Repair help with a very vintage calculator

[Nekopan]

I picked up this calculator with the hopes to restore it as I do with vintage computers. Seems moving from discrete logic chips into discrete transistor logic and a delay line was not so simple...

For reference, this is a Friden EC-132 calculator. It houses all the logic on four large circuit board modules, each with two single sided boards.


The calculator, as initially tested, wouldn't show any signs of life. After a deep cleaning and secondary inspection, it seems to be close to functional. There's a power neon light, that goes on. Usual transformer buzzing. The CRT displays the three rows of 0s for a fraction of a second, and then it seems to shrink and disappear. It does not show up again until the power is cycled. It also seems to not show up at all if I power it up "cold".

I managed to find a schematic of the power supply by emailing a few people who had worked on these back in the 60's (!), but didn't get any repair manuals. I'm all but certain those are lost. For the sake of my limited explanation, I've attached a PDF of the power supply schematic.
So, I went through and checked all the components on the power supply that I could. The capacitance test on my meter wasn't liking the caps on this board, so I just checked resistance to inspect for leakage. Also checked resistors against their colour bands, and the diodes. It all appeared to check out, from what I could tell.

I tried a couple tests. Nothing serious, I'm a bit squeamish working so close to ~3kv. I initially figured it had something to do with the neon bulbs in the circuit. That idea was shot down by someone much more technical than I, although I didn't formally test them. There was a resistor with some burn marks around the legs, so I checked a nearby capacitor out of circuit and tossed in a replacement. Nada. I checked a transformer for any erroneous continuity that was also nearby (both physically and in the circuit path). Checked out.

One interesting observation: The system lights up an "over flow lock" key when 9 digits are entered. This feature still works, even without the CRT displaying. I'm fairly certain that there's no control circuitry in the keyboard itself, so I'm hoping this is an indication that the logic is working. The problem with this reasoning is that, *when functional*, when an operation key is hit, it stays depressed until the operation is complete. Entering the "benchmark" operation (99999999 / 1) which is supposed to take the system 2 seconds to complete doesn't seem to hold the operation key down.

Many have mentioned that I should be replacing all the caps on the power supply regardless of whether they test out fine or not. There's quite a few and as they're very old caps, I'm worried about finding suitable replacements. For the time being that hasn't been done. A friend in the repair business suggested I try and turn the system on, and power cycle the CRT to see if the display shows up each time, or only at the initial power on (so as to narrow it down to a power supply or logic issue). Unfortunately, there's only a single switch for the whole system. I'd have to physically remove and insert the CRT neck board with the power on, which doesn't seem remotely safe. I had also considered probing the low voltage areas on the board to check the voltages against the values as per the schematic. Problem with that is the board is at the bottom of a large metal chassis, with the CRT nearly covering it. I could remove the CRT, but I'm unsure how that'll affect the voltages on the board. Lastly, there's a oscillation circuit in the power supply, my best guess is to provide a clock signal for the logic. But, without an oscilloscope, I'm out of luck for probing that.

This machine is big, bold and stunningly beautiful. It is over-engineered, and was obsolete just years after its launch, making it rather difficult to come across. I refuse to let this go to scrap, and so for now it sits in my closet, mocking me. If anyone has tips, I'm all ears. I'd even consider giving it to someone with the technical know-how to fix 'er up, if anyone is interested in a challenge.

[Rerouter]

As the display is fading out a few seconds after a warm power on, i have to suspect its the acceration voltage, measure pin 2 of T2, to make sure its still oscillating, preferably as its low voltage you could solder in a test lead before powering it on, to keep yourself from probing around 3KV.

This will help you cut the problem in 2,

The logic side of things still behaving (overflow) rules out some things,

On the connections for -12V, +6V and -80V, have you measured them after it fades to make sure its not one of them? The capacitor values imply electrolytics, its more common for the lower votlage ones to drop in value, but measuring the supplies would be a easy enough way to rule them out.

If you dont see a frequency on pin 2 of T2, see if there is one on the collector of Q6, its possible the oscillators capacitors may have gone marginal, you can get the flasher circuit oscialltors to lock up if both transistors switch half on at the same time.

[Kleinstein]

Elektrolytic caps can be a problem when old. However the really old one were relatively well made. Some tube based radios still work with there original caps. The good thing about replacing them with newer ones is that newer ones have a smaller form factor so one has room to use higher voltage types if in doubt.  Also electrolytic usually have a large tolerance - so there usually is no need to get the exact values (e.g. no problem using something like 470 µF instead of an old 500 µF or 68 µF instead of 60 or 80 µF).  With the caps it is more about keeping as much as possible of the old original - so at least keep the old caps.
Non electrolytic caps are usually still OK, though some also show degradation - usually more from over-voltage spikes than pure age.

The other part to look for in such old units are selenium based rectifiers. They degrade with time and chances are they would not work and could blow up with quite a mess / smell. So they should be replaced for safety reasons. It is possible to find them inside, but there where alternatives.

The oscillator circuit shown in the supply circuit is there to create the high voltage. It might also be used for horizontal deflection if the tube is using magnetic deflection. At least CRT based TVs usually did it that way. The unit is slow, but a 2 kHz clock would be really slow.

The picture getting smaller after some time somewhat suggests something like a power supply to turn down (e.g. a rectifier / cap failing when hot) or an oscillator stopping. Germanium based transistors are relatively sensitive to heat and turn on, when hot. They also show aging that can make them more sensitive.

Without a scope debugging such a unit is difficult. So it might be worth getting support from someone with more experience and a scope.

Before replacing caps or similar, please make some photos.

[Nekopan]

I'll get right on it when I've got a day off. I wasn't expecting such a quick response!

Problem is, when you ask to measure pin 2 of T2 for oscillation, what can I use for this? I have no scope with which to properly check it. My meter has frequency, but I think it's intended for mains frequency checking (defaults at 60hz). I'll look around, see if I can't borrow a scope from my friend. that'll be a few weeks down the road, with my work schedule being as tight as it is.

Before replacing caps or similar, please make some photos.

Photos of what? I can provide shots of the power supply, but I would figure there isn't much to see.

[Rerouter]

if your meter has a "Hz" mode, then that is all you need, I assume 60Hz when near an outlet, and 0 when you touch the tips together?
 
If not, you can cheat a little and use AC volts, it wont tell you the frequency, but it can atleast tell you that there is a frequency,

[Nekopan]

Yup! I'll have a go at it and see what it says.

[IanMacdonald]

As the CRT has a -3kV cathode potential this might well be electrostatic deflection. Magnetic deflection more typically uses a positive anode voltage. (The reason for the cathode having the HV is so that the deflection plates can be at or near chassis potential.)

It's germanium era stuff, and that suggests the HV rectifiers in the multiplier stack may well be selenium. I'd be surprised if selenium rectifiers of that age still worked.  If not, it should be possible to replace them with more modern silicon equivalents. In fact the voltage per stage is not that high and the frequency quite low, so common-or-garden 1N4007's would probably do as replacements. 

[Nekopan]

Would you be able to let me know which components are the rectifiers? If I understand which section is the high voltage section, then I was under the impression that it was all caps and diodes.

[Rerouter]

by rectifier, he means diodes. in the voltage multiplier stage.

[Nekopan]

Right. Yes, those all checked out but I've heard that they're basically pushed to the limit for that circuit. I'll be doing my frequency probing tonight, and if that checks out I'll look into replacing the diodes.

[Nekopan]

So, I finally got around to probing the unit. Now, I had to guess for the right spot, or else I'd have to tear down most of the unit. I found pin 2 off the transformer, since on the top of the board the connection lead right to a transistor. My multimeter measured 1852.0Hz, which seems to be fairly close to the rated 1.9KHz. is that discrepancy large enough to throw off the CRT voltage?

[Kleinstein]

The frequency is not that critical. So 1.85 kHz is OK. The interesting part would be when the unit get warm (and the display disappears) - in this case the frequency could change.

The high voltage diodes would be some of the first places to start with Si based diodes. Selenium based diodes are only low voltage each and would need quite a few in series to withstand 300 V. It was done (e.g power supply for tube based radios), but that were rather large parts. Changing those to modern Si Diodes would need more than just a diode, to compensate for the higher voltage drop.

A picture could give an indication what type of diodes / rectifier is used. The 32 V transformer winding might have an old one. A photo would also help to make sure to remember the polarity of parts and to keep an idea how it looked when original. With such old parts the original look might be an important aspect too - sometimes more important than functionality.