Activity Report - July 23, 2004 by Ed Thelen
Reforming Capacitors on the 1401
More than you ever wanted to know. - and never asked in Trivial Pursuits :-((
This report got so long and complex that I offer a Table of Contents.
1992 capacitors in this machine?? Forming and testing a whole bank at one time :-)) Pictures and log book Problems along the path Thoughts
Well friends - we have all looked at web sites, talked with friends, ... about reforming capacitors.
After playing with the leaked capacitor as related in an e-mail of 7/22/04,
- Capacitor reforming - So I "borrowed" that sick looking ROE brand capacitor - it also has IBM on the label - maybe made under contract to IBM
It seemed to take current reasonably (about 20 milliamps) as I raised the voltage at say 1 volt per second. and in less than a minute - the thing was across 30 volts and drawing no indicated current. Hmmmmm what is the meaning of that? Can't be that good!
After fooling around charging and discharging through a 110 volt 7 watt night light - I decided that the thing was acting as a capacitor so decided to measure its capacitance.
The capacitance of that sick looking puppy is about 20% above its marked 10,000 microfarads !! It takes about 13 seconds to discharge from 30 volts to 10 volts across a 1,000 ohm resistor.
Not that I think that sick looking thing that has oozed in the past all over some wires should go back into that power supply,
but it was interesting - If the good looking capacitors reform that fast, we are in Fat City !! I enjoy counting my chickens early :-))
Anyway, lookin' hopeful
So, I talked with LaFarr who had reformed a capacitor as part of a small home project. We decided to see if we could get some replacement capacitors at Excess Solutions http://www.excess-solutions.com/ on Brokaw in San Jose and see what we could do with the project's power supplies laying on the work bench at Computer History Museum. (see Activity Report Jul-21
1992 capacitors in this machine??
Excess Solutions had only one 30v capacitor in that form factor and we bought it. As we checked out, we explained to the clerk what we were doing and showed him the old oozed capacitor. He said "This 'old' capacitor is newer than the capacitor you just bought!" - WHAT? "Yes, look at the date codes, this 'old' one was made in 1992, and the one you just purchased was made several years earlier."
??? How could a 1992 capacitor be in this ancient 1401??? Then I remembered that Arnold Schweinsberg has said in an e-mail that he had taken the 1401 to a show in Dortmund. Maybe Arnold had replaced some capacitors with more recent production? ???
I called Mike Falarski that LaFarr and I were going to the CHM and likely going to be in the 1401 room into the evening.
Forming and testing a whole bank at one time :-))
LaFarr and I decided to put the 'old' oozed capacitor back into the power supply and reform all 7 capacitors at once. :-))
We disconnected the 25 ohm 25 watt bleeder resistor and the wires leading to the positive side of the capacitor bank. Using the handy 110 volt 7 watt night light as a current limiter, we connected my 0-to-30 volt power supply to the now isolated capacitor bank
Figure 1. Capacitor Charge Circuit
and watched the night light filament glow dull red and the voltage rise on the capacitor bank. In a short time the capacitors were at their rated voltage, and the night light was not glowing. :-)) Lookin' Good :-))
OH - in the above circuit, I did not show volt meters monitoring both the Power Supply voltage and the Capacitor Bank voltage. And ideally, you should set the voltage of the Power Supply to the working voltage the Capacitor Bank. Part of the game is to assure that you don't over voltage the Capacitor Bank much. If you do, you will reform the capacitors to a higher than necessary working voltage and reduce their capacitance.
OK - well now - How effective is that capacitor bank? Lets look at the leakage resistance of the capacitor bank. I disconnected the 0-30 volt Power Supply and watched the voltage. The voltage decreased so slowly that I got bored and figured the leakage resistance was high enough for practical purposes. :-))
I then placed a 10 watt 250 ohm across the charged capacitors. It took 30 seconds for the voltage across the capacitors to decrease from 30 volts to 10 volts (about 1 time constant). Using the formula,
TimeConstant=Resistance times Capacitance,
I figured that the capacitance of the capacitor bank was about 0.12 farads, well above the rated 0.07 farads. :-))
I did not figure an easy way to measure the effective series resistance of the capacitor bank. A practical evaluation of acceptable series resistance is to measure the temperature rise of the capacitors when handling rated or system current load
Should be very small - say less than 5 degrees F.
The above was very good news:
a) we had an effective capacitor bank, and likely a good power supply :-)) b) we had done all of the seven capacitors at the same time, a big time saving :-))
Pictures and log book
We left a big note in the power supply and wrote in the 1401 log book.
1st Power Supply
detail of the now solid leakage
And 1401 log book page 2
We did notice that the clamp-downs for the capacitors seemed designed for *eight* capacitors, and only seven were present. The eighth position seemed to contain an informal modification. I did not identify the red thing - but it was wired as part of the capacitor bank and was so tested.
Life was good. Lets see if we can do more !!
Problems along the path
Well - life got more complicated
- The other power supplies were different, the next was a 3 volt power supply, and:
- It will have to be disassembled to see the large capacitor ratings - especially the voltage rating
- There was a 75 microfarad electrolytic capacitor that was soldered in (probably used to help power the regulator circuitry.) I didn't bring a soldering iron.
- It had two little cards plugged in that I didn't care to get involved with at this point
- We decided to wait a bit on those more complicated power supplies
- The large bank of capacitors still in the 1402 are 70 volt rating - we need to bring a voltage source capable of capable of delivering at least that voltage to properly form or test its operation
- The two power supplies in the 729s are interesting.
- ? VAC input, tap and pot adjustable - page 03-01-1
+- 12 and +- 6 for logic
- 220 VAC input, tap adjustable power supply - page 03-02-1
+ 140 volt supply for clutches
+ 62 volt power supply to drive neon indicators
- 7.5 for address light
- 48 volt supply for relays,
with a 10 ufd 250 v AC capacitor across a transformer to consider :-((
So we stopped on a happy note, but not at final victory.
- LaFarr has an old DEC PDP-11 model 20 home that hasn't been fired up ( well let's use the words "turned on") since 1982 - bring that up will give another point of reference on our quest to bring up the power supplies in the 1401 system. He is planning to do that this weekend. Good Luck LaFarr!!
- I suspect that the 729 tape units were probably not exhibited at Dortmund - and the 729 power supplies might REALLY need re-forming.
- LaFarr and I wondered about bringing up the rest of the linear power supplies by exciting the input transformers with a resistor (say a 7 or 50 watt light bulb) in series to limit the current and help properly form any degraded capacitor with out excessive heating and excitement. (Using a resistor like this can reduce/eliminate complicated voltage ramping schemes - at a loss of some time and energy efficiency - but heck - we have hours and the few cents to do the job.
We decided to check with the rest of the group before doing this.
- We have 110 to 220 volt transformers to help do this job.
- We *think* that undervoltage excitation will not cause any damage to the linear power supplies
- We want an IBM C.E. around to verify connections to the power supplies, assumptions, etc.
- A rumor about capacitors - Later, when discussing the leaking capacitors, my son Randy remembered that in the early 1990s the Chinese had exported a huge number of capacitors that leaked in service. May be ROE is a Chinese product?