Single transformer for positive and negative ground circuits

[merlinb]

(The angry buzz is fairly objectionable.  Different parts of the core saturate slightly differently which creates field gradients and field gradients lead to high forces which cause the buzz -> I'm not sure about this now.)

With the massively lop-sided primary current the magnetostriction will be lopsided too, hence the buzz.

[Rob Strand]

With the massively lop-sided primary current the magnetostriction will be lopsided too, hence the buzz.

I'm not sure really.  I was thinking  Lorentz Force (the force on conductor).

Magnetostriction usually decreases with saturation.   The Lorentz Force  usually goes up under load (because the there's more current in the conductors) and under saturation (since more field escapes from the core and goes through the windings).  You get the same thing with a electric welder.  A hum when sitting there and a strong  noise under load.

For example, (there's few papers on the web about it).
https://www.comsol.com/paper/download/63432/kavasoglu_paper.pdf
EDIT: Actually this one was better,  (I've got a few but I didn't keep the links)
http://aip.scitation.org/doi/pdf/10.1063/1.4978759

I was going to pull out the old transformer and see if I could press on the windings, or even impregnate them with lacquer and see what happens.

[DimebuGG]

After weeks of being busy of my day job, I was able to work finally on the schematics suggested by Rob and Jack. It works flawlessly without any problems so far. I'm using a 500mA transformer dual 9V center-tapped which I paralleled. Thanks so much guys. 

[merlinb]

The invisible process going on is the continual increase in the core's B-field. 

This reminds me of bias shift in an RC-coupled amp that clips on one side. The core hits saturation on the 'tips' of the AC waveform and develops a DC component to balance it out, just like a coupling cap charges up until the clipped tips (charging) balance out the downgoing waveform halves (discharging).
If your imagination is only looking at the steady-state part of the waveform it's easy to miss the clipped tips and think your waveform is barely close to the clipping threshold (i.e. saturation), when in fact it is heavy clipping that has driven it into that state in the first place.

[merlinb]

Thanks for the heads up guys. I think the DMM draws about 50mA current and the distortion circuit probably draws around 10-15mA max.
Meanwhile, I found this:

What magazine is that from? Got a reference?

[Rob Strand]

This reminds me of bias shift in an RC-coupled amp that clips on one side. The core hits saturation on the 'tips' of the AC waveform and develops a DC component to balance it out, just like a coupling cap charges up until the clipped tips (charging) balance out the downgoing waveform halves (discharging).

Pretty much the same mechanism, one sided clipping.  Some RF oscillators use the one-sided clipping thing to "set" the output level.

This thread seems so long ago.