Center-tap Choke

Hi everyone.

Just made some tests with different inductors in place between the power transformer center-tap and the filter caps, with no choke or inductor in place (straight wire) the voltage over the caps are about +- 36v.

When i add an inductor the positive side gos to 39v and the negative goes to 34.4v, how is this possible? the power supply i tested this with were on a gainclone typ amp.

Thanks, Sonny.
I made a schematic for some clarity =) if i remove the choke (just a straight wire) the voltage over the capacitors become even, +36v / -36v.

With the choke in place the voltage becomes +39v / -34.4v.

Can try to change phases on mains/choke etc. but i doubt that is going to make any difference.

Thanks, Sonny.


  • pwr.jpg
    30.1 KB · Views: 176
Are you drawing exactly the same current from each rail? i.e. nothing in the circuit connects to the 0V rail, or if it does it draws only negligible current.

You have, in effect, two choke input supplies but with the added complication that the choke is common so it sees the difference current. Choke input supplies are very sensitive to current drawn until the current exceeds a threshold value, at which point the DC output is 90% of the secondary RMS voltage. Your supply will be very sensitive to the difference current.

What are you trying to achieve?
Just made some tests and measurements. There isn't much unbalance in capacitance between the positive and negative rail, about 7uF. Sure thing if you load a choke input supply lightly it will more or less become a capacitor input filter. I'm not sure if the LM3875 draws equal from + and - rail, but it should because it's not power grounded.

However it's strange that i get an increase in voltage on the positive rail from 36v to 39v :scratch:

I'm trying to achieve better sound ;) i want to isolate the signal ground from the noisy power transformer (and mains crap) and to decrease the CT ripple current.
If you take a balanced current draw then there will be no CT ripple current anyway. If this is present then it means an imbalance somewhere, either current drawn or the CT is not actually in the centre of the winding.

Reversing the choke connections will only have an effect if the choke is seriously magnetically coupled to the transformer, which is unlikely.

To get rid of mains noise just use a mains filter on the primary.
Just reversed the choke without any difference in voltage, same 39v positive and 34v on the negative rail. Yes the choke would be meaningless if the currentdraw was symmetric, but music is highly asymmetrical.

EDIT: If the transformer would have an unbalanced CT would it really create an even voltage over both caps without the choke in place? I hand a different transformer in that circuit once (W/O choke) and it gave some 0.5v voltage difference on the caps.
Last edited:
Put a load on the + and - outputs and see if the voltage becomes balanced.
This is very strange. There is an AC component on the center tap until the diodes conduct. If the voltage drop across the diodes isn't equal, it could cause some unusual voltage readings.

Tried to put a 170 ohm resistor over the +- rails (~73v) no big difference except some voltage drop, sure due to heavier loading.

I agree that its a little strange, it must be the mains transformer somehow, what else could it be?

Well i think it might sound more "headroomy" with some more aggressive bass dynamics after listening to this for some time now.
If the transformer secondary is unbalanced then you would get the same DC voltage on either side, equal to the peak voltage of the higher secondary side - in effect it would become a half-wave rectifier as the other (lower) half would never conduct. You could check with a scope - is the ripple at mains freq or twice mains freq?

Music being asymmetrical would not create an imbalance as music is still AC i.e. no DC component.
Try a pair of resistors. + to common and - to common. Keep the load equal on both outputs.
It's unlikely that the power transformer is the issue. The output is balanced when the choke is removed from the common connection.
I'm wondering if the core of the choke might have become magnetized.
Something is causing the choke's non-linear response.
I've got a couple of old chokes. I'm going to experiment.
Sorry i have no scope, would really like to have one right now =/
I did measure the transformers secondaries however, both "sides" did measure 27.0v AC and 38mH inductance so they seem very balanced indeed.

Can it be some kind of grounding problem? the amp is not grounded to any wall socket ground (neither is the DAC connected to the amp). But it's really strange that i get more voltage on one cap then possible (with cap input secondary vRMSx1.41)
A choke on the center tap of the secondary will not stop the noise spikes.
You might add a capacitor across the full secondary of the transformer.
With the proper capacitor value, you can resonate the transformer to your line frequency. This will increase the transformer's output voltage somewhat but the cap will also attenuate the voltage spikes, putting them on each half of the secondary equally.
I still got uneven voltages over the caps with a resistor from both + and - rails to common ground. I'm gonna try a different choke or two to the if there would be any difference, maybe even an air-core one to null any core "non-linear" problems.

Could it be the asymmetrical electrodes of the aluminum electrolyte capacitors? (though they are black gate F:s which are less asymmetrical then standard electrolytes)
The extra voltage on one side is caused by the voltage across the choke adding to one side and subtracting from the other. There is probably not much point in getting to the bottom of the issue, as the choke is not the best solution for your original aim anyway.

Drop the choke, and add a mains filter. In addition, add some proper grounding arrangements.