I'm in the process of repairing a piece of equipment with ferroresonant transformers supplying some large linear regulators. The transformer hums pretty loudly for a brief instant before the supply shuts off due to undervoltage protections kicking in.
The caps are 20uf at 660vac and 10uf at 660vac. They measure within spec on a handheld meter but that's hardly representative of their in circuit use. I connected them to a regulated high voltage dc supply and watched the current on my 5.5 digit bench meter. The leakage current is unreadable, it looks below the noise floor of my meter. How can that be? They also seem to hold a charge just fine. For reference a small 0.1uf oil cap shows steady leakage in the microamp range which is expected. I just can't believe these old caps have an undetectable amount of leakage. 😕
The caps are 20uf at 660vac and 10uf at 660vac. They measure within spec on a handheld meter but that's hardly representative of their in circuit use. I connected them to a regulated high voltage dc supply and watched the current on my 5.5 digit bench meter. The leakage current is unreadable, it looks below the noise floor of my meter. How can that be? They also seem to hold a charge just fine. For reference a small 0.1uf oil cap shows steady leakage in the microamp range which is expected. I just can't believe these old caps have an undetectable amount of leakage. 😕
If the caps are okay, problem is elsewhere.
The transformers are so called constant voltage transformers?
You can charge up the caps with an external supply, and discharge them with a small 15 watt or so bulb, see how long it takes to drain. That will give you some idea of their condition.
Do this when out of circuit with the rest of the equipment.
Read up on CVT if not familiar with them.
The transformers are so called constant voltage transformers?
You can charge up the caps with an external supply, and discharge them with a small 15 watt or so bulb, see how long it takes to drain. That will give you some idea of their condition.
Do this when out of circuit with the rest of the equipment.
Read up on CVT if not familiar with them.
Did your DC test supply go at least up to 1kV ?
Did you have a current limiting resistor in series with the DC supply - placing that in the grounded leg of your setup and measuring the voltage across that resistor is safer all round. Poly caps often start to leak in 'pops' as a pin-hole starts to conduct and then gets repaired - that may be the better indicator of how the dielectric has degraded or how far above nominal rated Vpk you can go.
Is it a poly encapsulated cap (ie. not metal can)?
Did you have a current limiting resistor in series with the DC supply - placing that in the grounded leg of your setup and measuring the voltage across that resistor is safer all round. Poly caps often start to leak in 'pops' as a pin-hole starts to conduct and then gets repaired - that may be the better indicator of how the dielectric has degraded or how far above nominal rated Vpk you can go.
Is it a poly encapsulated cap (ie. not metal can)?
The caps in question are metal can types and 40 years old. They spent some time in a warehouse without climate control too. trobbins I will try the resistor idea and measure leakage that way. My power supply is limited to 400vdc though. The ferroresonant transformers operate from 240vac.
This is all part of a large old computer. Right now there are currently three functioning examples remaining so getting this one restored to working order would be an accomplishment.
This is all part of a large old computer. Right now there are currently three functioning examples remaining so getting this one restored to working order would be an accomplishment.
These caps measure impossibly low leakage numbers. A 1meg ohm resistor is in series and the power supply is set for 100vdc. I measure 150mv across that resistor. That would mean the leakage is 1.5e-7 amps! For reference the datasheet for an equivalent modern oil cap says leakage should be 30uA maximum. Yeah something seems off here.
Which means *MINIMUM* 933V peak so bare minimum rating must be equivalent to 1kV DC which means they certainly must be HiPot tested to 3KV DC
You test them at
You call that *a TEST?*
Why do results surprise you?
Make a voltage multiplier getting 2 to 3kV DC out (at least) and connect its output to cap terminals through a 1M resistor made out of 10 x 100k 1W resistors in series which then I put inside a transparent plastic tube (fuel line type) filled with epoxy/grease/hot melt.
Leave 1 100k resistor free at one end to measure voltage drop across it to know leakage current and after capacitor is fully charged (current drops to a minimum), disconnect HV supply (unplug it from Mains) , let capacitor self discharge for a few minutes, say 5 or 10, and measure remaining voltage at the free resistor chain end to ground, accounting for the 1M series resistance and your meter input impedance.
This is a variation of what I use to measure *real* voltage rating for electrolytics.
After measuring, derate them approppriately, measured voltage is literally "breakdown" voltage, not to be approached under *any* circumstances.
Same high voltage resistor/probe (I use them to make HV probes and attenuators, specifically to scope plates of overdriven Guitar Tube amps where 1500V peaks are everyday fodder) can be then used to safely discharge capacitor under test.
When you measure voltage with the 1M resistor in series, it will visibly fall because you are discharging the cap, the initial value is your rating.
Did that testing include capacitance? Ferroresonant circuits need a certain level of capacitance to function correctly.
As they are metal can encapsulated, an additional test is for leakage to the can, which as JMF indicates should also be done at a suitably high voltage level (ie. at least 1kV).
It may be worth your while borrowing an insulation resistance meter that at least goes to 1kV and check all relevant parts and circuits in case it is not the caps.
Last edited:
Would that be an old DEC computer? A engineer friend said he used to slow an AC induction fan motor by placing a cap in series with the AC line. It shifted the phase of the current, so the motor ran slower. His claim for the technique was that the capacitor dissipated zero power.
Certainly 660V oil filled caps shouldnt give you much to worry about in a circuit like this. Perhaps you can test them this way - at least you'd have voltages and current levels you could - carefully - measure.
Unsure how you'd model the motor part in something like LTSpice, to verify your readings are what's supposed to be for the 10 and 20uF values.
Certainly 660V oil filled caps shouldnt give you much to worry about in a circuit like this. Perhaps you can test them this way - at least you'd have voltages and current levels you could - carefully - measure.
Unsure how you'd model the motor part in something like LTSpice, to verify your readings are what's supposed to be for the 10 and 20uF values.
Let's put these on the same basis.
1.5e-7 is 0.15uA, isn't it?
Which is 1/20th of the number you found on some other cap. (Caps for SOLAs are really required to be much better than other caps of "same" value and rating; it's a rough life and often a long one.) It is not unusual for a well-aged sample to be 20X better than the factory reject number. They try to sort-out the weak ones. And in metalized caps, part of the game is to burn-out the weak sections of insulation leaving only solid insulation, often in a "burn-in" period.
But don't obsess these caps. IMHO they are super unlikely to be your problem. Divide and conquer. A DC power is a chain of separable sections.
Hmmm so these caps may indeed be good. They read 10uf and 20uf respectively on a handheld meter and exhibit very low leakage and hold a charge just fine. I was testing at 100vdc because if a cap is leaky it's probably going to be leaky at any voltage.
No it's not a DEC system but another big name. We aren't the owner but it is on permanent loan to the museum. So I can't mention the details yet.
The documentation says that a leaky cap in the ferroresonant transformer will probably cause an undervoltage condition and that brought about my initial testing.
The system has an initial pilot supply that comes up fine and drives the control logic and supervisor circuits. There are 5 large constant voltage transformers with multiple secondaries all feeding linear regulators. There's probably 20 really large can capacitors inside these supplies. The error codes are saying the three voltages supplied from this transformer are not within spec during initialization so power is cut and the sequence stopped. The diodes and filter caps are good. So next I'm going to scope the regulator voltages and see what they're doing. There are no spare parts so troubleshooting has to be done with a light touch. We have excellent documentation but no actual schematics for the regulators.
No it's not a DEC system but another big name. We aren't the owner but it is on permanent loan to the museum. So I can't mention the details yet.
The documentation says that a leaky cap in the ferroresonant transformer will probably cause an undervoltage condition and that brought about my initial testing.
The system has an initial pilot supply that comes up fine and drives the control logic and supervisor circuits. There are 5 large constant voltage transformers with multiple secondaries all feeding linear regulators. There's probably 20 really large can capacitors inside these supplies. The error codes are saying the three voltages supplied from this transformer are not within spec during initialization so power is cut and the sequence stopped. The diodes and filter caps are good. So next I'm going to scope the regulator voltages and see what they're doing. There are no spare parts so troubleshooting has to be done with a light touch. We have excellent documentation but no actual schematics for the regulators.
I'd expect leakage to increase with applied voltage for what appears to be paper/foil caps in oil - in which case you won't really know if they are degraded until you can test with an insulation resistance meter (or equivalent).
Voltages from the transformer? Or on the capacitors? Or is the documentation ambiguous? (The tech writer leapt from effect to cause.)
I remember those BIG DC caps. IIRC they did not age well. I'd bring them up on a 12VDC supply through a 100 Ohm resistor and see if they went to 11.9V real quick or stalled at a few Volts. While an exact replacement for a failed big-can cap would be distressing, they beg to be hole-sawed open and filled with modern (smaller, maybe better) snap-caps. Looks vintage at arm-length, but up close no intent to deceive.
The basic SOLA CV transformer goes out of regulation if under-loaded. There were many variants and some were pretty good down to no-load. But some slight load may be good to try.
Last edited:
The cap reduces the voltage across the motor, nothing to do with phase, and not the best way to slow down an ac induction motor, but a lot easier than changing the freq.
Speed control methods of induction motor | electricaleasy.com
>There's probably 20 really large can capacitors inside these supplies.
Just dont get your wedding ring or a small screwdriver across the bars those things are connected to. I've heard it's happened!
Just dont get your wedding ring or a small screwdriver across the bars those things are connected to. I've heard it's happened!
Here the current production ceiling fan regulators use series caps of different values to get different speeds which are selected by a switch.
Earlier ones uses a BT136 or 131 triacs, switch would be fixed resistors or a pot.
The earliest of course were series resistors.
Large government orders now insist on capacitor based regulators.
I have been using them past 15 odd years, no problems.
Earlier ones uses a BT136 or 131 triacs, switch would be fixed resistors or a pot.
The earliest of course were series resistors.
Large government orders now insist on capacitor based regulators.
I have been using them past 15 odd years, no problems.
Another use I have seen is to put capacitors in the 3 phase windings of a 3 phase motor, so as to run it off a single phase supply, the first shifts phase by 120 to the second phase, and a second capacitor from that achieves a second shift to the third winding.
This is not recommended for continued use, but motor winders some times use it to test motors after repair.
This is not recommended for continued use, but motor winders some times use it to test motors after repair.