Transformer switch on saturation

[Mike Sulzer]

I made some measurements on a Hammond 182P12; this is a 120VA toroidal unit. Saturation is more frequent than you might expect. Higher voltage transients last longer as you would expect.

[Mike Sulzer]

Here is how these measurements are made:

A 0.5 ohm resistor is inserted in series with the primary of the transformer under test (TUT). The primary of a smaller power transformer is placed across this resistor, and it is its secondary (24V) that is fed to the A/D converter. This provides isolation from the line. Also a 4 diode clipper (~1.2V) with a series resistor is used to further protect the A/D. This transformer might slow down the rise time of the saturation transients, but I am not willing to risk the hardware. The data are collected with the scope tool of electroacoustic tool box with a sampling interval of about 5.2 microseconds using an Apogee Duet.

The current measurement is calibrated by connecting an 8 ohm resistor to the 24V secondary of the TUT and deriving a scale factor.

The magnetizing current of this transformer, once the transient has died out, is small.

[R.G.]

Nice work, Mike.

[nickb]

Here's a bit more data. I simply used a current probe so it should be fairly accurate.This was a 500VA toroid. The decay time is much shorter, I guess the L/R ratio is different. Note the scale 5A/div.

Updated: I forgot to switch the probe to DC and terminate the scope.

[trobbins]

Mike, if you use a 1 ohm sense do you get double the recorded signal level (that would indicate that the sensed signals are not being clipped).

What is the DCR of the transformer primary, and the mains voltage?

Were you just using a simple toggle mains AC switch?

[Mike Sulzer]

Nice work, Mike.

Thanks, R.G. My goal is to measure a transformer intended to supply a pair of EL34s at the 50 watt level, but not until after I get it mounted on a chassis (for safety reasons).

[Mike Sulzer]

Here's a bit more data. I simply used a current probe so it should be fairly accurate.This was a 500VA toroid. The decay time is much shorter, I guess the L/R ratio is different. Note the scale 5A/div.

Updated: I forgot to switch the probe to DC and terminate the scope.

[ATTACH=CONFIG]39511[/ATTACH]

That is a nice measurement!

[Mike Sulzer]

Mike, if you use a 1 ohm sense do you get double the recorded signal level (that would indicate that the sensed signals are not being clipped).

What is the DCR of the transformer primary, and the mains voltage?

Were you just using a simple toggle mains AC switch?

Initially I had severe clipping, but I thought I had attenuated enough to get of it. I checked again, and it appears that there still is clipping on some transients some of the time. I will try to get rid of that this weekend.

The transformer primary is about 3.1 ohms, about 117 volts, although the line variation is substantial here in Puerto Rico.

Edit: Yes, I am using an ordinary toggle switch for switching.

[R.G.]

The transformer primary is about 3.1 ohms, about 117 volts, although the line variation is substantial here in Puerto Rico.

To make some guesses at what you might possibly see, if the line voltage went up to 125Vac and the remanence of the core shifted the saturation point back to the peak of the AC line (both are only barely possible, but will help bound the problem), you'd see maximum possible peaks of 125*1.414/3 = 58.9A. So if you can read 60A without clipping, you're good to go.

[Mike Sulzer]

To make some guesses at what you might possibly see, if the line voltage went up to 125Vac and the remanence of the core shifted the saturation point back to the peak of the AC line (both are only barely possible, but will help bound the problem), you'd see maximum possible peaks of 125*1.414/3 = 58.9A. So if you can read 60A without clipping, you're good to go.

I just measured 33 A with 120V line. I am now using a .125 ohm series resistor and a +/- 2.4 volt clipper. (The highest peak is about a volt below the clipping threshold.) Plots coming sometime later.

[nickb]

That is a nice measurement!

Thx. I had originally done this about a year ago when someone came in with a Mixer/PA that would cause the breaker to pop at a particular venue. I couldn't find the original data so I had to do it again. To solve his problem I added a 'soft start' board which put 100 ohms in series for about 1 second and then shorts it out.

I guess we should calculate I^2.T to answer the original fuse question At least I think that's where this all started.

[Mike Sulzer]

Here is a big current transient. The secondary voltage is also shown. The red lines mark the approximate locations of zero crossings.

1. The switch was thrown.
2. There is a bit of bounce.
3. Transformer operation begins.
4. Core saturates and the primary current rises quickly.
5. Transformer operation is interrupted.
6. Current peaks and decreases as the primary voltage decreases.
7. By the next zero crossing the saturation certainly has ended and transformer operation resumes.

It seems to work as expected!

[mhuss]

Is this effect purely due to core saturation? I had a thought experiment (feel free to shoot it down ): Would not an un-energized transformer look like just the coil resistance? Until current starts flowing and there's some back EMF happening due to inductance, it's just a 3.1 ohm resistance.

I'm pretty sure all inductors surge a bit when AC is first applied; once a field forms (a continually varying field in the case of AC), then the inductive reactance kicks in.

[trobbins]

Mike, your measurement is an AC signal, so the lack of DC reference causes an AC averaging to be added to the 'current waveform' trace - which viewers will need to compensate for.

I suggest the 16ms repetitive current pulses are an on-going, but damped, inrush response. The magnetising current waveform is down in the noise given your earlier result for magnetising current.

If you apply something like a 50% or 100% full load rated resistive load to the transformer secondary, then you could determine a ratio between in-rush peak and full-load amp peak.

[mozwell]

Dear All
Transformer inrush can easily be 10x normal full load current.
The inrush will vary depending on where in the AC cycle the voltage is actually applied, and where in the AC cycle the voltage was previously turned off.
The inrush will also depend on how high the magnetic flux is being run on the transformer. Magnetic flux is determined by the mains supply voltage (and the transformers design itself)
If mains voltage is high, say 125V on a 110V winding, the inrush will be higher than if the supply was 110V.
In my experience toroidal transformers usually have a higher inrush than EI transformers as they have lower reactance.

So when the mains is turned off, depending on where the actual voltage was, we have a small "left over" bit of magnetism in the steel core & that will stay there.
When we turn it on again, if the voltage is the same polarity, we "add" to the leftover flux, and this can push the transformer into saturation. It takes a few cycles of the mains for the flux to "stabilize" back to normal levels. If the mains voltage is high, the extra push is higher & we go deeper into saturation & draw more inrush current. If the mains voltage is of the opposite polarity, we "subtract" the leftover flux, and we will have a lower inrush.
All this is very variable, unless you switch on / off at zero crossing.
Also, if the secondary is connected to a rectifier & big electrolytic capacitors, they will add to the inrush as we have to charge them from zero.
If you want to measure inrush only, do this with no load on the secondary.

If the inrush is a concern for you, then a soft start circuit, as others here have described, is the way to go.