DC heater elevation question

[R.G.]

That was one of my points.
Any ripple in the DC filament supply will be amplified.
In a lower gain amp, this is not a big problem.
In a high gain amp, you really need to get that ripple out completely.

Just to be accurate, any ripple on the DC filament supply will be amplified *after* any losses it has in jumping from the filaments to the signal path. This is just like the losses in getting the normal 6.3Vac from the heater windings to the signal path.

A straightforward 12AX7 amplifier does not get 6Vac of heater "ripple" into its signal path, but rather something very, very much smaller. I'll have to go look up the actual numbers, but I believe that it was tens of microvolts making the jump across from the heaters. Tubes are in fact designed to make the transmission of heater AC to cathode or grid be as small as it's practical to do, including using the center tapped heater on the 12A?7 series to get some internal cancellation of the hum field.

As a result is possible to make high gain amps with AC heater supplies that have very small ripple if you take great care with the wiring to make the transmission of heater voltage be only the transmission inside the tube.

So by using any heater signal with smaller ripple than the 8.9V peak of a 6.3Vrms heater supply, you get ripple reduction on the output to the extent that the ripple is less than 8.9V peak. If your DC heater supply has, for instance, 100mV peak to peak of ripple, then the ripple in the signal path of the tube is smaller by 0.05/8.9 = 0.0056.. or a reduction of over 99%. Likewise, any ripple induced from the wiring is reduced by the same factor. So any substantial reduction in the peak to peak AC signal on the heater supply reduces the feedthrough by the same amount. A 10:1 reduction in ripple on the heaters from the 17.8V peak to peak of a 6.3Vac signal to only 1.78V peak to peak reduces the transmitted ripple across the heater/cathode barrier by the same 20db.

Zero ripple is nice, but 20-40db noise reduction is easy and may be enough. In any thoughtful design, it is quite difficult to design to an absolute, like zero ripple. Good design knows when "good enough" has been reached.

Important to remember, the filament noise is not coming from ALL the preamp tubes. Just the first stages. Not the phase inverter.

Actually, the noise is coming from all the tubes, it's just that noise from any given tube is amplified by all the following tubes, so the noise of the first stage predominates by the gain of the first tube at a minimum.

It's also worth remembering that the reverb recovery tube is a high gain input when you're picking tubes for special treatment.

Yes, cancellation method DOES work, very well. You can't argue with results.

Yep, cancellation does work, but it's hugely dependent on getting the cancellation just right. I'll have to go look up the reference, as this is quite old.

It takes a slightly distorted 60Hz signal from its power supply secondary, then uses filters to extract the 60Hz, 120Hz, 180Hz, and 240Hz parts of the waveform. These are then individually run into pot-controlled allpass phase shift filters, and added together to produce a hum cancellation signal. In operation, one turns the volume down on all four channels, brings up the 60Hz a bit, then diddles the phase until hum goes down, then messes with the volume, then phase, etc. until hum is as low as possible. After that, you do the 120, 180, and 240. By the time this is done, the hum is significantly reduced.

It's a super deluxe version of what the old hum-balance pot on the heater windings tried to do, and works great... if you have the patience to retune it every time something changes, like amp location/orientation, pickup orientation/location, etc. It's pretty good for tuning out hum generated inside the amp though. Note that this is not an intricate fiddling with this wire here and this one there, it's a signal simply dumped into the amp at an input, and it's adjusted to cancel whatever hum is there. I've built one, and it works very well.

However, stopping hum is much, much better than trying to cancel it out once it's already in the amp, because if anything changes, cancellation adds noise, rather than reducing it.

[trobbins]

One aspect of ac on a heater is that transfer of the ac to cathode via capacitance coupling will accentuate any higher frequency components in the heater voltage. Mains frequency heater voltage is likely to have pretty low levels of harmonics (either originating from the mains itself, or coupled over from the HT rectification loading). Whereas rectifier generated ripple has both a low frequency content from the period, and a higher frequency content from the narrow charging pulse (without even getting in to diode recovery influences).

The transfer of ac to cathode via heater-cathode insulation resistance is tube and voltage dependant - tube to tube variances can be significant, as can the resistance variation characteristic with voltage. If the insulation resistance is luckily high then the capacitance pathway dominates the transfer level from tubes I've measured.

[soundguruman]

Just to be accurate, any ripple on the DC filament supply will be amplified *after* any losses it has in jumping from the filaments to the signal path. This is just like the losses in getting the normal 6.3Vac from the heater windings to the signal path.

A straightforward 12AX7 amplifier does not get 6Vac of heater "ripple" into its signal path, but rather something very, very much smaller. I'll have to go look up the actual numbers, but I believe that it was tens of microvolts making the jump across from the heaters. Tubes are in fact designed to make the transmission of heater AC to cathode or grid be as small as it's practical to do, including using the center tapped heater on the 12A?7 series to get some internal cancellation of the hum field.

As a result is possible to make high gain amps with AC heater supplies that have very small ripple if you take great care with the wiring to make the transmission of heater voltage be only the transmission inside the tube.

So by using any heater signal with smaller ripple than the 8.9V peak of a 6.3Vrms heater supply, you get ripple reduction on the output to the extent that the ripple is less than 8.9V peak. If your DC heater supply has, for instance, 100mV peak to peak of ripple, then the ripple in the signal path of the tube is smaller by 0.05/8.9 = 0.0056.. or a reduction of over 99%. Likewise, any ripple induced from the wiring is reduced by the same factor. So any substantial reduction in the peak to peak AC signal on the heater supply reduces the feedthrough by the same amount. A 10:1 reduction in ripple on the heaters from the 17.8V peak to peak of a 6.3Vac signal to only 1.78V peak to peak reduces the transmitted ripple across the heater/cathode barrier by the same 20db.

Zero ripple is nice, but 20-40db noise reduction is easy and may be enough. In any thoughtful design, it is quite difficult to design to an absolute, like zero ripple. Good design knows when "good enough" has been reached.


Actually, the noise is coming from all the tubes, it's just that noise from any given tube is amplified by all the following tubes, so the noise of the first stage predominates by the gain of the first tube at a minimum.

It's also worth remembering that the reverb recovery tube is a high gain input when you're picking tubes for special treatment.


Yep, cancellation does work, but it's hugely dependent on getting the cancellation just right. I'll have to go look up the reference, as this is quite old.

It takes a slightly distorted 60Hz signal from its power supply secondary, then uses filters to extract the 60Hz, 120Hz, 180Hz, and 240Hz parts of the waveform. These are then individually run into pot-controlled allpass phase shift filters, and added together to produce a hum cancellation signal. In operation, one turns the volume down on all four channels, brings up the 60Hz a bit, then diddles the phase until hum goes down, then messes with the volume, then phase, etc. until hum is as low as possible. After that, you do the 120, 180, and 240. By the time this is done, the hum is significantly reduced.

It's a super deluxe version of what the old hum-balance pot on the heater windings tried to do, and works great... if you have the patience to retune it every time something changes, like amp location/orientation, pickup orientation/location, etc. It's pretty good for tuning out hum generated inside the amp though. Note that this is not an intricate fiddling with this wire here and this one there, it's a signal simply dumped into the amp at an input, and it's adjusted to cancel whatever hum is there. I've built one, and it works very well.

However, stopping hum is much, much better than trying to cancel it out once it's already in the amp, because if anything changes, cancellation adds noise, rather than reducing it.

Your not going to stop it before it gets in the amp, especially with single coil pickups...effects units, etc...

The ripple is amplified regardless. The noise drives guitar player crazy.
But I disagree wholly that it's the same as a hum balance pot. The hum balance is only effective to a degree. Not to the degree that guitar player wants, or would prefer.

The DC or cancellation method cleans up what the hum balance leaves behind. What the hum balance can't hope to do.

Cancellation method has the advantage because there are no parts, or DC supplies and filters, being added to the circuit. That's the beauty of it.

The cancellation method is only used by more experienced builders...
The DC method is more friendly to inexperienced.
AND the cancellation method is not "as" applicable to PC boards, which have poor layout, and parts in fixed locations.

When you have a PC board with poor layout, the DC method becomes one of the "few" choices. Or, the only choice.
And that's why Bogner, Boogie, Carvin, Peavey, etc. are all using it. The alternatives are impractical.

So, in my example...the Marshall cathode follower V2 is quite noisy, in all plexi amps, there is a very loud BUZZ. Always has been...
And my choice is to run the filament on DC for V1, V2. OR to use cancellation method.
AND cancellation solved the problem without adding any parts, and without installing a DC supply and without spending more money...

So, that was a beautiful, simple, cheap solution that worked very effectively. The filament buzz is reduced dramatically, without spending more money.
I get what I want for free.

[R.G.]

Your not going to stop it before it gets in the amp, especially with single coil pickups...effects units, etc...

Not the point. Hum inside the signal coming into the amp is signal, and will get amplified. It *is* signal when it comes in on the input cord.

And AC is going to be inside the amp from the AC mains. Stopping it means making the coupling mechanisms that get it into the signal path be very inefficient. The difference between no hum and hum 120db down from the signal is more philosophical than practical.

The ripple is amplified regardless. The noise drives guitar player crazy.

Excellent non-sequitur. Yes, any ripple that gets into the audio path is amplified. HOW MUCH gets in is a big deal, as is HOW MUCH it gets amplified. Ultimately, if it's inaudible, guitar players stay sane. And amplified ripple does not necessarily drive guitar players crazy; it's a matter of degree. And beyond that, many people would argue that guitar players are already crazy in general.

But I disagree wholly that it's the same as a hum balance pot.

You may want to re-read my post. I actually said a number of things, none of which were that anything was the same as a hum balance pot. What I said was that there was a device I'd seen that allowed you to cancel out not only the mains hum, but it's harmonics as well as shifting the phase of each harmonic to get the lowest possible hum. I've built it, and it's very good. It can, for instance, even cancel hum coming in from a single coil, at least to the degree that the single coil hum doesn't change, and that's a big deal. You may disagree if you like, but do try to read and understand what you're disagreeing with first.

The hum balance is only effective to a degree. Not to the degree that guitar player wants, or would prefer.

The hum balance of WHAT? A humdinger? The variable-phase multi-harmonic canceller? WHAT? One if the problems with a humdinger is that it cannot shift the phase of the induced signal, so if the hum in the signal path has been shifted a bit, complete cancellation is impossible. And a humdinger can only cancel the mains frequency that the heaters run on, not the second, third,... harmonics from other sources. The balance I was talking about can, and does.

So again, unless you state what you're talking about, saying that "hum balance is only effective to a degree. Not to the degree that guitar player wants, or would prefer," doesn't mean anything. If you mean a humdinger, yeah, I agree.

The DC or cancellation method cleans up what the hum balance leaves behind. What the hum balance can't hope to do.

?? What hum balance? Which one?

Cancellation method has the advantage because there are no parts, or DC supplies and filters, being added to the circuit. That's the beauty of it.

If by this you mean diddling the wires to get a lower overall hum by reinjecting mains by wire placement, well, OK, that's cheap. But it's as marginally effective as a humdinger, and has the issues that someone could move a wire and permanently damage the balance. A more sophisticated canceller can cancel harmonics as well, which the mystical wire routing method can't. The lowest frequency in the B+ is two times mains frequency, for instance. No amount of mystical wire tinkering will get rid of that.

The cancellation method is only used by more experienced builders...

Only folks as incredibly good as you are, right?

I get what I want for free.

In general, or only as regards amp hum?

[R.G.]

Found it.

Audio Filters Theory and Practice

Clever device. Uses the mains and filters to generate 1x, 2x, 3x, 4x and so on times the mains frequency. Then a phase shift stage lets you null out phase differences and a variable level per harmonic lets you null out individual harmonics. I put two phase shift stages per harmonic in mine so I could skip the business about flipping the mains plug and do it in the electronics. It's mildly amazing to see this thing in operation. Even BAD hum can be nulled out almost completely. The problem is it can't follow changing hum amounts and phase, so it only works well on unchanging hum problems. But it can be used on the end of a multifaceted chain, like a PA, and null out hum from whatever.

[Enzo]

So it could null out ground loop hum in a system - a hum source that is stable - but be worthless for the hum coming from the guitar pickups - like them picking up the power tranny in certain positions.

[R.G.]

It's like all cancellation schemes - cancelling a static signal source is much simpler than cancelling a changing one. It is possible to cancel almost all the hum in an amp by plugging this widget into an unused input. That is how I used it. It is also possible to null out the amp plus any static source - like a ground loop to another amp - this way. If the guitar is putting out an almost unchanging hum, then it would cancel almost all of it. If the guitar is moved toward and away from a leaky transformer, that would be a problem.

This is not a substitute for low noise/low hum amps, but it's a great problem solver. It is better for amps to not let the hum get into the audio path in the first place if you can avoid it.

[trobbins]

Neat bit of kit for the test bench. It may be great for those guitar amplifiers that have been 'hi-fied', as it could be used to dial in some missing 'tone' that has been inadvertently supressed by overdesign of the power filtering.

[Lee_ranaldo]

Hello guys.I've been busy and not following your replies. Sorry about that

I Finally got a new power transformer witha separate winding for the DC heater circuit and the AC wires are fed with a 6.3V Center tapped
secondary @ 5A, a bit more than what 4 6L6GC's need for but makes the PT run cooler . Now the background hum is fairly normal.

The amp was oscillating in a few settings so i re-did the whole shielded wire arrangement : Hooking the shield straight to the chassis instead of the circuit ground.

The DC heaters are fed with a 7812 regulator that i stabilized so there is almost no ripple at the output.I took great care of this design to make sure
no wrong values on the circuit could cause ripple , even though there was a regulator included.

I'm feeding every 12AX7 heater with 12.6 Volts DC average and almost no ripple at all.

So now the only tiny issues are.

1) i get a high pitched tiny squeal when i turn the standby sometimes. I Changed the whole set of tubes and the issue remained constant. I am thinking of adding a snubber cap
across the standby switch terminals to solve this problem. I wonder if any of you had a similar issue.
2) I can take a 20 volts DC average reference for the power tube heaters that run on AC. I wonder if i'm gonna have issues
since the DC heater circuit (provided by a separate tap from the power transformer) could feed part of the AC from the heaters into the circuit and create issues.

This would be to elevate the heaters to a positive voltage reference and avoid hum being induced from the heaters to the power tube elements.

Everything else runs fine, The output section seems stable and runs fine

Any suggestions?

[Lee_ranaldo]

Hello guys.I've been busy and not following your replies. Sorry about that

I Finally got a new power transformer witha separate winding for the DC heater circuit and the AC wires are fed with a 6.3V Center tapped
secondary @ 5A, a bit more than what 4 6L6GC's need for but makes the PT run cooler . Now the background hum is fairly normal.

The amp was oscillating in a few settings so i re-did the whole shielded wire arrangement : Hooking the shield straight to the chassis instead of the circuit ground.

The DC heaters are fed with a 7812 regulator that i stabilized so there is almost no ripple at the output.I took great care of this design to make sure
no wrong values on the circuit could cause ripple , even though there was a regulator included.

I'm feeding every 12AX7 heater with 12.6 Volts DC average and almost no ripple at all.

So now the only tiny issues are.

1) i get a high pitched tiny squeal when i turn the standby sometimes. I Changed the whole set of tubes and the issue remained constant. I am thinking of adding a snubber cap
across the standby switch terminals to solve this problem. I wonder if any of you had a similar issue.

2) I could elevate the heaters using the DC voltage from the pre-amp heater circuit. This would be done before the regulator and before the first filtering cap (3 3300uf caps in parallel). The voltage is around 20 volts AVG or more. I wonder if i would have any issue or any signal fed into the circuito from the AC heater center tap and so on.


Any suggestions?

[trobbins]

I think you are saying that the 12VDC regulated supply comes from a completely isolated transformer winding, with no other connections to any circuitry, other than to the preamp valve heaters, and to a single 0V grounding point (preferably a local distributed star point related to the preamp valves).

It is worth noting that most of us would be amazed if elevating an isolated output stage heater winding would make any measurable difference at all so this discussion is likely to be just a 'thought experiment'.

An alternative is to take the 6.3V CT for the AC powered heaters directly to the 12VDC regulated supply. With this scenario, the most negative going part of the 6.3VAC signal is still going to be elevated by about +8VDC (assuming the 0VDC from the preamp heaters is grounded). And you are introducing less higher frequency ripple to the output stage heaters, due to the performance of the regulator, so only mains frequency hum would couple across output stage heater-cathode parasitic capacitance and resistance (which may go directly to ground anyway in a fixed bias arrangement).