PDA

View Full Version : Have you got a humming valve amp?



Simon Barden
27-04-2017, 12:31 AM
If you've got a humming valve amp, especially if it's a commercial model with PCBs , then (if it's not an open ended single power valve class A amp which doesn't have one) it's quite likely that it's down to the phase-inverter/splitter valve. (If it's a hand-wired turret-board boutique-style amp it way well be the case as well, but with a lot more cables running around inside this sort of amp that may be badly dressed and running near noisy sources, there are multiple possible hum sources that may be the cause, so what I'm discussing here may only help cure some, if not all the hum on this type of amp).

This PI (Phase Inverter) valve will typically be an ECC83/12AX7 or ECC81/12AT7. Both of these are dual triode valves - that is, two independent triodes in the same glass package.

Unless you've got a small wattage class A or boutique class A amp with multiple output tubes then it's very likely that you've got two, four or six power amp valves operating in a push/pull Class AB mode (apologies to the technical wizards if I'm generalising too much here). They work in pairs, with one half of each pair being the 'push' side of the amp whilst the other valve is acting as the 'pull' mode (as we are amplifying an AC signal, the push and pull operations swap round as the AC signal changes from a positive to negative or negative to positive, part of its cycle).

One side of the PI valve feeds half of the output valve(s), the other side of the valve feeds the other half of the output valves(s), but in one half the signal is inverted compared to the other half so that the two sets of power amp valves amplify the difference between the two signals.

Now others may be able to tell you exactly what is going on here (and I may have got it slightly wrong), but all I know is that if the two halves of the PI valve don't have exactly the same gain, then the signal sent to each set of power tubes is a bit unbalanced. With a fully balanced signal, any 'common mode' mains noise picked up by both signals in the power stage is cancelled out (just like a balanced input on an audio interface). With slightly different signal levels, the mains noise isn't fully cancelled, and the more the two halves' gain differs, the more hum you get.

On some amps, the gain of one of the triodes in the PI can be adjusted by using a variable resistor (I believe in its bias circuit), So if your amp has a 'hum balance' pot, then use it to get a hum-free clean sound (If you can't get it perfectly hum-free, then its due to noise pickup elsewhere within the amp, but at least you can minimise it). Note well that the 'hum-balance' pot is not the same as any 'bias adjust' pot you may have on your amp, as the 'bias adjust' pot sets the bias for the power amp valves only, and should only be adjusted if you know what you are doing (otherwise leave well alone or take it to an amp specialist to be set correctly if you ever fit new power valves).

But if you haven't got a hum balance pot (and a lot of amps don't have), and your amp hums quite badly, then it's worth purchasing a 'balanced' valve for the PI valve. This is just a standard valve that's been tested and both sides found to have nearly exactly the same gain and drive characteristics. It normally costs a few $ extra for a balanced over a standard valve, but it's worth it IMO. There's no benefit to using a balanced valve in other positions; the slight variations in gain won't be noticeable there. It can only cure hum picked up in the output stage of the amp; but often this can be reduced to inaudible levels with clean amp settings by using a balanced PI valve. Apparently the valve won't remain balanced with the power amp starting to distort (and beyond), but by that stage you are very loud and the background hum will be hidden by all the musical noise the amp is making.

A balanced valve won't reduce hiss (certainly not by any noticeable amount), which is mainly caused by the valves themselves. Just the amp's own hum. And it can't remove any hum picked up by a guitar.

Double check with your amp manual as to which valve the phase inverter is, but if lined up in a row in the chassis, it is almost always the one right next to the power tubes. Even without a new valve, it's worth swapping round any valves you've got of the same type to the phase inverter position to see if any of them decrease or increase the amount of hum you get. You may find that one you already have gets rid of the hum (almost) completely. But if you do notice a difference in hum levels, then next time you replace some pre-amp valves (or immediately if it's a bad hum), it's well worth making sure that you get a balanced valve for the PI position.

I've now done it to two of my amps, and it's made both of them hum free. You can get used to a slight amount of hum in an amp, but once it's gone, you'll wonder how you ever coped with it!

When searching for info on balanced valves on the web, I've seen posts that say you don't need a balanced valve for the PI position. Now it may not improve the amp's sound and if you've got a hum balance pot then it's certainly of no benefit; but without a balanced PI valve, in amps without a hum balance feature, then you could be getting a lot more hum than you need to - especially if you want a nice clean sound for recording. I know this through my own empirical testing.

If you ever order a complete set of new valves for your amp from a good valve supplier, you should find that one of the pre-amp valve boxes has a 'balanced' or 'PI tested' indication on it. If so, use it in the Phase Inverter position for best results.

As it's been tested, then it may also have a 'V1' indication (or a tick against common V1 printing on all the valve boxes). In addition to any supplied balanced valve, there should be at least one other valve with just a 'V1' on it. This valve will have been tested for low noise characteristics, and is suitable for use as the first pre-amp valve (hence the V1), which is almost always configured for the most gain and therefore the valve that will produce and amplify the most noise, so a valve with low self-noise is best used in this position.

Hope this is of some use to you. It's not something I've seen mentioned in articles before and I've only just discovered the joys (and importance) of balanced PI valves, so hopefully this will benefit some of you. It now seems pretty obvious to me, but you do need to know something about how valve amps work before you'd work it out without being told.

DrNomis_44
27-04-2017, 05:46 PM
There's another way to balance the PI valve, basically it involves tweaking the values of either, or both of the plate-load resistors of the PI valve, typically in most guitar amps that have a PI valve, such as a Fender Twin, or a Marshall JTM45, or Plexi, the two plate-load resistors are 100k types, I've seen some guitar amp schematics where one of the plate-load resistors is an 82k (some Mesa/Boogie amp circuits for example), this is usually done so that the output valves clip more evenly when they are overdriven, and I have also seen schematics where a small trimpot is placed in series with the 82k plate-load resistor so that it enables the waveform-symmetry to be tweaked to a certain extent.

Simon Barden
03-05-2017, 02:12 AM
Yes, that trimpot is the same thing as the hum-balance pot, especially if it's brought to the back of the amp so it can be adjusted easily. It seems such an easy thing to provide that I'm surprised that it's not a lot more common. It cost almost nothing to provide if designed in from day one, and saves having to order a specially tested valve each time you replaced the PI. Especially when 99.9% of the manufacturers don't bother to fit tested ones as standard in the factory.

Simon Barden
04-05-2017, 06:55 PM
Well, it seems I was just about 100% wrong on that. I've just been put right on this on the SoundOnSound forum by my mate Dave, who's an ex-Blackstar amp designer.

What causes the hum in the PI valve position is leakage between the heater and cathode. Now that fact that I've bought 'balanced' valves for use in this spot and the hum has gone, is nothing to do with the 'balanced' bit. As Dave pointed out to me, it's unlikely that the gain control resistors for the two halves of the valve are any better than ±5% tolerance, and most likely ±10%, so that matching the triode gains to 99.9% isn't going to make any difference.

So it's most likely the fact that in the testing and selection process, any valves with high heater/cathode leakage are being weeded out, so that only those with little or no heater/cathode leakage make it through. Now, I have no idea whether the same criteria are being used for other valves that aren't being 'balance' tested, and that probably depends entirely on the mindset of the valve suppliers. They may have to reject a lot of their valve stock if they were too rigorous! But I'd still recommend paying a bit extra for a PI valve that has been tested for use in the PI slot.

And hum balance pots (if fitted) go in the heater line and are nothing to do with the phase inverter circuit.

DrNomis_44
10-05-2017, 06:01 AM
Well, it seems I was just about 100% wrong on that. I've just been put right on this on the SoundOnSound forum by my mate Dave, who's an ex-Blackstar amp designer.

What causes the hum in the PI valve position is leakage between the heater and cathode. Now that fact that I've bought 'balanced' valves for use in this spot and the hum has gone, is nothing to do with the 'balanced' bit. As Dave pointed out to me, it's unlikely that the gain control resistors for the two halves of the valve are any better than ±5% tolerance once, and most likely ±10%, so that matching the triode gains to 99.9% isn't going to make any difference.

So it's most likely the fact that in the testing and selection process, any valves with high heater/cathode leakage are being weeded out, so that only those with little or no heater/cathode leakage make it through. Now, I have no idea whether the same criteria are being used for other valves that aren't being 'balance' tested, and that probably depends entirely on the mindset of the valve suppliers. They may have to reject a lot of their valve stock if they were too rigorous! But I'd still recommend paying a bit extra for a PI valve that has been tested for use in the PI slot.

And hum balance pots (if fitted) go in the heater line and are nothing to do with the phase inverter circuit.



I seem to remember reading in an old Valve data book that the 12AX7/ECC83 spec for heater/cathode maximum voltage is something like 100V DC max before the heater insulation punctures and leaks, I may be wrong though.

FrankenWashie
10-05-2017, 01:13 PM
Well, it seems I was just about 100% wrong on that. I've just been put right on this on the SoundOnSound forum by my mate Dave, who's an ex-Blackstar amp designer.

What causes the hum in the PI valve position is leakage between the heater and cathode. Now that fact that I've bought 'balanced' valves for use in this spot and the hum has gone, is nothing to do with the 'balanced' bit. As Dave pointed out to me, it's unlikely that the gain control resistors for the two halves of the valve are any better than ±5% tolerance once, and most likely ±10%, so that matching the triode gains to 99.9% isn't going to make any difference.

So it's most likely the fact that in the testing and selection process, any valves with high heater/cathode leakage are being weeded out, so that only those with little or no heater/cathode leakage make it through. Now, I have no idea whether the same criteria are being used for other valves that aren't being 'balance' tested, and that probably depends entirely on the mindset of the valve suppliers. They may have to reject a lot of their valve stock if they were too rigorous! But I'd still recommend paying a bit extra for a PI valve that has been tested for use in the PI slot.

And hum balance pots (if fitted) go in the heater line and are nothing to do with the phase inverter circuit.

Well good on you for following through and updating mate. This sort of stuff is fascinating in passing as 70% is beyond my ken so to speak. I do like to try and understand it all though and occasionally (very occasionally) something relevant sticks.

Simon Barden
10-05-2017, 04:32 PM
FW, I probably know a bit more about electronics than you, but a lot of this stuff confuses me just as much as it does you, especially when it's being described without the aid of a diagram. So I do my best, propose something based on my empirical testing and then wait for others to expose the errors in my argument. Sometimes being wrong is the only way to learn something. I often find that if you write something incorrect, people will pile in to correct you, whereas if you just ask a question about the same thing, you rarely get more than a few one- or two-line answers.

Doc, probably, though I'd imagine that's for complete catastrophic breakdown (as probably what happened in the noisy valve that started off this thread), which may also be affected by other events (e.g. the two times I ran the amp for a few seconds without a speaker connected). Plus although one manufacturer's 12AX7 should be the same as any other manufacturer's 12AX7 to meet the 12AX7 specification, we know that they can be very different in sound and behaviour. And also there must be significant large manufacturing tolerances, otherwise there would be no need for suppliers to test the valves before making sure they fit into an acceptable category, and no need to specifically match output valve pairs if they were all made exactly the same.

FrankenWashie
10-05-2017, 04:46 PM
FW, I probably know a bit more about electronics than you, but a lot of this stuff confuses me just as much as it does you, especially when it's being described without the aid of a diagram. So I do my best, propose something based on my empirical testing and then wait for others to expose the errors in my argument. Sometimes being wrong is the only way to learn something. I often find that if you write something incorrect, people will pile in to correct you, whereas if you just ask a question about the same thing, you rarely get more than a few one- or two-line answers.


True statements all.
Electrickery is one of my least understood facets of what we all do here. Always happy to learn more and have knowledge expanded. Even happier if I can do it without getting electrocuted in the process. :)

stan
10-05-2017, 07:07 PM
Really insteresting Simon, thanks

DrNomis_44
10-05-2017, 07:30 PM
FW, I probably know a bit more about electronics than you, but a lot of this stuff confuses me just as much as it does you, especially when it's being described without the aid of a diagram. So I do my best, propose something based on my empirical testing and then wait for others to expose the errors in my argument. Sometimes being wrong is the only way to learn something. I often find that if you write something incorrect, people will pile in to correct you, whereas if you just ask a question about the same thing, you rarely get more than a few one- or two-line answers.




Doc, probably, though I'd imagine that's for complete catastrophic breakdown (as probably what happened in the noisy valve that started off this thread), which may also be affected by other events (e.g. the two times I ran the amp for a few seconds without a speaker connected). Plus although one manufacturer's 12AX7 should be the same as any other manufacturer's 12AX7 to meet the 12AX7 specification, we know that they can be very different in sound and behaviour. And also there must be significant large manufacturing tolerances, otherwise there would be no need for suppliers to test the valves before making sure they fit into an acceptable category, and no need to specifically match output valve pairs if they were all made exactly the same.

That's very true, most electronic devices and components are made to a certain tolerance, simply because it's impossible to make them exact, in the real-world you don't really need to have components measuring exactly as per their marked values, since most electronic circuits aren't that critical unless they are for specialized applications.


I'm following this thread with a lot of interest.


Servicing a piece of electronic equipment is a lot like trying to solve a mystery, you can regard fault-symptoms as clues as to what may be causing the fault, and by using deductive-reasoning, you can gradually eliminate all the possible causes until you're left with whatever happens to be causing the fault.