Transformer Core Saturation

[poynt99]

OK guys.

I believe what we are seeing here is core saturation. Would you agree?

Luc is running phased-shifted current through a MOT as shown. With the secondary shorted, we see relatively clean sine waves for the current and voltage. When the short is lifted, we get what appears to me as core saturation. I don't know why it would be going into saturation though.

Orange is voltage, blue is current, red is v * i.

[BEP]

IMO, you're seeing the inductor the transformer became when the secondary was shorted.
When the secondary is open you're seeing the transformer doing a very poor job of trying to be a CVT because the capacitance of the magnetron isn't there.
What looks like saturation is probably the magnetic shunts kicking in.

This will limit output current to a very small extent and may make the output appear like pulsed DC.
If all he wants is HV it may be worth a try to knock the magnetic shunts out.

These things are useful but deadly.

[poynt99]

Thanks WW.

It looks like it might be core saturation indeed, but not sure it is caused by the magnetic shunts.

http://wiki.4hv.org/index.php/Microwave_oven_transformer

The transformer is designed to be as cheap to manufacture as possible, with no regard for efficiency. This is because it is the manufacturer who pays for the copper and iron, but the user who pays for the energy consumed. Thus the iron area is minimised which results in the core being taken well into saturation with result high core losses. The copper area is also minimised, resulting in high copper losses. The heat that these generate is handled by forced air cooling, usually by the same fan that is required to cool the magnetron. The core saturation is not part of the non-ideal classification, it is merely as a result of the economics of manufacture.

[Magluvin]

I was thinking that the second shot with the sec shorted is the thin pos and neg traces of power and current, are just the fact that the cap is charging up sooner than the cycle of the input.  With the sec not shorted, the primary inductance is high, so cap charges slower.  Then when sec is shorted, the primary inductance goes down, cap fills quicker, so action stops till the input goes out of phase. So putting bigger caps in, if the spiky bumps widen with sec shorted, most likely thats the case.  

Mags

[Magluvin]

What Im basically saying is it seems the second shot is of a shorted sec and the first open.  Doesnt make sense the other way. But maybe this is not a normal circuit.

Mags

[BEP]

IMO, the wiki paragraph pertains to a MOT powering a magnetron. Yes, they do saturate easily when loaded and during initial inrush, but when unloaded?

Edit:

Ok, I see what you mean. Yes, that is evidence of saturation - even unloaded.

My mention of the magnetic shunts was because they affect core saturation with built-in leakage. The effect is part of the constant voltage function but problematic when that expected capacitive load isn't there.

[poynt99]

WW, Mags,

I think it may be a combination of the two; saturation and power pulsing due to the phase shift.

Thanks for your input guys.

[ion]

I am in total agreement with this post from Poynt:

The transformer is designed to be as cheap to manufacture as possible, with no regard for efficiency. This is because it is the manufacturer who pays for the copper and iron, but the user who pays for the energy consumed. Thus the iron area is minimised which results in the core being taken well into saturation with result high core losses. The copper area is also minimised, resulting in high copper losses. The heat that these generate is handled by forced air cooling, usually by the same fan that is required to cool the magnetron. The core saturation is not part of the non-ideal classification, it is merely as a result of the economics of manufacture.

While this is true of most application specific transformers where the expected steady state load is known, it is not true of general purpose wide range  laboratory or industrial transformers that must have sufficient primary inductance loaded or unloaded to support the mains voltage and frequency without saturation.

Note that a transformer that begins to enter saturation with no load will come out of saturation  when loading is introduced and when loaded up to the normal transformer rating.

Dual path transformers (MOT's or Neon sign transformers with shunts) tend to have high leakage inductance by design due to the nature of the loading.

[Peterae]

Hence the fan needed to cool the transformer even though run times are relatively short while cooking things.

[BEP]

Hence the fan needed to cool the transformer even though run times are relatively short while cooking things.

Yes, these things are probably the worst example of transformer design just to save money on manufacture. By design, they are intended to shift the cost to the consumer. Hence the same fan that cools the magnetron generally also cools the transformer.

However, they are the easiest to modify for other applications. Removing the designed-in magnetic leakage is normally the first thing I do with them. It is rare that I need constant voltage out in a project. I've even been known to add iron to the core.

They also make pretty good ballasts when you short the secondary. It helps prevent a single MOT from blowing the mains breaker when it doesn't have the load it is designed for.

For comparison, I have a 'calibrated' PT (potential transformer) formerly used in a calibration lab. It has 120/240 on one side and is tapped for 120/240/480/600/1200/2400/3300/4160/10k/13.8/14.4kV on the other side. When the lab 5S'd it, it was I all I had in me to put it in my truck and carry it down to my basement bench. I doubt I'll ever saturate the bastard and it is only rated 250VA.

Not having powered it up yet, I'm worried that an inrush limiter may be needed.

[poynt99]

EM,

I think from the scope shots that the current (blue) is leading the voltage.

[BEP]

Is channel 1 truly measuring mains instead of transformer voltage - as shown on the schematic?

[poynt99]

I believe so WW.

[poynt99]

I am in total agreement with this post from Poynt:

While this is true of most application specific transformers where the expected steady state load is known, it is not true of general purpose wide range  laboratory or industrial transformers that must have sufficient primary inductance loaded or unloaded to support the mains voltage and frequency without saturation.

Note that a transformer that begins to enter saturation with no load will come out of saturation  when loading is introduced and when loaded up to the normal transformer rating.

Dual path transformers (MOT's or Neon sign transformers with shunts) tend to have high leakage inductance by design due to the nature of the loading.

Thanks for your insights ION.

[gyula]

....
When the short is lifted, we get what appears to me as core saturation. I don't know why it would be going into saturation though.

Hi poynt99,

To answer this,  we have to consider the AC impedances of the components and their combined resulting Z impedance as is known for a series RLC circuit. Luc used 25 uF capacitor, this has about 106 Ohm reactance at 60 Hz, this would (in itself) draw 249/106=2.34 Amper from his 249V mains.
But as per Luc's scope shots, the current draw from the 249V source was 3.27 A for the shorted secondary and 7.68 A for the open secondary MOT circuit (a 10x multiplier should be used for the CH2 scopeshots as Luc wrote).

It is clear that for the shorted secondary case, the primary coil of the MOT must have a lower X_L inductive reactance than for the unshorted secondary case of course.

So in the shorted case, the 106 Ohm capacitive reactance was reduced only by a smaller amount of inductive reactance than in the unshorted case, thus the resulting total Z impedance of the series RLC network was higher in the shorted case than in the unshorted case, this is because the absolute value of the difference of (X_L-X_C) remains higher when X_L becomes smaller and X_C is constant.

For the open secondary case, the X_L inductive reactance of the primary coil surely was higher than in the shorted case so that the difference of the (X_L-X_C) value becomes lower, this means the resulting Z impedance also becomes lower vs the shorted case: this means a higher current draw from the mains (I=V/Z), than in the shorted case.  And a higher current draw inherently biases the MOT core more heavily than a smaller current, this is why we can see the sharp current peaks in the open secondary scopeshot, indicating the beginning of core saturation.

Putting all this otherwise: by connecting any coil with an inductive reactance between 0 and 106 Ohm (at 60 Hz) in series with a 106 Ohm capacitive reactance, then the resulting Z impedance of this series RLC would change like this: the closer the coil reactance approaches the 106 Ohm capacitive reactance, the lower the Z impedance becomes. When the coil reaches 106 Ohm reactance, then series resonance occurs at 60 Hz and the maximum current flows into the series circuit, limited mainly by the DC resistance of the coil and that of the AC source (and by the series load in Luc's circuit).

Gyula

[poynt99]

Hi gyula,

Thanks for the explanation. I had not sat down to think about the details, but I agree 100% with your analysis.

[ion]

Very nice and accurate analysis Gyula.