Common mode filtering for heater supplies.

[jnb]

I am curious as to how others filter common-mode noise from their heater supplies.

I am using balanced capacitors at the heater pins and want to introduce a common mode choke per supply. I have read that they could be used before a regulator, after a regulator or at the heater pins.

Which of these locations is more appropriate, or is all three recommended?

[zigzagflux]

What I do, and if I recall, what Morgan Jones recommends, is to have a common mode choke after the regulator. Keep your caps right at the pins of the tube socket.

You end up with a L-C in each leg to ground. Common mode noise is reduced, and it would seem to me any spurious regulator noise is attenuated as a bonus.

I suppose you could use them prior to the regulator as well, but I question its effectiveness. I use a choke prior to the regulator for 120Hz ripple reduction, not RF filtering.

[artnace@cox.net]

Think about this for minute, the noise generator is the heater soure - transformer/diodes and possibly anything that makes it across the power transformer. Depending on the quality of the transformer, one could get coupling from the high voltage windings due to audio loading. This could justify a separate transformer for the filaments. The heater is the load. We want to reject the common mode noise before it gets to the heater from the soure. Thus, we must add some series impedance between the soure and the heater (cm choke) and lower the impedance from the heater source to ground. Thus, we must add a cap on the source side to ground. The capacitive reactance of the cap should be much lower then the inductive reactance of the choke at the frequency of interest, say 60 Hz. Adding caps on the heater side will work if the cm
capacitance is greater than the capacitance from the filament to other tube elements. I would suggest 10X the highest capacitance. The common mode noise is usually 60Hz and harmonics plus any stuff comming across the power transformer. Thus, the -3bd break point of the filter should be 6 Hz of less. I suggest using LTspice to simulate the filter. With a quality capacitance meter, you should be able to measure the capacitance from the heater to the other elements.