DHT Filament Hum Cancellation

4. Composite indirectly-heated driver tube

1. Abstract

In this article we SPICE several designs providing cancellation of DHT AC filament induced hum tube amplifiers. The results are demonstrated with a 2-stage SET topology but are applicable to a wider range of tube amplifier designs.

2. Introduction

Directy heated triodes (a.k.a DHT) with AC-fed filaments are prone to generate unfiltered residual 120Hz (in counties with 60Hz mains) hum, as sjown in []. The hum is most troublesome in in low-power single-ended triode (a.k.a SET) amplifiers working with high-eeficiency speakers. In such settings 1mV of hum on the output is troublesome. in [] we have shown that harmonic ditortion is the main reason for the hum.

Here we present several designs, all being variations of classical 2-stage SET topology, that reduce or completely eliminate the DHT hum. We preset each ide with SPICE simulation. most of these ideas were implemented with modified Moondog 2A3 SET amplifier and resulted in reduction of amplifier's hum to or below audible levels when tested with very high efficiency speakers (sensitivity is above 103 dB/W/M).

3. Hum-complimentary DHT stages

If 120Hz hums is a property of Directy Heated Triodes (DHT), how about cancelling the noise of one with the noise of another. This is, in fact, exactly how it happens in push-pull DHT stages, where hum is very small. See, for instance, [hf60-dht].

Another direction to go opens up after one can realise that typical SE topology involves signal polarity flips. That is, a plate-loaded driver stage inverts signals injected into its grid or cathode. Thus, we can develop a hum-like signal on driver which can cancel hum generated by the output tube - iff the phase and amplitude match perfectly.

4. Composite indirectly-heated driver tube

We've shown that composite tube models help to explain and model DHT AC hum. We also built composite models for several popular triodes. A related technique is to emulate a DHT tube by means of indirectly heated triodes (IHT) with AC signal gradient present on cathodes, for whic hmore than one tube is required. We've shown that 2 tubes is sufficient []. In this section we show several practical designs based on the idea of using a composite tube with AC on cathodes as a driver for a DHT power singel-ended (SE) stage.

The first design uses AC comping from the taps of the center-tapped transformer of the split-load B+ rectifier circuit. Such circuits are very common in SET amplifiers and the Moondog uses that, too.

Here, best cancellation happens for R14 and R15 equal to ~ 97K

Above, the value of parameter R runs from 80K to 120K by 5K, and as shown, cancellation occurs with R ~ 97K. When cancellation happens the AC voltage on cathodes of 6SN7s is ~ .95V amplitude.

The next design is very similar but uses a low-voltage secondary tap of the amplifier's transformer. Any winding that is floating and provides > 1VAC will be sufficient.

R14 and R15 are ~ 1...2k when cancellation happens.

The winding that feeds the heaters of 6sn7 tube is an excellent candidate to be wired this way.

A very elegant variation of this approach is that the taps are connected directly to the cathodes of the tubes, thus perfectly modeling a DHT, but this has proved to be sensitive to quality of the transformer; in some cases this injects RF noise into the circuit; still such variation is temptive as it is very close to how a DHT driver works. One addition twist is to use the output tube winding. This would elevate the DC level of the cathodes of 6sn7, though. Here is the idea:

Note this wiring bypasses the driver tubes. Yet the cancellation will happen, too, The values of R14 and R15 are ~ 1k.

Each design presented will work "as-is" only when the driver triodes are a perfect match; otherwise some balancing using R14 and R15 is needed.

5. Conclusions and future work

6. References

[hf60-dht] 6B4G PP amp based on Mullard toplogy: a mod to Eico HF-60

Author: Dmitry Nizhegorodov (dmitrynizh@hotmail.com). My other projects and articles