Novel method to reduce filament hum in SE Amp?

[dshortt9]

Suppose I use two DHT triodes in parallel for a SE amp.
More power, reduced Rp, etc.

Instead of using one filament transformer I use two.

One end of each filament is grounded and the other end is fed by a seperate transformer but 180 degrees out of phase.

Will this effectively cancel the hum and/or AC filament induced IM products?

Can I then use AC instead of DC for the filaments?

Should I have another cup of coffee?

[tomchr]

There is no such thing as too much coffee. That would be like breathing too much air. You should definitely have another cup!!

Now, about the hum cancellation. If all the parasitics are equal (which they won't be) and all the induction paths for the mains hum are identical then I'd say the hum should be reduced.
But I wonder what would happen if the two signals aren't perfectly 180 degrees out of phase. I bet you'd introduce a bunch of new intermodulation products. I suppose one could get some math together in Mathcad/Matlab/Octal/whatever and figure it out that way.

You could also go with a DC supply. No hum there (assuming decent line regulation and such).

~Tom

[Miles Prower]

DC heater power would be a better way to go. No hum, so no need to balance out hum. Just make sure that the heater DC is well filtered and regulated. Dirty DC is much worse than clean AC for heating DHTs. With clean AC, you're just adding the one frequency, whereas dirty DC includes lots of harmonics of the ripple frequency. Lots more frequencies to intermodulate with the signal frequencies. Not a good thing.

[danielrs]

In some occasion, after biased the tube with the hum-pot , there is usually small level of hum if you listen closer to the speaker > 95dB and above.

Using AC Filament with lower hum level.

Below is a forced balanced filament supply circuit, which can lower the hum level to the significant level (1/3 of original value). This case, you don't need a hum-por but you can use fixed balancing resistors and plus additional use of 2 diodes in series with resistors like arrangement in the attached picture. The value of the series resistor is about 2-3 times of the fixed balancing resistor.

[Hearinspace]

My take on it would be - why not build a one channel breadboard and find out?

[korneluk]

Originally posted by danielrs

Below is a forced balanced filament supply circuit, which can lower the hum level to the significant level (1/3 of original value). This case, you don't need a hum-por but you can use fixed balancing resistors and plus additional use of 2 diodes in series with resistors like arrangement in the attached picture. The value of the series resistor is about 2-3 times of the fixed balancing resistor.

What kind of diodes do you recommend for this circuit?

Thx,

[danielrs]

You can use any kind of diodes in the circuit. In my case for single 300B, i use ultrafast diodes MUR420 4A 200V from motorola.

[revintage]

Did a sim with a 6B4G SE A and the proposed resistors/diodes and in this case ripple caused by the heater at output of the amp is reduced by 50%. From 35mV to 16mV peak. Still I would go for DC-heater fed by a CCS.

[kevinkr]

Originally posted by revintage
Did a sim with a 6B4G SE A and the proposed resistors/diodes and in this case ripple caused by the heater at output of the amp is reduced by 50%. From 35mV to 16mV peak. Still I would go for DC-heater fed by a CCS.

I second the CCS, use it all the time, and my amplifiers are silent. (Total measured noise including hum and ripple on 8 ohm tap <0.4mVpp..)

[Michael Koster]

With DHTs there is 60 Hz ripple introduced by the filament voltage
which can as shown be reduced by balancing each filament; no
need to cancel it tube-to-tube.

There is also 120 Hz (power) hum that can't be balanced out.

Best way to reduce both is DC ;-)

Michael