I'm unsure if anyone has seen this: US5343080A - Harmonic cancellation system
- Google Patents
In short it's for cancellation of power harmonics within 3 phase power. However the approach essentially adjusts the phasing to cancel out harmonics within the transformer core.
So I'm wondering if that's possible to cancel harmonics in an audio transformer in the same way? (either tube or magnetic amp)
- Google Patents
In short it's for cancellation of power harmonics within 3 phase power. However the approach essentially adjusts the phasing to cancel out harmonics within the transformer core.
So I'm wondering if that's possible to cancel harmonics in an audio transformer in the same way? (either tube or magnetic amp)
Harmonic cancellation is possible in most parts of an amp, although I think the transformer would be the last place I would try it! Not being a transformer winder, it is the thing over which I have the least control. I suspect you could achieve a similar end-result by looking at the system as a whole, and making a tweak somewhere within the valve circuit instead. It's a lot easier to change a resistor or two, than to add or remove turns.
Moreover, I suspect the method in the patent can only be tuned to work at one fundamental frequency.
Moreover, I suspect the method in the patent can only be tuned to work at one fundamental frequency.
Last edited:
Assuming it can be done you will surely lose efficiency because the extra windings take space and you will have to use smaller wire for any winding. So you will need a bigger transformer for same efficiency and power handling. In OPT it has to work on a wide bandwidth equally well. Typical phase shift at 20Hz and 20KHz in OPT's is of the order of few degs in best cases so around there the shift you apply on purpose will not work. Not just at 20Hz and 20KHz but even 40Hz and 15KHz, so to speak. I would not think that one does it in midrange and doesn't care about the low and high ends. Not only the increased frequency differential in distortion might make things worse but there could be other side effects given the complexity of such a transformer. A really good regular OPT is already complex respect to a power transformer. The AC mains works at fixed frequency. Totally different animal. That's my first impression of course. I might be wrong but can't see it as a practical solution.
The Berning ZOTL is a much better solution. If want a challenge try to replicate this!🙂
The Berning ZOTL is a much better solution. If want a challenge try to replicate this!🙂
Agreed - I did have the fact it's focused on a single frequency at the back of my mind and that the effect may not be linear with frequency, so you may not be able to apply it as a function.
Yep I did see the ZOTL schematic - it reminds me of a current chopper 🙂. I've used PWM current choppers to control low voltage servo phases (2.3V) with 12+V. The extra voltage overcomes the EMF (required to ensure the servo switches step) and the high frequency chopper prevents the current from cooking wire. At 12V the servo would be too hot to touch and would fail, with the chopper the servo is both cold to touch, has more torque, is more accurate and will not fail. I use one on the focus system of my telescope where it holds several lbs worth of equipment on the end of the focuser in a vertical position without batting an eyelid. The torque does need end limit switches and a safety to stop things being crushed 😱 The system then auto-focuses very precisely.
I digress I'll look at the ZOTL in more depth.
Edit: Quick look here: https://davidberning.com/images/microZOTL/microZOTL_Manual.pdf looks like he's using a full rectified full bridge (Q1-4 as one such section) to get 450V but limit the current. It also looks like he's using a back of inductors to store energy (that's the bussed inductors up the side). Combo booster 🙂 That bank of boosters can be varied using the mosfets associated. If the power is directly connected then this would then alter the impedance seen through the power supply by the load (speakers).
Yep I did see the ZOTL schematic - it reminds me of a current chopper 🙂. I've used PWM current choppers to control low voltage servo phases (2.3V) with 12+V. The extra voltage overcomes the EMF (required to ensure the servo switches step) and the high frequency chopper prevents the current from cooking wire. At 12V the servo would be too hot to touch and would fail, with the chopper the servo is both cold to touch, has more torque, is more accurate and will not fail. I use one on the focus system of my telescope where it holds several lbs worth of equipment on the end of the focuser in a vertical position without batting an eyelid. The torque does need end limit switches and a safety to stop things being crushed 😱 The system then auto-focuses very precisely.
I digress I'll look at the ZOTL in more depth.
Edit: Quick look here: https://davidberning.com/images/microZOTL/microZOTL_Manual.pdf looks like he's using a full rectified full bridge (Q1-4 as one such section) to get 450V but limit the current. It also looks like he's using a back of inductors to store energy (that's the bussed inductors up the side). Combo booster 🙂 That bank of boosters can be varied using the mosfets associated. If the power is directly connected then this would then alter the impedance seen through the power supply by the load (speakers).
Last edited:
In OTs the primary source of harmonic generation is the core related magnetizing current, loading down the output tube's Zout.
One can use CFB to pre-program in the extra tube current for driving the magnetizing current. Or, similarly, one can just put a small inductor (shunted by an R) in the cathode, with the same core material and flux density, to pre-program in the extra tube current. (acting like CFB, except speaker load effect not included)
Low output Z will work similarly to suppress magnetizing current effects too, of course, but its just so much more elegant to pre-program in the correction in the first place. (feed-forward) One can also use a little current Fdbk from the output tube cathodes to the driver stage to null out the OT winding resistance. You never see these corrections being made in tube amps, but these can remove the need for global N Fdbk altogether. (local N Fdbk used to linearize the tubes, or Crazy drive for full feed-forward)
One can use CFB to pre-program in the extra tube current for driving the magnetizing current. Or, similarly, one can just put a small inductor (shunted by an R) in the cathode, with the same core material and flux density, to pre-program in the extra tube current. (acting like CFB, except speaker load effect not included)
Low output Z will work similarly to suppress magnetizing current effects too, of course, but its just so much more elegant to pre-program in the correction in the first place. (feed-forward) One can also use a little current Fdbk from the output tube cathodes to the driver stage to null out the OT winding resistance. You never see these corrections being made in tube amps, but these can remove the need for global N Fdbk altogether. (local N Fdbk used to linearize the tubes, or Crazy drive for full feed-forward)
Last edited:
How do you propose to phase-shift by the same angle over 20Hz..20kHz? Its a transformer, not a Hilbert transformer (signal processing joke!).
Harmonic cancellation occurs in many 3 phase systems.
Notably, a change in connection, delta vs Wye/Star, will give cancellation of some harmonics, if I recall, 3rd is depressed.
It's worth noting too, that some harmonics are cancelled at the Neutral node in a Star connected system.
I don't think this technique is in any way useful for audio, as it is just a foible of the way 3 phase systems work. A happy coincidence for power quality applications.
Notably, a change in connection, delta vs Wye/Star, will give cancellation of some harmonics, if I recall, 3rd is depressed.
It's worth noting too, that some harmonics are cancelled at the Neutral node in a Star connected system.
I don't think this technique is in any way useful for audio, as it is just a foible of the way 3 phase systems work. A happy coincidence for power quality applications.
I may. I may not.
The biggest issue, would be maintaining any cancellation, across the bandwidth of operation, as is usually the case.
The biggest issue, would be maintaining any cancellation, across the bandwidth of operation, as is usually the case.
Removing the OT primary winding resistance is -part- of the solution to removing transformer distortion (magnetizing current and hysteresis current drops across the prmary winding resistance).
That just takes some current sampling resistors in the output tube cathodes, fed back to the driver stage cathodes to make some negative R predrive.
The other part of the problem is the magnetizing current and hysteresis current cause a V drop across the Rp of the output tubes. This can be fixed with either lowering the Rout of the tubes, using CFB, or by placing a selected sampling inductor (made with same material and flux density as the OT) across the sampling resistors in the output tube cathodes. This will develop a pre-drive signal for the driver stage that provides the extra current waveform to supply the OT with feed-forward magnetizing current correction.
A small capacitor across the cathode R can remove the effects of leakage L in the OT as well, by genreating a compensating HF pre-drive V.
The secondary winding resistance of the OT is best removed using a global N Fdbk sampling resistor, since that only has load current thru it (no magnetizing current)
All the complexity of switchmode ZOTL (Berning) OTs and filtering are un-necessary. Just a few components added to a conventional tube amp's cathode circuitry can fix the same problems, without introducing a bunch of high frequency transients, poor output impedance (leakage L in the ferrite xfmrs suffered at the HF carrier) and phase problems for global N Fdbk.
SET amps desperately need these fixes, with the high magnetizing currents there. But apparently the OT 2nd and 3rd H distortion is what the listeners like.
Last edited:
I guess that has some truth.
Myself, I will try and minimise THD in SET amps (I have yet to build a PP)
I am generally happy if I can get the THD to less than 1%, with as low 2nd and 3rd as I can achieve, using my less imaginative methods. (Copying what many others have done)
But then, I will typically include plate to grid feedback in ALL stages, which I guess is something most people would not agree with, either.
Myself, I will try and minimise THD in SET amps (I have yet to build a PP)
I am generally happy if I can get the THD to less than 1%, with as low 2nd and 3rd as I can achieve, using my less imaginative methods. (Copying what many others have done)
But then, I will typically include plate to grid feedback in ALL stages, which I guess is something most people would not agree with, either.
Well, although I agree that your solution is a lot more convenient I disagree on the "desperately". It's more a technical issue than practical. It becomes relevant more and more at frequencies below 30Hz where there is almost nothing to reproduce and most speakers barely reproduce too, in the best case. I don't think people like distortion, it's just a minor issue usually inflated for marketing reasons....
My emphasis was meant for the extreme purists who want every little phantom flaw fixed. Being able to actually hear them clearly is another matter.
I just wanted to point out how relatively easily these issues can be fixed for ordinary OTs, without resorting to extreme measures like switch-mode hysteresis fixes. P-P OTs have so little magnetizing current to begin with that no one even considered fixing the problem, let alone the hysteresis portion (small % of magnetizing current) until modern times.
It's also intellectually interesting that most distortion or output impedance mechanisms in classical tube amps can be fixed fairly well using "local" feedbacks or feedforwards without having to go through an expensive OT with high bandwidth for global N Fdbk. (Although some global N Fdbk is quite helpful for finishing details, and it's available to a safe and useful extent generally.)
It really should be possible to make a high performance tube amp with some cheap stuff and elegant design fixes. (however, I'm not too excited by using toroid power xfmrs for OTs with insufficient primary inductance and poor bass response. For a one time $20 or $30 extra, its worth having a decent OT that will sound good for years and years.)
I just wanted to point out how relatively easily these issues can be fixed for ordinary OTs, without resorting to extreme measures like switch-mode hysteresis fixes. P-P OTs have so little magnetizing current to begin with that no one even considered fixing the problem, let alone the hysteresis portion (small % of magnetizing current) until modern times.
It's also intellectually interesting that most distortion or output impedance mechanisms in classical tube amps can be fixed fairly well using "local" feedbacks or feedforwards without having to go through an expensive OT with high bandwidth for global N Fdbk. (Although some global N Fdbk is quite helpful for finishing details, and it's available to a safe and useful extent generally.)
It really should be possible to make a high performance tube amp with some cheap stuff and elegant design fixes. (however, I'm not too excited by using toroid power xfmrs for OTs with insufficient primary inductance and poor bass response. For a one time $20 or $30 extra, its worth having a decent OT that will sound good for years and years.)
Last edited: