DC heater elevation question

[Lee_ranaldo]

Hello forum

I've encountered this issue for quite a long time and i've debugged this whole amp to a certain point
the only issue i have is heater induced hum. Hum does not increase when i increase the signal that goes into the
power stage. It remains constant and reduces the dynamic range quite considerably.

I tried balancing the windings of the CT transformer. This Transformer delivers 12VAC RMS that goes into a Full Bridge rectifier
that feeds DC into the PHASE SPLITTER and PRE AMP TUBES.

I'm evaluating elevating the heaters with some DC to reduce the background HUM. This amp has 6 12AX7 Tubes that run DC on their heaters
and 4 6L6GC tubes.

This is a partial snapshot of the actual design:



This is what i have in mind



I wonder if anyone tried elevating heaters to reduce the HUM produced on the output stage. The Pre-amp stage is very quiet.
I tried balancing the windings to reduce the hum (Aka Humdinger) with no positive results. Considering the array how the heaters are connected is unusual and
might contribute to the HUM i'm giving the DC elevated heaters a try. I wonder if this would work or the peavey approach (5150 heater circuit) would work best feeding
the power amp stage instead.

The capacitors will just block the DC feed into the windings before it enters the full wave bridge rectifier.
I don't know how should i calculate its capacity accordingo to the amount of current they have to pass through their elements.

What do you think about this? Have you encountered a similar issue in your design?

Thanks in advance.

[Mick Bailey]

Have you tried this?



You also may need to reverse the heater connection for one tube of each pair to get the best hum-cancellation.

[soundguruman]

Hello forum

I've encountered this issue for quite a long time and i've debugged this whole amp to a certain point
the only issue i have is heater induced hum. Hum does not increase when i increase the signal that goes into the
power stage. It remains constant and reduces the dynamic range quite considerably.

I tried balancing the windings of the CT transformer. This Transformer delivers 12VAC RMS that goes into a Full Bridge rectifier
that feeds DC into the PHASE SPLITTER and PRE AMP TUBES.

I'm evaluating elevating the heaters with some DC to reduce the background HUM. This amp has 6 12AX7 Tubes that run DC on their heaters
and 4 6L6GC tubes.

This is a partial snapshot of the actual design:

[ATTACH=CONFIG]28705[/ATTACH]

This is what i have in mind

[ATTACH=CONFIG]28706[/ATTACH]

I wonder if anyone tried elevating heaters to reduce the HUM produced on the output stage. The Pre-amp stage is very quiet.
I tried balancing the windings to reduce the hum (Aka Humdinger) with no positive results. Considering the array how the heaters are connected is unusual and
might contribute to the HUM i'm giving the DC elevated heaters a try. I wonder if this would work or the peavey approach (5150 heater circuit) would work best feeding
the power amp stage instead.

The capacitors will just block the DC feed into the windings before it enters the full wave bridge rectifier.
I don't know how should i calculate its capacity accordingo to the amount of current they have to pass through their elements.

What do you think about this? Have you encountered a similar issue in your design?

Thanks in advance.

Before you go on with this:

The filament noise does not enter at the phase splitter.
It's in the first 2-3 stages of the preamp. This is where a small amount of AC enters the audio path, and gets amplified.

1. Because of layout. Keep the filament wires separated from the audio path.
OR design a layout for the wires that cancels the hum.

2. Mick Baily is right on.
Reversing the filament connections to a preamp tube can result in cancellation of the HUM.
Routing the filament wires, and the grid wires, in a certain path - can cancel HUM. 100% true.

Using this method, it is every bit as effective as DC heaters, only more complicated.
The wires must be positioned "just right." The position of the wires MUST be permanent and fixed.

WHEN using DC for the filaments:
It's the first 2 preamp tubes, not the phase inverter.
BUT you are on the right track.

Even the SMALLEST amount of AC that enters the first preamp stages...
WILL be amplified.

AND this is where the filament BUZZ comes from, in almost any tube amp.
It's almost never the output stage.

HUM in output stage?
The output tubes are out of balance.
The bias supply has too much ripple.
The B+ supply has too much ripple.

[markr14850]

While your DC is probably reasonably ripple free with respect to itself, it may still have an AC component in relation to your circuit ground. You might consider having only the DC side of your circuit referenced to ground - and to do that directly, rather than through balance resistors.

[Tom Phillips]

...The capacitors will just block the DC feed into the windings before it enters the full wave bridge rectifier.
I don't know how should i calculate its capacity according to the amount of current they have to pass through their elements.
What do you think about this? ...

I think the idea is not practical. The capacitance value would need to be unusually large to pass the current required.

Are you absolutely sure that the hum you are hearing is 60 Hz and not 120 Hz?

[soundguruman]

It usually works out quieter with the DC floating.
But I am not sure why...It's just testing that reveals this.

But if you reference DC to ground, and the DC isn't regulated,
then reversing DC and ground on the tube "may" also give you quieter amplification. Depends on which way you have it polarized.

There is something else important too:

As far as AC heater buzz / layout is concerned, 12AX7 has a noisy side and a quiet side.
Reversing usage of 123 and 678 can reveal quieter operation, as far as filament buzz is concerned. And this can be a pretty big difference, depending on layouts.
The first stage of amplification better be the quietest side of the tube.

But with DC heater, it depends more on "how" little ripple there is in the power supply. How "battery like" it is?
Can you get all the ripple out? (I suppose that's easier when the DC is regulated down from a higher voltage, hence the advantage of regulated filaments)...

I see it as being that the filament is almost a short circuit....and that condition is harder to maintain complete ripple rejection in an unregulated supply, especially.

1. So you are either going to use massive filter capacitor(s) in an unregulated supply...(boogie method) (McIntosh Method)

2. Or you are going to use a regulated supply, with an input voltage high enough....to reject all the ripple...

or you are going to compromise and put up with some noise...
or you are going to use AC buzz cancellation methods, arranging / positioning wires and components to cancel noise.

Then the 123 678 difference in noise won't be so cumbersome a restriction.

So a good approach is
6 volt regulated filaments, from 12VAC filament transformer.
Then you concievably have enough ripple free current, enough ripple rejection, for the first two preamp tubes.

[Lee_ranaldo]

Thanks everyone for your replies!

The pre-amp modules were designed on PCB. This is not my design but i got this amp for quite a while and i've been pulling my hair testing and trying every way possible to get rid of the hum / Buzz.

First of all i traced the signal with an oscilloscope and got rid of most of the buzz by using proper shielded wire and fixing a few issues on the pre amp section. Second the filter caps were replaced cause
i saw the B+ had extra ripple at the phase inverter and later stages. So since its on PCB there is little modificacion i can do to the wiring. The pre-amp tubie sockets are directly mounted into the PCB so all i can control in terms of signal routing is the heater supply, ground, b+ wires and signal wires that are already shielded properly.

This amplifier has a tube effects loop and the return is acting as a master volume control for both channels. The amount of buzz that i get on high gain stages AKA dirty channel of the amp is more than acceptable but the constant hum on the background ruins everything. I have a Line filter installed to avoid noise that could be feed into the amp by the mains power.

I have considered reversing the wires on the AC heater supplies. Also the bias circuit arrangement is similar to the one found on JCM800. the cap values are around 10uf each so that might contribute to the
background hum i hear. I did try running 2 power tubes in series to be able to supply 12VAC to each pair and i had no positive results, in fact i think the hum was worse.

The preamp and phase splitter are grounded with a single connection point straight to the chassis,

The hum dissapears when i remove the phase splitter tube though.

I'll run a few tests and let you guys know how it goes.

[Lee_ranaldo]

I've been running some tests

1) reversing heater wires on one of the power tubes seemed to reduce the hum , (not by much) but still noticeable
2) removed the balancing resistors to reference the ground from the DC heater supply and connected the negative side to ground. Then i saw the bias voltage was being
modulated by the DC heaters and vice versa (they share the common ground on the PCB. I soldered a wire from the negative side of the caps from the DC heater
supply straight to the chassis and this issue stopped
3) i still see ripple on the DC heaters but not as much
4) turning the amp and the hum remains constant and now i can hear the pop generated by the relays when i change channels. The relays are connected to the DC heater supply.

I could assume feeding the phase splitter with AC instead of DC would get rid of induced hum

[trobbins]

Many people on many forums have highlighted hum problems trying to introduce a DC heater supply using rectified heater windings that also power other heaters. Not only are some heaters moved very far away from symmetry around 0V and with added switching rectified waveforms on them, but rectifier hash is introduced and splattered around. Technically it is far better to revert to AC powered heaters for all valves (to removed the local rectifier hash issue, although HT rectifier hash can still ingress if not managed well) and use a pot humdinger to tune residual hum levels on the first preamp stage, and the general techniques for keeping high level signal wiring from sensitive areas (applies to not only heater wiring).

The heater-cathode DC resistance is affected by using DC elevation - of the tubes I've measured, that resistance can vary widely among samples of old 12AX7, so it may be worth tube swapping if there is still some noticeable hum via heaters (which can only be really evaluated by doing a simple but clean comparison with heaters that are temporarily powered from batteries, and assumes you really have got other ingress issues like wiring well managed).

If DC heaters must be used, then imho that requires a separate power supply transformer, or at least a separate winding for generating the heater DC voltage - and due care in managing rectifier noise.

[Mick Bailey]

Just to make sure the grounding of the power tube heaters isn't causing the hum; remove all four power tubes and check for hum on the 'send' socket.

[Lee_ranaldo]

Just to make sure the grounding of the power tube heaters isn't causing the hum; remove all four power tubes and check for hum on the 'send' socket.

I just verified that. The grounding system i used before (using a 12VAC / CT through balancing resistors) did gave me a few issues.
Without load voltages were fine. With loads the heater voltage would go up to 10 volts AC instead of 5.6 VAC (that i've measured without load)

So i used this arrange.

i am powering 5 12AX7 Tubes with a regulated 12,6 VDC (measures 12,3 VDC under load).



DC heater supply has way less ripple than before

Without regulation = 0.28 VDC
With regulation =0.08 VDC

I attached the 78X12 Regulator straight to the chassis (aluminum chassis btw) totally isolated through insulation
so there is no electric connection between the heatsink and the regulator itself.

I plan to reverse the DC supply on a few stages in order to reduce hum induced through out of phase cancellation as i've seen sugested here.

I've also seen peavey's approach of adding 2 100Ohm balancing resistors on the heater circuit but they add a 100ufx25 volt Cap from the CT to ground

I did try this and according to the oscilloscope it seems to improve the offset and reference the CT to "true 0 volts". The waveforms of the Heater supply
on the scope looked simetric which would improve heater induced hum reduction through out of phase cancellation. Although when i powered
it up with tubes in it the cap seemed to "warm up" a bit so i didn't test it further.

The last test would be elevating the heaters as they are wired on the image i've attached and see how many volts will actually provide
good hum reduction

Thanks everyone for their replies and suggestions and , i know, its another HuM/DC question but i hope this turns out helpful for someone
that encountered the same issue as i did.

[R.G.]

Since you have an oscilloscope, use it to get a picture of the hum on the output. Is the hum primarily AC line frequency (50 or 60Hz, depending on your country/location) or twice that? If the hum is 2x mains frequency, it's coming from the power supply operation. If it's 1x mains frequency, it's probably grounding and/or shielding issues, or possibly (1) a nonfunctioning half rectifier in the DC/B+ circuit making the supply be half wave rectified or a high-ripple bias supply as noted before.

One quick and dirty test is to get the amp warmed up, listen carefully to the hum, and flip the power switch off. This stops the rectifiers entirely for lack of power in, and the amp then coasts down as the tubes eat up the remaining B+ power before the filaments cool. You only get a second or so of this. But if the hum disappears when the power switch is flipped off, it's coming from the AC power into the amp. If it does not, then it's caused by reception of radiated hum from other places.

Note carefully that the wiring of transformer, rectifiers, and first filter cap is CRITICAL in terms of hum. Ordinary silicon rectifiers can generate a squark of RF every time one turns off, and this can be radiated to other places in the amp. The only way to get rid of this is the (1) wire the rectifiers to the first filter cap correctly, negative wire directly to the cap minus and/or (2) snub the rectifiers to kill RF emission or substitute fast/soft turn off diodes.

There are some other low-probability things it might be. The disappearance of the hum when the PI is pulled and non-participation of the preamp is a clue, just not a very clear one at present.

[Lee_ranaldo]

One quick and dirty test is to get the amp warmed up, listen carefully to the hum, and flip the power switch off. This stops the rectifiers entirely for lack of power in, and the amp then coasts down as the tubes eat up the remaining B+ power before the filaments cool. You only get a second or so of this. But if the hum disappears when the power switch is flipped off, it's coming from the AC power into the amp. If it does not, then it's caused by reception of radiated hum from other places.

Note carefully that the wiring of transformer, rectifiers, and first filter cap is CRITICAL in terms of hum. Ordinary silicon rectifiers can generate a squark of RF every time one turns off, and this can be radiated to other places in the amp. The only way to get rid of this is the (1) wire the rectifiers to the first filter cap correctly, negative wire directly to the cap minus and/or (2) snub the rectifiers to kill RF emission or substitute fast/soft turn off diodes.

There are some other low-probability things it might be. The disappearance of the hum when the PI is pulled and non-participation of the preamp is a clue, just not a very clear one at present.

Thanks exactly what i encountered checking the DC heaters. Their negative connection to the first cap was done on the PCB : The ground is being shared by the energy reservoir (2 caps in series with bleeder resistors) and the bias circuit. The signal @ the output was totally erratic and it seemed that the circuit had poor regulation. I Ran a wire straight to the ground point in the ciassis and it cured the problem .

Also i usually shut the amp off by its power switch and let the caps drain. The hum goes away as soon as i shut it down. It doesn't sound like the buzz you hear on a poorly designed amp. It sounds like 50hz (our main's frecuency over here) and not 120hz. Also i installed caps accross each diode on the HT rectifier section (100nf * 1KV ceramic disc cap) to reduce the switching noise a bit. I do have some UF4007 laying around that i might try if i notice that i could improve the S/N ratio even more.

One of my concerns now is this: By elevating the Heaters and use a DC voltage as a reference (Instead of ground) i might add a DC component that will be passed through the rectifier on the heater section. I don't know HOW MUCH this would raise the voltage received by the regulator.This is , of course, an assumption that i might test.

[Enzo]

What RG was asking though was whether the hum was REALLY 50Hz instead of 100Hz. 120Hz in the USA is simply twice 60Hz, and will be the result of power supply ripple. In your case 100Hz would come from ripple. Note that 50Hz and 100Hz are the exact same "note" musically, just one octave apart, and it can be difficult to tell them apart. Due to overtones, I always check with my scope.

[soundguruman]

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.
It is not uncommon to use 10,000 - 20,000 uF caps for your supply filters. Boogie is using 24,000 in a lot of designs. (3X 6800 in parallel)
McIntosh was using 20,000 for preamps in the 1960s, just for 2X 12AX7.
So, 4700 (here) is under-designed. I wound go at least 10,000.

Important to remember, the filament noise is not coming from ALL the preamp tubes. Just the first stages. Not the phase inverter.
So usually the first 2-3 tubes with DC filament is all it takes. The rest of the preamp tubes usually can run AC filaments.
This can save current, and reduce the amount of filtering required. The smaller the load, 3 tubes instead of 5 tubes, the less ripple you get.

And now, an example of tweaky AC filament noise cancellation methods:




Follow the blue wire here, it's the input to grid pin 7, on this Marshall V2 cathode follower stage.
The wire is 600 volt insulated Teflon, and it has another 600 volt tubing over it. (1200 volt effective insulation)
The audio signal from the circuit board enters between pins 9 and 8...
goes UNDER the plate resistor, and the jumper, then between pins 4&5, then does a reverse turn, over to the grid, pin 7.

When the grid wire travels under the plate resistor, this creates capacitance between the plate and grid, reducing HISSSSS. Yes, it works.

This deliberately induces filament AC onto the grid wire (from pins 4&5)...and balances it by making the grid wire parallel to the opposite filament wire (pin 9).
Producing a cancellation of AC on the pin 7 grid.
The grid wire is held in place at the exact correct distance (to produce cancellation) by glue and wire ties.
Once you have positioned the wire to cancel the noise, it needs to be secured permanently. It can't be allowed to move...

In the original Plexi design, V2 is producing about 2 V PP filament noise at the amplifier output (with volume pegged)...a LOUD buzz.
AFTER re-routing the wires...
The filament noise is reduced to about 20 mv PP (or less) at the amp output. A HUGE improvement. All by just moving the wiring.

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