Transformer Inrush Currents 4

The reason for the inrush current is that the transformer core sees more volt-seconds than it was designed for.

As you probably have noticed, the inrush current seems to come randomly. Sometimes there is no inrush at all and sometimes it trips the breaker. The reason is that connection to the 240 V is sometimes made at the beginning of the sinewave, sometimes on top of the sinewave and sometimes somewhere in-between.

It is when connecting at the beginning of the sinewave that you get the most inrush (core sees about twice the design flux), so arranging the switching so that connection is on top of the sinewave will reduce inrush to zero.

There are SSRs with the ability to switch on top of the sinewave - in fact they were designed to eliminate inrush current. There are also NTC thermistors that can be used for the same purpose. Google, google - and google.

Gunnar Englund

A 1000A seems a little high for a 10A transformer, you would normally expect it to be around 20A if the secondary was unloaded. The secondary load could be the problem, charging capacitors, which means you need a different solution, there are several easy ways to kill capacitor inrush currents.

An old ham radio trick (used in power supplies) is to have a series resistor on the AC line to limit the in-rush / start-up current surge, and a relay that shorts out that series resistor once one of the conveniently low-voltage secondary DC voltages rises sufficiently to pull-in the relay. So, when you switch on the PS, there is a second click a couple of seconds later as the relay kicks in.

Skogs has provided links towads more-modern solutions...

Long long ago, not far from here I learned that the inrush current in a transformer was due to the core magnetization remaining from when it was last switched off.

If you switch it on and it happens that the incoming volts are crossing zero and will tend to magnetize the core in the same direction then the core saturates.

Once it has saturated, no inductive effect & thus all you are left with is the dc resistance of the winding.

Which will give you a very large current indeed.

This is the same sort of effect that is used in mag amps and saturable reactors and stuff like that, so I understand.

You could just install a larger circuit breaker upstream (along with requisite larger conductors). This is what is normally done, since the inrush is normal behavior.

I would just throw that transformer into the nearest trash can.

It is not difficult to design a transformer that has an acceptable inrush current from the unavoidable flux doubling that skogsgurra describes.

The problem is that the magnetic core can go into hard saturation, leaving only the dc resistance of the primary winding to limit the peak inrush current over the first few cycles. The way to reduce the problem is to use a magnetic core material with a much softer saturation characteristic, and design it for a lower operating flux density.

The best way to do it is with E and I laminations, using ordinary cheap transformer iron, and keep the flux density low by using more primary and secondary turns (more turns per volt). The resulting transformer will be larger and heavier but it will be far less aggressive at turn on.

The worst possible way, is to use a high tech grain oriented silicon iron tape wound core, and run the working flux density right up very high. Many transformer manufacturers these days, do it that way to end up with a small hot running transformer at the lowest possible cost to them. It is cheaper, because there will be far less turns and less copper even though the core will be more expensive.

It should be possible to find something commercially available if you know what to look for, or go and talk to a transformer winding company and tell them your problem.

Resistors, relays, NTC thermistors will all work, but may be unreliable or inconvenient. much better to find yourself a decent transformer rather than trying to band-aid rubbish.

scogs,

>It is when connecting at the beginning of the sinewave that you get the most inrush (core sees about twice the design flux), so arranging the switching so that connection is on top of the sinewave will reduce inrush to zero.

Am I being a bit thick here? You say the worst place to switch a transformer on is when the supply voltage is at zero volts. Huh? Surely the worst place to switch it on is at the peak of the sinewave? I accept that an unloaded transformer will have the current and voltage 90 degrees out of phase when the AC transient has decayed, making the worst place to switch it off at zero volts, but I disagree about the switch-on point.

Maximum peak magnetic flux is the point of zero change, and that corresponds to the zero voltage point.

If you switched it off there, the remnant flux remains as residual magnetism in the core. Next time you turn it on, if you try to drive the flux further in the same direction, the core will saturate.

It is confusing, because transformer magnetising current (and flux) are always exactly ninety degrees out of phase with the voltage. When one reaches a peak, the other is always at zero.

Its a small enuff transformer..just install a proper breaker ..and call it a day as dpc said. What is the rating of the breaker that trips?

HI logbook, scogs is quite right about the best point to switch on the transformer.

Ok so we all agree that the worst place to switch off the transformer is the highest current point which occurs at the voltage zero crossing point.

Scogs, warpspeed, and Cbarn have the worst switch on point at zero crossing and I am outnumbered three to one in thinking that the if you turn a transformer on at zero crossing there will be the least possible transient. That sort of situation indicates a misunderstanding (which may be mine!) but which is worthy of discussion.

There is a lot of talk of the current and voltage being 90 degrees out of phase, but that does not occur at switch on. Remember that at switch on the volatge is sinusoidal but the current only becomes sinusoidal after ther transient has died out.

Now I could simulate this on SPICE using a linear inductor. Would this result be satisfactory or are we saying that the major problem here is the non-linearity and remnant magnetism in the core? I am not sure if my simulator can handle non-linear inductors.

Hi logbook,

In the absence of other replies, I'd suggest that the transient component causes the problem because of core saturation. The transient is (or can be) a significant DC offset which will shift an ungapped core into saturation. Most transformer cores are ungapped unless they are specifcally designed to handle a DC component.


----------------------------------
I don't suffer from insanity. I enjoy it...

Hi logbook, you seem to have stumbled upon the answer yourself in your 3rd paragraph. You agree that conditions at switch on for the "zero turn on case" are different from normal operation, ie. no voltage and no flux. Now consider the "max voltage on case" and you will see that you have the same conditions at swich on that you would have if the transformer was operating normaly, ie. max voltage no flux.

Sorry that I wasn't awake to catch some of the misunderstandings early enough. But I see that several other workers have done most of the answering for me. Thanks. I still think that a clarification may be needed.

First: Transformer cores are designed to work between a negative maximum flux (Bmin) and a positive maximum flux (Bmax). Outside these limits, the core is saturated and loses its inductivity so that impedance goes to zero - or very close to.

Second: In steady state (i.e. not at switch on), the flux moves between these two limits. And since induced voltage is proportional to flux derivative it follows that the flux is at its maximums when the derivative (or voltage) is at zero.

Third: Since flux is at its maximum at zero volts, there is a flux change equal delta-B = Bmax - Bmin (B used for field intensity) or twice Bmax. This is what saves the transformer's day; it has 2xBmax available to handle the positive half-period's volt-seconds and -2xBmax to handle the negative half-period's volt-seconds.

Fourth: If the transformer is connected at the beginning of the period (flux is zero and not -Bmax) it will have used up the available flux change shortly after 90 degrees (a quarter period) and goes into saturation - which results in inrush current. If the breaker doesn't trip, it will take a few periods before the DC component has died out and current is normal again.

Fifth: This has very little to do with when the transformer was disconnected. The remanence in a modern transformer core is low and the state when disconnected usually doesn't have much influence on the inrush current. It used to be a "popular" explanation several decades ago. It may have been valid then, but it is not now.



Gunnar Englund

Ok, after some simulations of the AC transients I see what is going on. An inductor as a load gives no current at all initially at switch on, regardless of the supply voltage phase. If it is switched on at peak supply there seems to be no additional transient, the steady state condition being reached immediately.

If it is switched on at zero volts the transient shifts the whole current waveform up, doubling the peak aiming value of the current/flux. Any normal core would then saturate making the situation worse. Having a higher series resistance damps the AC transient more rapidly. The inductor then simulates an unloaded transformer.

If the transformer is loaded such that the load is several times the magnetising current the "inductor" effects seem to get damped very rapidly. The switch-on current is then pretty much independent of the switch-on phase.

But all of that is with linear magnetics.

We therefore have unanimous agreement
Thanks guys.

Great! Can you also make the transformer core non-linear? Saturating at about 1.5 Tesla?

Gunnar Englund

Hi skoggs, nearly right but the core does not have to go into saturation to produce inrush current.

You better give me a very good explanation there before I believe you, cbarn. Do you mean that a linear core (not saturating) could produce a 10 times, or more, inrush current?

Are you not confusing the DC component with inrush?

The DC component exists as soon as you connect anywhere different from 90 (or 270) degrees and it can produce some extra current. It is only when the DC component saturates the core that you get inrush current. So you really need a saturable core to get inrush.

Gunnar Englund

No skoggs that isn't what I said. You do not need a saturating core to get inrush it's a basic function of transformer action and amounts to a maximum of 2 times steady state current. If the core does saturate then the current will rise further than that also the inrush properties of the load on the secondary can add even more to it. Logbook has his simulator going so I'm sure he can verify that for you or you can just get a transformer and experiment if your still not convinced.