AC heaters ultimate hum limit

I have been running some noise/hum tests on my  classic mic pre circuit. Despite achieving a measured EIN of around -120dBu at 70dB gain, when using AC heaters there is a significant 50Hz and 100Hz component of hum:

https://drive.google.com/file/d/0B_n67A1hN3qtaF9XTThRcWR5RlU/view?usp=sharing

When using DC heaters on the same PCB these components completely disappear:

https://drive.google.com/file/d/0B_n67A1hN3qtR2d3cEFoSXdSemM/view?usp=sharing

So the hum is clearly due to AC heaters. I have tweaked the heater centre tap and can reduce the hum by a further 5dB but no more. Removing heater elevation worsens the hum.  -120dBu is less than 1uV at the mic input and with 70dB of gain the hum  will be swamped by acoustic noise. But, since there is a 1:10 mic transformer, it implies there is 10uV hum at the grid of the first stage. Tubes like the EF86 claim to be able to keep hum down to 1.5uV at the grid but they take all sorts of special precautions in the build of the tube to achieve this none of which apply to the 6922 tubes used in the classic design.

The question is therefore, is this as far as I can go with AC heaters?

Cheers

Ian

If the AC heater is introducing a known repeatable hum signal, is it possible to cancel it out by introducing an opposite version of that signal into the input?

JR

PS: I am not a tube guy... in case it isn't obvious.  8)

JohnRoberts said:

If the AC heater is introducing a known repeatable hum signal, is it possible to cancel it out by introducing an opposite version of that signal into the input?

JR

PS: I am not a tube guy... in case it isn't obvious.  8)

Click to expand...

PPS: you mean an inverting opamp to phase cancel a tube amp hum  8)

PPPS:Now the question is: opamp with AC or DC rail  :-X

PPPPS:

best
Zam

time to switch to DC heaters  8)

how much hum at around 50-60 db gain?

zamproject said:

JohnRoberts said:

If the AC heater is introducing a known repeatable hum signal, is it possible to cancel it out by introducing an opposite version of that signal into the input?

JR

PS: I am not a tube guy... in case it isn't obvious.  8)

Click to expand...

PPS: you mean an inverting opamp to phase cancel a tube amp hum  8)

Click to expand...

no...  even tubes are differential devices , while I'll leave it to others to work out the details.

PPPS:Now the question is: opamp with AC or DC rail  :-X

Click to expand...

I didn't say anything about op amps. Since it's your idea, you can answer yourself.

PPPPS:

best
Zam

Click to expand...

It seems a DC heater may be easier, while a circuit pooge to cancel out the hum might be cheaper if the hum is stable. If the hum changes with other variables, then canceling it could be impractical

JR

Hi Ian
I really think the problem is EMF. The current in the heater leads creates a magnetic field. The field couples to the input circuit and the current noise is them amplified and heard as hum. If the heater leads are on the PCB disconnect them and use twisted leads ( try solid wire as this makes it easy to keep the wires in a specifies place) from the transformer spaced from the board. Make the leads go vertically away from the tube socket by several inches. Measure the circuit and move the leads closer to the PCB if the hum increases them you need to make a magnetic shield.

Another test is make a single insulated loop connected to the input connector. Now move this close to the ac heater leads and listen to the signal. Rotate the pickup loop 180 degrees and see if you can cancel the hum.
Good hunting.

Duke

JohnRoberts said:

If the hum changes with other variables, then canceling it could be impractical

JR

Click to expand...

ie: different location in planet earth : 50hz  60hz  ?

I pretty much refuse to use AC heaters anymore unless I build something point to point, which I'm not sure if I ever plan on doing.

Thank you for the numbers!

buildafriend said:

I pretty much refuse to use AC heaters anymore unless I build something point to point, which I'm not sure if I ever plan on doing.

Thank you for the numbers!

Click to expand...

would P2P be reducing AC heaters hum at 70db gain on same mic_pre ?

edit:

buildafriend said:

I pretty much refuse to use AC heaters

Click to expand...

1+
DC heaters are easy and cheap enough these days...

JohnRoberts said:

If the AC heater is introducing a known repeatable hum signal, is it possible to cancel it out by introducing an opposite version of that signal into the input?

JR

PS: I am not a tube guy... in case it isn't obvious.  8)

Click to expand...

In theory yes, in practice partially.  The hum dinger is the usual way to achieve this. However, the null point for each tube is slightly different so complete cancellation never occurs using this technique. It might be possible to make a per tube adjustment but I do not think it has been tried. Food for thought.

Cheers

Ian

Audio1Man said:

I really think the problem is EMF. The current in the heater leads creates a magnetic field. The field couples to the input circuit and the current noise is them amplified and heard as hum. If the heater leads are on the PCB disconnect them and use twisted leads ( try solid wire as this makes it easy to keep the wires in a specifies place) from the transformer spaced from the board. Make the leads go vertically away from the tube socket by several inches. ....

Click to expand...

I fully agree with you, excellent ideas. Also there is an idea of a screening method  used in RF: using metal plate (Fe-Sn)  mounted on the bottom of the PCB with holes for  heather and +B wires.

The hum cancellation in SE tube amps is well known procedure. I believe it can be  applicable here to some extent.

http://www.tubecad.com/april99/page2.html

Thank you everyone for the suggestions.

@kambo: I normally do use dc heaters but for the classic solo and the LunchBox projects I found that AC heaters gave the same noise measurement with my Lindos and you certainly cannot hear the hum. It was not until I started using the QA400 that I could actually see the size of the hum. The hum is largely proportional to gain. I suspect the hum at the output of each stage is about -77dBu and each stage has a gain of about 28dB so with the gain up full the second stage raises the -77dBu hum of the first stage up to about -50dBu. For 50dB gain the hum would be about 20dB lower - about -70dBu.

I agree for small projects dc heaters are fairly easy and fairly cheap but - you need to up rate the heater winding and heatsink the regulator. For larger projects it begins to become bulky and expensive. A fully loaded LunchBox needs about 3A at 12V DC which is a chunky heatsink. For a small tube mixer you probably need at least twice that so a linear regulator becomes a PITA so you start thinking about using SMPSUs but then you find they often cannot supply the inrush current and sit stubbornly in foldback current limit. I recently bought a Lambda 150W 12V SMPSU so I can do some testing with a SMPSU made by a reputable company. Ripple is of course a concern but the Lambda data sheet says I can hang as much as 10,000uF across the output - sorry drifting OT.

What would be interesting would be to get hld of a V72 or a REDD 47 clone and subject it to the same tests.

@Duke: I also am beginning to think the cause is primarily magnetic filed - of course it is hard to twist tracks on a PCB so I will try your idea of disconnecting tracks and replacing with twisted wire. The problem is made more difficult because the heaters of the two tubes are wired in series. I love the insulated loop idea - I'll try that too.

@moamps: Thanks for the SE cancellation link. I will check it out.

Cheers

Ian

ruffrecords said:

But, since there is a 1:10 mic transformer, it implies there is 10uV hum at the grid of the first stage. Tubes like the EF86 claim to be able to keep hum down to 1.5uV at the grid but they take all sorts of special precautions in the build of the tube to achieve this none of which apply to the 6922 tubes used in the classic design.

The question is therefore, is this as far as I can go with AC heaters?

Click to expand...

I think you've answered your own question.  Designers of valves like EF86 will have thought long and hard about this problem and the construction of the valve will have been optimised  for low hum.

I believe, EF86 heater is not just "twisted" but in fact in a "hum-bucking" configuration.

If you are not using a valve where attention has been paid to this,  DC heaters are probably the simplest solution overall.

You don't need DC on all the valves .. only on the sensitive inputs.  And twisted heater leads are essential for AC heaters.

And there's simple reliable circuits like in CJ's first post http://groupdiy.com/index.php?topic=57166.0

But if the LINDOS doesn't pick it up and you can't hear it, why bother?  8)

Don't forget FFT type instruments like the QA400 used for "noise spectrums" will exaggerate 'single' tone noise like hum.  Try different size FFTs on noise if you can on the QA400 to see what I mean.

ricardo said:

But if the LINDOS doesn't pick it up and you can't hear it, why bother?  8)

Click to expand...

Because I prefer to understand and it is an interesting challenge!

Don't forget FFT type instruments like the QA400 used for "noise spectrums" will exaggerate 'single' tone noise like hum.  Try different size FFTs on noise if you can on the QA400 to see what I mean.

Click to expand...

The QA400 has different widowing options; Hann, Hamm, Rect, Bartlett and Flat top - I do not recall these making any difference to the amplitude. Is that what you are referring to or are you suggesting using a smaller number of points?

Cheers

Ian

How clean is the 50Hz? I'm guessing it's it a rather corrupt waveform, which actually contains higher frequency components that are more easily coupled to the grid etc. Anything you could do to make it more of a pure sinewave would likely lower the noise floor.

Wiring the heaters in a different order or with different relative phases may also help to optimise, but obviously there is a limit to what you can do here, given the voltages and pinouts.

Guard traces around the heater pins?

It was not until I started using the QA400 that I could actually see the size of the hum.

Click to expand...

Welcome to the world of analyser-induced neurosis. When I just used the Lindos I had a good handle on what represented a 'performance window' for hum / noise. Since getting an FFT analyser I've found it's enabled the paper specs of my gear to get much better, but at the expense of reduced productivity. Has owning the FFT made projects receive a better reception from clients?

NB - having the FFT has made me re-evaluate my attitude towards SMPSs. Assuming you get a grip on filtering the HF hash, they can make life a lot easier in terms of keeping LF mains-related noise at bay. I wonder if there's something to be said for a  SMPS for heater duties, coupled to individual current sources? 

JohnRoberts said:

zamproject said:

JohnRoberts said:

If the AC heater is introducing a known repeatable hum signal, is it possible to cancel it out by introducing an opposite version of that signal into the input?

JR

PS: I am not a tube guy... in case it isn't obvious.  8)

Click to expand...

PPS: you mean an inverting opamp to phase cancel a tube amp hum  8)

Click to expand...

no...  even tubes are differential devices , while I'll leave it to others to work out the details.

PPPS:Now the question is: opamp with AC or DC rail  :-X

Click to expand...

I didn't say anything about op amps. Since it's your idea, you can answer yourself.

PPPPS:

best
Zam

Click to expand...

It seems a DC heater may be easier, while a circuit pooge to cancel out the hum might be cheaper if the hum is stable. If the hum changes with other variables, then canceling it could be impractical

JR

Click to expand...

JR, I'm just joking  ;D
I stop spamming this topic, unfortunately I'm neither a tube guy, I need to learn.
Best
Zam

ricardo said:

Don't forget FFT type instruments like the QA400 used for "noise spectrums" will exaggerate 'single' tone noise like hum.  Try different size FFTs on noise if you can on the QA400 to see what I mean.

Click to expand...

ruffrecords said:

The QA400 has different widowing options; Hann, Hamm, Rect, Bartlett and Flat top - I do not recall these making any difference to the amplitude. Is that what you are referring to or are you suggesting using a smaller number of points?

Click to expand...

Yes.  Try different number of points.

There's a number of ways to "calibrate" an FFT when measuring noise but you should see the relative change between broadband noise & 'monotonic' noise (eg hum/buzz) quite clearly.

In my 1970s/80s Calrec days, hum/buzz had to be "inaudible".  This required hum/buzz to be well below  the measured A/broadband/CCIR noise level.  But the final arbiter was always headphones and the gain right up.

ruffrecords said:

I prefer to understand and it is an interesting challenge!

Click to expand...

1+
you rock Ian