on the Cromemco) are screw connectors. I will choose to disconnect the positive side of the capacitors on the positive voltage rails and the negative side of the capacitors on the negative voltage rails.

The Cromemco BlitzBus is purely passive and doesn't contain any capacitors. So , all I did here (with the DC connections removed) was to check for short circuits between the power rail bolts.

On the Cromemco, there are fuses in the DC side of the power supply. These are 30A (on the +8V rail) and 15A (on the +18V and -18V rails). These are WAY too large for powering up an unloaded S-100 bus. Change them for something like 1A...

Powering up should give sensible (but un smoothed) DC voltages on the S-100 backplane bolts.

Depending upon the visual state of the smoothing capacitors, and whether you are happy to attempt to use the existing capacitors or not, will dictate the next steps. I elected to reform the four (4) smoothing capacitors. When I used a VARIAC and suitable resistors, this resulted in all four (4) capacitors being saved.

The only problem I came across was one of the bridge rectifiers (on the +8V rail) decided to die on me after I reassembled the entire unit. It took me a while to work this out, but a trip to the local electronics shop and I was away.

For testing cards, I built myself a single S-100 connector that I can wire up to my bench power supply unit. I can raise the voltages on the S-100 connector whilst monitoring the voltage outputs from the on-board voltage regulators. Checking for 'stress' as I increase the voltage. I have a bench power supply that I can limit the current on. Set the current limit at the maximum specified in the manual for the card. Make sure the output voltages from the on-board regulators are increasing (this would indicate no short circuit is present) and make absolutely sure that the voltage limits at that rated for the voltage regulator (e.g. +5V for a +5V regulator). This would prevent damage to the card should a voltage regulator be short circuit (from input to output). An unlikely scenario (for a linear voltage regulator) but I had one on one of the disk drive voltage regulator boards feeding a floppy unit. This would have blown the floppy disk unit electronics to bits!

Soak the cards on the test rig one at a time at the rated DC voltage.

Put the cards back into the chassis one at a time and give it a test. Ideally, you would want to build the system up in a logical manner.

OK, your system may be different (e.g. electronics on the S-100 backplane etc.). In this case, you just have to look at the risk and the best way of mitigating against the most likely faults.

Sometimes, it is just remove all the cards, take the unit outside, power it up and cross your fingers!

Dave
 
Last edited:
Thanks Dave!

Also - I note it seems like a variac is not always the answer.. when is it absolutely NOT the answer? Is there a certain situation that for sure you don't want to involve a variac?
 
johnx993 said:
Amazing how many people don't recap machines during the restoration process.
Click to expand...

I'm not bulk recapping something where capacitors are testing fine. Tantalums be tantalums, I ain't spending a week speccing, ordering and replacing every single one unless every time I power up another one is going bad somewhere. Same goes for lytics. If they tested fine for ESR, capcitance and physical leakage they are probably fine.
(Actually I did to this for my PR-7820 laserdisc player...but that was almost two years ago and the job sucked so much I've still not actually recapped it)


You go into the larger machines and linear supplies and the cost to recap becomes exponential. One PSU recently cost me $270 for six capacitors. That's not something I can write off as "eh, it had to be done". they had to be extensively tested before I determined it had to be done. Knock on wood but I have not actually recapped my Datamaster. The PSU was extensively checked before I said "it's fine" and it's held up to that word for the last year and a half. the idea of recapping every machine I have in the collection would both drive me insane with"take it apart, find all the caps, order them, wait a week, install them and test it" would drive me insane and the costs would bankrupt me.
 
>>> when is it absolutely NOT the answer?

Usually with switch mode power supplies. Linear supplies are generally OK.

My DEC PDP-11/45 has multiple supplies, all with big and expensive electrolytic capacitors. I reformed every one fine, except for two capacitors. These were rusty and physically damaged, so I have replaced these alone. I did price up replacement capacitors. No thanks...

Dave
 
FWIW, here’s my old thread on the Northstar.

Eudimorphodon

Thread 'My "new" Northstar Horizon: where to get started?'

It's been kind of a running joke for five years that I owned an S-100 computer that I'd never actually seen in person before. (A story primarily comprised of a long-distance adoption followed by several years of "how in the world am I going to safely ship this thing?") But now, thanks to some of the awkward circumstances our current world has imposed on traveling my Loch Ness Monster of a machine is now in my hands. And, man, I'm simultaneously excited and totally at a loss.

I haven't had any chance to clean it up so I'll save pictures of the outside, but a quick inventory of the inside...

I guess I spent something like a week and a half reforming the big capacitors in the PSU, gradually stepping down the resistance from 42Kohm, until testing the output for ripple with the Christmas light dummy load connected. All indications were it worked perfectly to reform those big caps, and all it cost was a few cents worth of resistors.

I also didn’t have any tantalum caps go nuclear in subsequent testing (despite the machine being riddled with them), so while they may be spectacular when they go I don’t know how widespread the problem really is. I guess I did leave off, among other things, worried the floppy drives might be fried, the 12v regulator powering just them was way out of spec, but maybe they’re okay if it went bad while it was out of action…
 
I got 1,500 cards made for work (analogue input multiplexer cards) to replace some old cards - and hence extend the life of the equipment.

The highest failure rate was the very expensive 68uF solid tantalum capacitor on each analogue input (16 inputs per card). The next highest failure rate was the tantalum bead capacitors on the power supply rail (+24V rail feeding a +15V linear regulator). The regulator input and output each had a capacitor. The shorted capacitor on the input would operate the power supply protection.

All of the above faults were on our burn-in rig.

We checked the failure rate of the NEW tantalum capacitors with the manufacturer's datasheets, and we were still within their projected infant mortality rates.

So, even buying new parts, and replacing them, doesn't necessarily eliminate the issue!

I can provide detailed figures if anyone is interested...

We didn't have any power supply tantalums going super nova though - probably due to the power supply protection.

Dave
 
falter said:
Yes sorry, I should have been clearer. I did not expect random S100 cards to operate well together. I just was concerned about changes to things like power pins. I don't want to plug in a card and have it go 3 Mile Island. I will of course read extensively before even thinking about plugging one in, but as I am not an EE I don't always have 100% confidence that I have read things right.
Click to expand...
Well, a true story about all boards not playing nice... I had an 8K Ithaca Audio RAM board that I asked a friend of mine to check out. He called to tell me when he plugged it in and turned his machine one, very shortly afterwards, he could see heat rising off the board. He shut it down and found that pins 13 & 14 that the Ithaca Audio board used for 'standby power' to power the memory, just happened to be, at the time, listed as 'unused' and the very ones he used for the low voltage AC he used for the power supply for the programming voltage for the EPROM programing board he had in his machine. So all the the memory chips on my board got hit with about 12 volts AC across the power pins. And lo and behold, not a single chip failed. The board worked fine for many years.
 
daver2 said:
Oh yes, beware the 'unused' or 'reserved' pins - they always get used...

Dave
Click to expand...

That is definitely one of the pitfalls that applies a little more to S-100 than some other busses. There were, what, like two dozen or so? pins in the original Altair layout that were open season until IEEE-696.
 
NeXT said:
I'm not bulk recapping something where capacitors are testing fine. Tantalums be tantalums, I ain't spending a week speccing, ordering and replacing every single one unless every time I power up another one is going bad somewhere. Same goes for lytics. If they tested fine for ESR, capcitance and physical leakage they are probably fine.
(Actually I did to this for my PR-7820 laserdisc player...but that was almost two years ago and the job sucked so much I've still not actually recapped it)


You go into the larger machines and linear supplies and the cost to recap becomes exponential. One PSU recently cost me $270 for six capacitors. That's not something I can write off as "eh, it had to be done". they had to be extensively tested before I determined it had to be done. Knock on wood but I have not actually recapped my Datamaster. The PSU was extensively checked before I said "it's fine" and it's held up to that word for the last year and a half. the idea of recapping every machine I have in the collection would both drive me insane with"take it apart, find all the caps, order them, wait a week, install them and test it" would drive me insane and the costs would bankrupt me.
Click to expand...
So if you were in my shoes and did what I did and had one tantalum from one bouquet go bang, you'd just replace that one and carry on? I just worry about these things going off in my face while I've got the hood open.
 
If you absolutely don't want to replace all the tantalum capacitors, maybe consider putting something on top of them that would protect plugged in boards and most of the motherboard from debris from failing capacitors, perhaps? Don't know what would be suitable, like a screw on cap from a typical carbonated drink but non combustible yet insulating.

But also: The plug in boards should also have capacitors. Are all capacitors on the motherboards even necessary?

On the third hand, it looks like there are 17 tantalum capacitors including the failed one. I assume that some are for the unregulated +8V rail and some are for the unregulated +16V and -16V rails. Thus there are likely two different specs. I.E. two different 10-packs of capacitors would likely be a decent replacement. I would think that modern low ESR electrolytic capacitors would work fine, and if you really want to go full overkill you could solder "nanofarad" capacitors in parallel, on the solder side of the PCB.

I wouldn't bother doing this if it was say a Pentium 4 socket 775 PC motherboard from 2005 as there are plenty of those. But for something as old and valuable as an S-100 computer in such great shape it seems like a good idea to replace those capacitors.

============

Btw, in case anyone wonders, the tester Daver2 refers to for checking insulation between ground and live/neutral are commonly called a megger in many places of the world, although that is afaik technically a brand and not the correct name.
 
On a bare backplane you should just be able to use a meter to see if you have anything that looks like a dead short between voltage and ground. You probably grenaded the only one already…
 
MiaM said:
If you absolutely don't want to replace all the tantalum capacitors, maybe consider putting something on top of them that would protect plugged in boards and most of the motherboard from debris from failing capacitors, perhaps? Don't know what would be suitable, like a screw on cap from a typical carbonated drink but non combustible yet insulating.

But also: The plug in boards should also have capacitors. Are all capacitors on the motherboards even necessary?

On the third hand, it looks like there are 17 tantalum capacitors including the failed one. I assume that some are for the unregulated +8V rail and some are for the unregulated +16V and -16V rails. Thus there are likely two different specs. I.E. two different 10-packs of capacitors would likely be a decent replacement. I would think that modern low ESR electrolytic capacitors would work fine, and if you really want to go full overkill you could solder "nanofarad" capacitors in parallel, on the solder side of the PCB.

I wouldn't bother doing this if it was say a Pentium 4 socket 775 PC motherboard from 2005 as there are plenty of those. But for something as old and valuable as an S-100 computer in such great shape it seems like a good idea to replace those capacitors.

============

Btw, in case anyone wonders, the tester Daver2 refers to for checking insulation between ground and live/neutral are commonly called a megger in many places of the world, although that is afaik technically a brand and not the correct name.
Click to expand...
Back in my Navy tech days we used the 'megger', which was an insulation breakdown checker to check the breakdown resistance of aircraft antennas. If memory serves, anything less than 300,000 ohms was considered short. The thing operated like hurdy-gurdy and required 2 people as one had to crank. Primitive but effective. Railroads used to use those things too.
 
Several pins used by the Altair computer were redefined with the IEEE 696 spec. Plus as the S100 bus was in it's early days many other computer makers used some pins for special purposes that weren't allowed later on with the IEEE 696 spec. A couple of pins in the spec's were also left undefined so makers could do more or less what they pleased with these pins. There are a few somewhat common uses for one or two of these undefined pins. But no, the Altair and the IEEE-696 are not the same even if most of the signals are shared.
 
You must log in or register to reply here.
Back
Top