Detecting DC Component in Transformer Flux

Hi,
I am driving a transformer at 15khz with a square wave of +-200Volts. I wound a auxiliary coil on the transformer to measure what is applied to the primary. I want to be able to detect a DC component in the voltage that I apply to the primary and correct it in the next cycle by applying positive or negative voltage for a longer time to balance the flux.
My first idea is posted here. The aux winding is 1:.5 and i used a diff op amp configuration to step down voltage to +-10v. Next stage is a MFB low pass active filter -3db set at 150Hz. The gain of MFB filter is inverse of the step down stage. I have attached the schematic. It does simulate well, but as of now not working in hardware. I an open to ideas to detected the flux in a transformer.

I would rather not sense both pri and sec currect to get magnetizing current because i would be treating the consequence instead of the cause.

How do you expect the auxiliary coil to pick up the DC component of the primary winding without actually being attached to it?

Bearing in mind that a current is not generated in a wire by a static field only a dynamic field.

You might see some symmetry if the BH curve of the core had been shifted but not other wise.

If you apply 202 volts for half a cycle and -200 volts for the next half cycle, there would be a 1 volt DC component. I want to measure that by using a LPF on the aux winding which is coupled to primary, i should see that same wave on aux coil and be able to detect the offset.

Don't think in absolute volts with an assumed zero crossing point.

Think in terms of change of flux which will be the same positive as it will negative.

Or to put it another way, in old valve radios they had transformers with +/-20V of AC signal sitting on around 200V DC, the output of the transformer at the speaker certainly had no DC component in it.

Seriously get your oscilliscope out and check the waveforms or if you have a multimeter that works across the audio band use that.

There are only two ways you will get a DC component out of a normaly working transformer. The first is if the AC waveform is significantly asymetrical the second is if the core starts to saturate on one half cycle.

You need a step down integrating differential amplifier with its inputs connected to the winding you are driving, the diff amp can have filtering and you can post filter also - this will give you a DC value (delayed a bit by the filtering action) which corresponds to the net DC volt-seconds applied to the Tx wdg and hence the net DC flux, also it will be in phase with the DC imbalance. If you then create a ckt to alter the volt seconds applied to the wdg so that the net effect is to always move the o/p of the diff amp towards zero then you will achieve your ends - we have implemented this for grid connected inverters (3kW) to keep the DC injected into the mains < 5mA. If you want to absolutely keep the average flux zero in the core, you need to keep the average current in the wdg zero, using a sense resistor and the same method as above (with a low frequency isolating Tx to give isolation from the driven winding), or you can have a seperate cut toroid with an hall effect sensor in it and a suitable winding on the toroid in parallel with your driven winding, the toroid (or gapped E core) mimics the Tx.
Hope this helps, Regards, Orson Cart.

Orson thanks for you comments, i would like to know more, how do they isolate the opamp that is sensing the Vprimary? Using an LEM? opisolator? isolated opamp?

After further thought, you are correct i cannot detect dc component using a aux coil. I am going to run my primary currents through matlab and see where the average lines up.
Thanks for you help. Orson i would like to discuss more on implementation side.
Also as you see orson i step down in one stage and in the second stage there is a gain to offset the step down and its a LPF. I am not confident thise should be done because i will get more noise.