Hi,
a delayed on and instant off for the bypass relay is OK for a few seconds of no power.
For a few hundred mS of loss you would need a fast acting instant off and a fast acting relay to get the resistor and/or NTC back into circuit in time.
For a one cycle loss, one would need a missing pulse detector and a very fast bypass, probably not a relay, maybe a FET bypass.
But, I had not realised the significance of a missing part cycle.
Would some kind of phase comparator be able to release in time for the re-establishing of power?
a delayed on and instant off for the bypass relay is OK for a few seconds of no power.
For a few hundred mS of loss you would need a fast acting instant off and a fast acting relay to get the resistor and/or NTC back into circuit in time.
For a one cycle loss, one would need a missing pulse detector and a very fast bypass, probably not a relay, maybe a FET bypass.
But, I had not realised the significance of a missing part cycle.
Would some kind of phase comparator be able to release in time for the re-establishing of power?
kl, Is this a hypothetical question or do you have a problem with poor mains. I used a 600va toroid for years, but found I needed at least a 6.3 AT fuse. Even then it "parted" once on switch on.
It is possible as Andrew say's to have a missing pulse detector, but is it worth it all ? Providing the secondaries are suitably protected any risks are minimal with a large primary fuse, indeed the toroid manufacturer possibly has info on recommended input protection. Relays are OK but the contacts need to be suitably rated to prevent arcing.
What seems a simple problem can either be " a non problem " or if you are determined to do something about it can become a fairly major design exercise.
It is possible as Andrew say's to have a missing pulse detector, but is it worth it all ? Providing the secondaries are suitably protected any risks are minimal with a large primary fuse, indeed the toroid manufacturer possibly has info on recommended input protection. Relays are OK but the contacts need to be suitably rated to prevent arcing.
What seems a simple problem can either be " a non problem " or if you are determined to do something about it can become a fairly major design exercise.
A PTC thermistor and relay to short it out gives the ideal result. Burst control using triacs can generate RF noise and "irregularities" at the zero crossing. It's a no win situation really. The thermistor and relay offer no delay in the event of a mains drop out, but does that matter, the caps in the amp wil be charged so the toroid isn't seeing a low impedance.
it's not the smoothing cap load that is the problem.
That is solved differently.
It's the loss of flux in the transformer on power failure that we are trying to solve. A relay cannot drop out instantly. The best are of the order of 2mS and big relays that can handle input currents are very possibly >10mS to drop out.
This drop out time gets much worse the more effectively we snub (with a diode) the back EMF of the coil.
If the problem is as bad as that posted graph shows, then a relay can never solve that quarter cycle current surge. A whole cycle loss of power may even be too short to detect and for the relay to open in the 20mS available to detect and drop out.
That is solved differently.
It's the loss of flux in the transformer on power failure that we are trying to solve. A relay cannot drop out instantly. The best are of the order of 2mS and big relays that can handle input currents are very possibly >10mS to drop out.
This drop out time gets much worse the more effectively we snub (with a diode) the back EMF of the coil.
If the problem is as bad as that posted graph shows, then a relay can never solve that quarter cycle current surge. A whole cycle loss of power may even be too short to detect and for the relay to open in the 20mS available to detect and drop out.
No easy answer is there ? To be honest I think it's a non problem really. I mentioned the reservoir caps because from fully discharged it must increase the primary current over many cycles I would imagine. If the secondary circuit wiring is low impedance it may well " overload " the transformer briefly, although I have never done any measurements.
Relays will never work for drop outs of a cycle or two.
Anyone any data from the toroid manufacturers on what they recommend ?
Relays will never work for drop outs of a cycle or two.
Anyone any data from the toroid manufacturers on what they recommend ?
Mooly said:
I don't know what toroid manufactures recommend. But last night I got the idea of using PWM to address all the toroid problems at the same time. I can maybe begin to work on it when I get my new computer to replace the failed, one year old, one. This computer is 4 years old and simulations take too long.
megajocke said:
Motors and EI core transformers usually have enough leakage inductance, it seems, to limit inrush current due to series impedance. But that area-wide mains loading-down effect you mentioned would seem to help mitigate the problem.
When I got my 2kva toroids, I thought they would be easy to tame. I sure was surprised when, even though I had a 2 amp circuit breaker in series with the primary, it had such a dead-short impedance when I plugged it in, the little circuit breaker didn't trip while the 20 amp service panel breaker did.
Don't create a problem if you not have any.kl said:I found some info at noinrush.com. It seems that conventional soft start circuits and NTC are useless in case of voltage dips. I will try to redesign my softstart.
My design with a relay with a fast reset will pick most situations. Loss of part of period, how common is that really?
all this is nicely solved by using a device ( TSR) that brings the operating point at maximum remanance by pulsing the transformer with unipolar pulses then switch on. This avoids the high inrush current happening if the core is saturates while switching on.
http://www.emeko.de/index.php?id=29&L=1
JPV
http://www.emeko.de/index.php?id=29&L=1
JPV
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.