The IMSAI restoration Part I - Power Supply: let there be light (literally)

507 views
Skip to first unread message

P- Lab

unread,
Nov 6, 2020, 1:15:24 PM11/6/20
to IMSAI 8080esp
Hi all,
as promised I'll try to write here the progress of the restoration of the real IMSAI of my friend Andrea, the owner. I am just helping him remotely.
If you think that this is off-topic just let me know and I'll stop...

Pictures comes from WhatsApp, so the quality is low.
Better photos and descriptions in a better English will be released at the end of the work (hopefully...)
Make yourself comfortable: long post.

The IMSAI appears in a really good shape (excellent, I'd say), with no dust/rust at all inside. All is extremely clean, which is quite a surprise for a 44yo machine.
If I remember well it is equipped with:
- MPU-A rev.4 with purple ceramic 8080
- N* Disk controller MDC-A2
- 1x Seals El. RAM board (8 kB)
- 4x Godbout EconoRAM II (8 kB each)
- MIO CArd
- Video terminal card

I will write about the individual components as we get there.

We decided to adopt a Socratic approach (I know that I know anything) and therefore to use schematics and manuals, at least for now.
Power supply is very simple, our only concerns were about the electrolytic capacitors.

They were all removed from PCB and their internal dielectric individually "reformed" via a very slow process. The process is based on increasing the voltage applied to the capacitor very slowly, constantly measuring the leakage current and giving the dielectric the time (many hours) to reform.
It took one day for each 95000uF, and half a day for the others.

The biggest priority, however, was to remember to gently discharge the capacitors at the end of the work. Can you imagine what would happen to a metal object dropped by mistake on the terminals of a fully loaded 95000uF capacitor? :-O

At the end of the process and the nominal voltage, the leakage current has been found negligible. No bulges nor other strange things were found.

We were not able to determine clearly if the transformer is the 7V/10A or the 7V/28A model.
In doubt, we decided not to drain more than 10A from the +7V branch.
Based on the diameter of the output cables and my experience, I would say that it is the 10A model, but I could be wrong.

In order to properly test +7V branch along with +13.5V (4.5A) / -13.5V (4.5A) we did some math on the max declared power of the different sections.
To simulate the load we decided to use 8x halogen car headlights 12V/55W each.
The connection were like:
- A SERIES of two bulbs between +13.5V and common GND
- A SERIES of two bulbs between -13.5V and common GND
- A PARALLEL from one to four bulbs between +7V and common GND to simulate different loads. Moreover, tungsten filaments acts as PTC resistors so we preferred not to stress +7V too much for the first switch on.

Andrea built a plywood frame to keep all the bulbs in place and to avoid short circuits, connecting the bulbs with properly sized wires.
He also placed a large fan in front it to avoid burns caused by the heat of the lamps (in the pics they appear brigther than they are).

The test started with a single bulb on the +7V branch, progressively adding them up to four and monitoring current, voltage and ripple.

The total load was about:
- 36% of the maximum load for the +7V branch (if transformer is 28A model), 100% if 10A.
- 66% of the maximum load for branches +13.5V/-13.5V
and has been maintained for two hours, with adequate cooling to avoid overheating damages to the insulation of the windings.
Better safe than sorry.

Finally, Andrea reconnected the power supply to the boards and...  no smoke! Success!
The display also ligthed up! Great success!!
A very quick test showed that the switches and buttons were working perfectly.

In the next post we will see some functional tests, and this is where the IMSAI 8080esp came to great help.

See you in the next post!
Claudio.

IMG-20201022-WA0046.jpg
IMG-20201021-WA0019.jpg
IMG-20201021-WA0020.jpg
IMG-20201021-WA0022.jpg
IMG-20201022-WA0046.jpg
IMG-20201021-WA0018.jpg
IMG-20201022-WA0048.jpg
IMG-20201025-WA0020.jpg
IMG-20201022-WA0047.jpg
IMG-20201022-WA0012.jpg
IMG-20201022-WA0044.jpg
IMG-20201022-WA0037.jpg

P- Lab

unread,
Nov 6, 2020, 2:09:02 PM11/6/20
to IMSAI 8080esp
* I know that I know nothing

Ted Kekatos

unread,
Nov 6, 2020, 8:45:29 PM11/6/20
to IMSAI 8080esp
Thanks for posting your IMSAI 8080 restore story.   

The Oracle

unread,
Dec 27, 2020, 11:48:50 PM12/27/20
to IMSAI 8080esp
Claudio,

Please don't use reformed cap's, you can get replacements from here https://www.tedss.com/

And I don't know if that's a photo of the fan they are using, but here is a photo of the optional fan kit from IMSAI.

IMG_9049.jpg

Thanks,
Lynn

P- Lab

unread,
Dec 29, 2020, 7:00:31 AM12/29/20
to IMSAI 8080esp
Hi Lynn,
Andrea reported that original caps were healthy after few days of reforming, no need to replace them.

By the way I am currently restoring another IMSAI 8080 (not mine) which definitely needs new caps.
Thank you for the link, I'll forward it to the owner.
The original fan is ok, smooth and quiet as it should be, but thank you anyway. :-)

Regards,
C.

The Oracle

unread,
Dec 29, 2020, 10:22:45 PM12/29/20
to IMSAI 8080esp
Claudio,

If the cap's were bad reforming them is only a temporary fix (they are a ticking time bomb). The IMSAI uses an analog power supply the capacitors are in a form regulating the voltage coming from the power supply. If they go the voltage going to the mother board will go way up, the voltage regulators on your S-100 cards will not be able to handle it and will hopefully burn out before any spick gets to the rest of the card.

Not trying to make a big deal, but there are only a few original IMSAI's left in the world and we have to be the best stewards we can be.


Thanks,
Lynn

curt mayer

unread,
Dec 30, 2020, 12:06:54 AM12/30/20
to The Oracle, IMSAI 8080esp
I'm confused.  Caps will fail open or fail closed.  If they fail closed, you will have a short across ground and power, with high current inside the cap and it will blow up as the electrolyte boils and vaporizes. the high current upstream may also cause the rectifiers to blow.  Again, all ending in open circuits.  I don't see a failure mode where you get voltage spikes.  Am i missing something?

--
You received this message because you are subscribed to the Google Groups "IMSAI 8080esp" group.
To unsubscribe from this group and stop receiving emails from it, send an email to imsai8080esp...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/imsai8080esp/d86c1485-bc90-4c4d-b009-b7c0cc9de6dfn%40googlegroups.com.

P- Lab

unread,
Dec 30, 2020, 4:26:10 PM12/30/20
to IMSAI 8080esp
I am even more confused.
There are two IMSAIs getting restored around here in these days (and none of them is mine, unfortunately... --.-- ).

I helped restoring the first (Mr. Andrea's) from a distance --> reform was fine --> no need for new caps --> yay!
Due to the successful restore, another local friend gave me his bad-in-shape IMSAI to try a restore, so I am taking a good care of it --> tried reforming --> not ok --> big caps need replacement --> will do, thanks for the source.

In the meantime I am using other external power supplies to continue with the restore, which is going quite well, by the way.
BTW, I agree with who wrote that there cannot be overvoltages involved in a caps failure.
Regards,
C.

curt mayer

unread,
Dec 30, 2020, 6:19:12 PM12/30/20
to P- Lab, IMSAI 8080esp
Transformer failure could you overvoltage, but not capacitor failure in simple linear power supplies.

--curt

The Oracle

unread,
Dec 30, 2020, 6:25:16 PM12/30/20
to IMSAI 8080esp
I'm looking at a catastrophic capacitor failure, as you refenced in your first line. during that failure (we are talking microseconds) the power draw will increase substantially and will continue to output power to the mother board until complete failure. 

lets look at that a minute. note the placement of C0, C1, C2, C3 in the IMSAI power supply there is nothing to stop the voltage from going straight into every thing pugged into the mother board.

PSC_2.jpg


Thanks,
Lynn

The Oracle

unread,
Dec 30, 2020, 6:26:25 PM12/30/20
to IMSAI 8080esp
We are NOT dealing with a linear power supply in the IMSAI.

Thanks,
Lynn

The Oracle

unread,
Dec 30, 2020, 6:30:25 PM12/30/20
to IMSAI 8080esp
Claudio, 

That's good, you should keep moving forward in what ever way you think is best for your customers. I was trying to help.


Thanks,
Lynn

curt mayer

unread,
Dec 30, 2020, 11:03:19 PM12/30/20
to The Oracle, IMSAI 8080esp
Capacitors are not regulators.  They are layers.of conductors separated by insulation.  If the insulation fails, then they fail closed, conducting current across a low resistance, giving off heat,.smoke, poof.  Something usually blows up upstream too, if you are lucky a fuse, but it might be a rectifier diode.  When these blow, there is an open circuit. No voltage.

Without a charge pump somewhere, you cant get large voltagea on the bus from this power supply.

curt mayer

unread,
Dec 30, 2020, 11:12:49 PM12/30/20
to The Oracle, IMSAI 8080esp
It is exactly a linear power supply.  Regulation shunts straight to ground.  It's grossly inefficient and almost foolproof.  Cap failure gets you excess ripple if open, dramatic electrolyte boiling if closed.  Where exactly does this high voltage come from? The secondaries are behind rectifier diodes.  Insulation failure in the secondaries could get you high voltages, but that's a remote probability.

Richard Deane

unread,
Dec 31, 2020, 3:27:10 AM12/31/20
to curt mayer, The Oracle, IMSAI 8080esp
If the capacitors fail such that smoothing doesn't occur then surely you get overvoltage to the peak of the unsmoothed chopped sine wave from the bridge rectifier. I would expect this to be higher than the desired voltage of the smoothed signwave, and I would have expected the board electronics to be designed for the smoothed voltage.
Richard

udo....@freenet.de

unread,
Dec 31, 2020, 4:07:09 AM12/31/20
to IMSAI 8080esp
No matter what, the secondary of the transformer is 8V max etc. and the regulators on the
boards can deal with that of course.

Mark Cummings

unread,
Feb 27, 2021, 9:27:18 AM2/27/21
to IMSAI 8080esp

I just want to clarify a few misunderstandings and terms in the power supply design. I am not referring to an incorrectly setup supply.

The IMSAI 8080 PS-28U power supply is an unregulated power supply. THE PS-28U User's guide states this on page 2-29. The unregulated voltage depends on the AC mains input voltage, the primary taps selected on the transformer and the load on the secondary. According the the User's guide the unloaded output voltages should not exceed +11V on the +8V output, +18V on the +16V output or -18V on the -16V output. 

The regulation of the voltages down to +5V, -12V and +12V or whatever is required on the card is done on the individual interface cards themselves, not on the PS-28U. The regulators on the interface cards remove the AC ripple from the unregulated supply from the PS-28U.

The large secondary capacitors are also required to sustain the output voltage at least 2V higher than the +5V or +12V voltage required on each interface (or 2V lower than -12V for the negative rail), and this to ensure the regulators do not go into dropout (meaning that they can no longer regulate the voltages on the interface cards).

The normal mode of failure for electrolytic capacitors is to go high impedance (or resistive) which would affect their ability to deliver current during the dips in the rectified AC secondary output. This would show up as a larger dip in the voltage which could cause regulator dropout. The peak voltage of the secondary is the maximum that the capacitors can charge up to, not exceed. As the peak of the AC waveform decreases the capacitors are their to maintain the average voltage output and supply current to the load and in doing so this will cause a certain amount of ripple. Under normal circumstances the capacitance is selected to hold the voltage up above dropout during the dips in the AC waveform on the secondary. The higher the load the higher the ripple, however the peak of the ripple will not exceed the output peak of the secondary on an unloaded supply.

In short, a failed secondary capacitor (in it's normal failure mode) is extremely unlikely to cause an excess voltage on the interface cards and therefore should not damage them electrically. If anything the voltage would dropout on the cards and cause incorrect operation and other symptoms.

Mark.
Reply all
Reply to author
Forward
0 new messages