[Niteshade]

Hum can needle  the best of us. No matter how pure your DC supply is, how good your grounding scheme may be- Hmmmmm can still exist.

The typical ways of eliminating filament hum:

1. 100 Ohm resistors connecting both side of the filament to ground, or a hum balance pot.
2. Putting the filament voltage center tap to ground (if it has one)
3. Twisting the filament wires
4. Using filtered DC

One of the reasons filament hum exists is because the filament will be closer to ground than the cathode, thus the EMI migrates to the cathode.

Possible Answer: Making the filament a higher potential than ground! I haven't seen this done before, but it certainly makes sense! Apply around 40 volts to one side of the filament supply windings, tie the other side to ground with a high value resistor. The credit for this idea doesn't go to me, I found it on another BBS and thought it was ingenious. Can't wait to try it! Has anybody tried it?

FYI: This is great for indirectly heated cathodes. It will not work for direct heated cathode tubes. DC is your best way forward there.

[JohnR]

I've seen it somewhere, but don't remember where. I'm really not clear on how it would reduce hum though...

[Niteshade]

Think of it as biasing the filament. You're raising the filament's reference voltage above ground. Granted, biasing typically brings a reference below ground level, but the opposite can be done to prevent voltage flow.

I've seen it somewhere, but don't remember where. I'm really not clear on how it would reduce hum though...

[JohnR]

I realize that, but how does that reduce hum? It's indirectly heated, right, so...

[Niteshade]

You're right. Even though it's indirectly heated, the filament can be considered in this noise issue/model as another separate part of the tube. The cathode can draw energy (EMI) from the filament and due to their proximity to each other, it unfortunately happens very often. Cathodes are often above ground potential, making them filament EMI magnets.

[JoshK]

John Broskie suggests to typically raise heaters to 1/4 B+, Pretty easy to do with a voltage divider in your B+ and reference heater ground to this with a cap to help hold the voltage steady.

It supposedly does help reduce hum.  It reduces the potential between the cathode and heaters causing less draw for leakage currents through the insulator (not perfect after all). 

I've seen it reference as a common trick on diyaudio tube forum, probably the result of Broskie suggesting it.

[JoshK]

P.S. I think Gordy said it helped reduce hum on his Aikido.

[Niteshade]

I'll let you know how it works. For some reason the paraphase phase splitter that I use sometime is very susceptible to filament noise. Only certain tubes have filament noise issues. The reissue TungSol's are horrible! My new Chinese 6SN7's are dead quiet. But- this issue has cropped up in the most unfavorable places, like preamps and I have been looking for an easy blanket fix.

[JoshK]

Another idea to try is adding a common mode choke near the heater pins.  If the noise on the heater wiring is common mode, it should cancel quite a bit.  If the noise is differential, then mount a 0.1uf or appropriate value close to but before the CMC, to turn the differential to common mode. 

CMCs aren't the cheapest solution, but might work if nothing else does.  JW Miller and Coilcraft make ones that we use for the Felix.  You might be able to find cheaper ones for lower voltage, since not as many watts will be pushed through the windings.  I would think there would be plenty of available solutions with SMPSs so prevalent.

John at the clarisonus blog has explored the idea some there.

[JoshK]

In case you haven't seen it...

http://www.clarisonus.com/blog/?p=169

and the measurements...

Note: my memory was slightly off.  I thought that the article also mentioned CMC chokes, it doesn't mention CMC chokes, but goes into choke input supplies in heaters, which could be also useful.  However, I have read reports with CMCs from Thoersten Loesch and others on tube diy forums.

[Niteshade]

Thanks for the info Josh. Filament noise has at times been a real issue. Some news to report: I tried elevating the filaments to around +100v and it's NOT a good idea! One of my 6L6's has a conductive path between the filament and cathode and caused one of my fail safe resistors to fry. It was measured at around 2,000ohms. Other than this, the 6L6 works great.

A good link: http://www.google.com/url?sa=t&source=web&ct=res&cd=25&url=http%3A%2F%2Fvacuumtubedesigns.com%2FVTD%2Fdatapdf%2FHumHissNoise.pdf&ei=XmL8SK_RGZeoeMeD-JoH&usg=AFQjCNG8d8xwNtyzs0Vb2qooWSpUBCI6kQ&sig2=-DxudN8q0byKatSlg47SPQ

[ghpicard]

I use it in my SET45 driver stage and it works... it's usually suggested for low signal level stages where the also low signal from the parasitic rectifier effect and the not-so-low cathode impedance gets combined and thus a common cathode stage regarding to the input signal also behaves as a grounded grid one for the noise (coming through that parasitic diode). Driver stages are not really so sensitive, depending of course of the tube used, but for some phono stages, it becomes almost essential if we don't want to resort to DC heater power.

Another use for biasing the filaments to a higher than ground potential is to prevent filament-cathode rupture when the cathode-filament maximum allowed voltage is lower than the cathode potential (i.e. some OTLs, some cascodes).

Using this technique for high signal level stages (such as a power output) is IMO not useful at all, unless used as told above to avoid cathode-filament rupture (arcing).

However, not recommending a valid technique due to usage of a faulty tube does not sound (pun intended) right.

[Niteshade]

"However, not recommending a valid technique due to usage of a faulty tube does not sound (pun intended) right."

I hear you- but the tube actually works 100% and is noiseless. I don't know what insulates the cathode from the filament with indirectly heated tubes- but there must be some impurity in it/defect with the tube I was using. I as said earlier- the tube works perfectly otherwise. My concern is not so much with the filament biasing method as with how filaments are insulated from cathodes. If there are quality issues, then this approach might not be advisable since the filament bias supply can be easily shorted out. My concern can be treated as more of a question since this method of treating filament noise is uncharted territory for me. I only tried it once.

What is your approach to biasing the filaments? How much voltage did you use?

[Niteshade]

There are two kinds of noise I typically have to deal with: Hum and/or a high frequency buzzing.

Of the two I find the buzzing most difficult to deal with. It would appear to travel through the ground, even with star grounding.

What have been your experiences with this?

[ghpicard]

I don't know what insulates the cathode from the filament with indirectly heated tubes- but there must be some impurity in it/defect with the tube I was using. I as said earlier- the tube works perfectly otherwise. My concern is not so much with the filament biasing method as with how filaments are insulated from cathodes. If there are quality issues, then this approach might not be advisable since the filament bias supply can be easily shorted out.

The heater is insulated by a specific coating that's a compromise between heat conductor and an electrical isolator.
It is specifically designed to work that way, and the breakdown voltage is always specified in the manuals.
For the 6L6-GC, for example, the heater-cathode breakdown voltage is 200V with the heater either positive or negative respect to the cathode.
I use indirect heated cathode rectifiers (6AX5) to kill two birds with the same stone, and they have a breakdown voltage of 450V. This means I can run my heaters from the same secondary of the power transformer without any problems at all.
Again, what you have there is a faulty tube that developed a heater-cathode leak. As soon as you float the heather winding of the power transformer, all is OK, otherwise IMO you are probably injecting AC in the cathode. If the cathode circuit impedance is low enough then the noise injected (assuming an AC driven heater) *could* be masked. If the impedance is high... hummmmmmmmmmmmmmmmmmmmmmmmmmmmmmm

A simple way to check whether the 6L6 is injecting noise or not (the leak could be perfectly non-linear regarding to the K-H voltage) is simply replacing the tube for a non-defective one.

What is your approach to biasing the filaments? How much voltage did you use?

I apply a voltage that's about 20V higher (more positive) than the cathode DC voltage to the heater. The voltage itself does not matter, and 20V is about 4-5 times the voltage I have at the driver tubes cathodes and it's positive.
Regarding to the values of the B+ divider I use ? A 220K/20K voltage divider with a 220 uF electrolytic capacitor in parallel with the 20K resistor, that's on the ground side. The B+ can go from 250V to 350V, unregulated but variable with taps on the power transformer winding.

But prior to doing any changes I would replace that 6L6... perhaps you are suffering so much for something that could be as simple as a plug-out and plug-in

[Niteshade]

Thanks for all the information!

Allot of noise travels through the ground system too. I am going to various methods of eliminating it.

Most of my noise problems are buzzing noises- which must be square waves. I rarely have hum issues.