AC filament induced hum

[KE6DF]

I have a few questions that fall under the general topic of hum introduced into audio or RF circuits from AC on filaments.

I've noticed that shielded wires are sometimes recommended on runs to the filaments of low level audio tubes like 6SJ7s used as first audio stage voltage amplifiers.

Does this make a difference? Would twisted pairs work just as well? Or I suppose shielded and twisted would be better yet.

Given that it's best to keep filament lines short around low level stages, does it make sense to have a separate filament transformer near the low level stages?

Or does this just introduce more problems from the magnetic fields generated by the transformer.

If you are going to use transformers near low level stages, do potted transformers will steel cans work better (in terms of hum) than open frame units?

Or are you better off sourcing filament voltages for a speech amp from a power supply on another chassis? But, of course, this works against the idea of keeping leads short.

Do higher heater voltage tubes have more of a problem with hum (e.g., 12SJ7s rather than 6SJ7s).

And, finally, most circuits to power 2A3 PP modulator stages run the filaments from a dedicated 2.5V winding with the center tap grounded through a resistor. This resistor is effectively the cathode bias resistor for the 2A3 -- a typical value might be around 1000 ohms.

But, what if you have a 2.5V winding with no CT? Can you just tie the cathode resistor to one side of the 2.5V filament? Or would it be better to use two resistors tied to ground -- each of 2000 ohms in this example?

And what if you have a "spare" 5V 3A CT winding that could power a pair of 2A3s by connecting one between the CT and each side of the 5V winding? If so, how would you handle grounding in that case?

Maybe this is too many questions in one topic.

Any help on any of these questions would be appreciated.

[Opcom]

I have a few questions that fall under the general topic of hum introduced into audio or RF circuits from AC on filaments.

I've noticed that shielded wires are sometimes recommended on runs to the filaments of low level audio tubes like 6SJ7s used as first audio stage voltage amplifiers.

Does this make a difference? Would twisted pairs work just as well? Or I suppose shielded and twisted would be better yet.

This is more to avoid the noise from the wires from getting into the sensitive input circuit nearby. STP (shielded twisted pair) is great if you have it. YMMV

Given that it's best to keep filament lines short around low level stages, does it make sense to have a separate filament transformer near the low level stages?

Or does this just introduce more problems from the magnetic fields generated by the transformer.

If the wires are shielded or well-twisted and run in the corner of the chassis or flat along it, they can be long if necessary. Sometimes you can tie them to the chassis with little tabs. But yes, shorter is good, just as long as the transformer does not create a magnetic field in the tubes.

If you are going to use transformers near low level stages, do potted transformers will steel cans work better (in terms of hum) than open frame units?

Or are you better off sourcing filament voltages for a speech amp from a power supply on another chassis? But, of course, this works against the idea of keeping leads short.

Some transfomers have better shielding yet look much alike. I can't answer.

Do higher heater voltage tubes have more of a problem with hum (e.g., 12SJ7s rather than 6SJ7s).

I have not noticed this. Those indirectly heated tubes have the filament bound so tightly together, the hum if any would be coming fromthe leads of the socket and inside the tube. I have never done measurements - audio tube experts speak up!

And, finally, most circuits to power 2A3 PP modulator stages run the filaments from a dedicated 2.5V winding with the center tap grounded through a resistor. This resistor is effectively the cathode bias resistor for the 2A3 -- a typical value might be around 1000 ohms.

But, what if you have a 2.5V winding with no CT? Can you just tie the cathode resistor to one side of the 2.5V filament? Or would it be better to use two resistors tied to ground -- each of 2000 ohms in this example?

And what if you have a "spare" 5V 3A CT winding that could power a pair of 2A3s by connecting one between the CT and each side of the 5V winding? If so, how would you handle grounding in that case?


If you run a separate 2K res. to gnd. from each side of the transformer, you will lose the degenerative effect of the otherwise common 1k resistor. I have done OK by putting a resistance across the filament winding and tapping it in the middle, This tap then goes through the bias resistor to ground. The tapped resistance should be low enough to not interfere with the bias level and high enough not to draw too much current from the filament ransformer. Perhaps 50 or 100 ohms tapped in the center, and the 1000 Ohm bias resistor to ground. You could bypass the bias resistor with a big cap too, maybe 50-100uF but it is for fidelity (bias regulation of sorts) and not necessary for hum reduction.

In my modulator of two 3-500Z's, I have a non-tapped fil. transformer. It is a zero bias sertup so there is no cathode bias reistor.
I put a 10 Ohm resistor from each side to GND and this gives no noticeable hum (although it would be hard to measure and differentiate from other noises in the high powered amp there).

Putting the filaments in series and sending the middle through the resistor works, but I think it is asking for more hum than the other method because the hum is out of phase to each tube, rather than in phase to both.
Common mode (in-phase hum) is ok, differential (out of phase, like push-pull hum) is going to get amplified since the tubes won't cancel in the output transformer.

Some people also say to try swapping the filament leads around on one tube, to see if the hum increased or decreases.

The previous statements are for directly heated tubes.

[KM1H]

Fully enclosed transformers or those with an electrostatic copper shield are quiet.

Ive always noticed more hum from HV filament tubes such as in AC/DC sets even in high end Zeniths such as the MJ-1035-1. Their prior version MJ-1035 used a filament transformer with 6V tubes and AC/DC for the B+ and is noticably hum free.

The best cure is to run filaments on DC.

Carl
KM1H

[KA1ZGC]

A few minor points on filaments and the wiring thereof.

WRT shielded cables, I've never seen this done. Like Patrick said, STP is the way to go if your going to go that route, but I haven't seen many STP cables that were built to handle any real current. That said, you could always twist your own with an electric drill and a vise, and slip some copper braid over it.

One thing to consider is that a lot of designers think it's perfectly harmless to feed filaments in an unbalanced configuration (tie one side of the filament to ground, and one side of the iron to ground). While this may complete the circuit, it also invites filament hum issues. Unless the tube itself has one side of the filament tied to some other element, best practice is always feed filament AC in a balanced config with a twisted pair straight from the iron. Fewer things can couple in or out of the filament supply that way. A single wire feeding a filament will induce AC hum into any conductors that find themselves running parallel for any distance.

The recommendation of a shielded filament line sounds to me like it came from someone who was trying to overcome the shortfalls of having run their filaments unbalanced in the first place, but that's just a guess.

As far as the more general concept of shielding goes: all the approaches discussed are only effective against hum that is picked up electrostatically. Faraday shields do not protect against electromagnetic interference, unless they're made from mu-metal. That's why it's good practice to run the filament voltage in a twisted pair, it allows any magnetically-induced hum or ripple to manifest as common-mode and get cancelled out.

If you experience filament hum with a twisted-pair balanced feed, that's usually (though not always) indicative of problems in the filament iron knocking the feed out of balance.

One trick I've seen in a few Harris broadcast transmitters is to use a pot or variac across part (or all) of the secondary of the filament iron with the wiper arm to ground to "dial in" the ideal filament balance, not unlike setting the carrier null on a phasing sideband rig. Mind you, these transmitters have filament leads 3/4" thick or more, so twisting was never an option.

The moral of the story is that good design fundamentals are all it takes to keep your filaments nice and quiet without any real tricks or gimmicks.

[k4kyv]

Sometimes it is the tube that introduces the hum.  For low level pre-amplifier stages, using high mu triodes like the 6F5, 6SF5, 6Q7, 6AV6, 12AX7, etc, I have always found it best to acquire about half dozen tubes and select the quietest one(s). I would most likely do this with pentodes as well. Mark the "hummy" tubes and keep them for emergency use; keep the rest on hand for spares.

Use a separate power supply located some distance away from the pre-amp. But check for excessive voltage drop in the filament line, and use larger gauge wire if needed. A steel chassis with a steel bottom cover should be used.  Twisted pair for the filament line is more effective than shielding.  Braid type shielding has little or no effect on 60~ a.c. magnetic coupling. But it is effective against electrostatic coupling. So if you use external grid cap type tubes, use a shielded grid lead, and use one of the little metal shielding caps that fit over the top of the tube. I have seen unshielded grid leads and grid caps pick up hum and noise when they are excessively long or extended outside the chassis. This was due to electrostatic, not magnetic coupling.

If after all precautions the hum persists, try DC on the filaments.  If that improves the problem, make the DC permanent. My dual-channel audio pre-amp/mixer for the D-104-dynamic mic combination uses DC on the filaments, as does the oscillator tube in my T-368 master oscillator unit. I use an old 12 volt supply that was built for a 2 m. transceiver that I no longer use, with a series resistor to drop the voltage down to 6.3 v. One of those temporary solutions that became permanent.

If the filament winding is not midtapped, a low value resistor to each side of the filament line will make an effective artificial midtap.  Just be sure each resistor is very close to the same resistance value.  The value of resistor depends on the filament voltage.  The resistors are in series with respect to the filament voltage, but are in parallel for the DC cathode current.  Use as low a resistance as possible, keeping it just high enough that the drain on the filament voltage will be negligible, and the DC voltage drop will also be negligible.

I once read of a unique solution someone used to get rid of the a.c. hum in a mic preamp for an old style condenser microphone, which operates at extremely low level.  Instead of using DC, they built a 50 kHz oscillator, followed by a power amplifier, and used 50 kHz a.c. to run the filaments of the low level stage.  This was pre-WW2, when high current/low voltage rectifiers and high capacitance low voltage electrolytic capacitors were practically non-existent, making a DC filament supply rather difficult, and even then the ripple voltage may have been excessive.

I would not recommend running a pair of 2A3s, with each tube getting its filament voltage from half of a 5v winding.

Some transformers designed for high quality audio use are potted in special alloys designed for magnetic shielding.  Thin steel cases are probably at least better than open frame.  Older broadcast quality audio transformers are sometimes potted inside cast iron cases.

[K6JEK]

The CE 100V has 12.6V center tapped filament winding with half of the 6.3V tubes on one leg, half on the other leg and the 12V tubes like the mic pre-amp using both legs.

There is something I don't understand.  They run all three wires everywhere. No filament uses the chassis as the return path.  But they ground the center tap at the transformer.  Why?

I've wondered if it wouldn't be better to lift that ground and put in a couple of hum balance pots, one across each 6.3V half of the winding.

It is a bit of a hummy beast so this isn't a completely academic question. 

[WD5JKO]

There is a Central Electronics reflector at:

http://mailman.qth.net/

Look at the archives..some recent CE posts talking about the 100V/200V power supplies, with input from Nick Tusa.

If you can get a 60db down carrier null, then you don't really have a hum issue. Others claim that these rigs can do that, so the design must be OK.

Jim,
WD5JKO

[K6JEK]

There is a Central Electronics reflector at:

http://mailman.qth.net/

Look at the archives..some recent CE posts talking about the 100V/200V power supplies, with input from Nick Tusa.

If you can get a 60db down carrier null, then you don't really have a hum issue. Others claim that these rigs can do that, so the design must be OK.

Jim,
WD5JKO

Thanks Jim.  I  occasionally post on that reflector.  In fact I started that discussion you mentioned about the virtue or vice of plug-in silicon rectifiers since they jump the voltages up.  Nick is con.  Bob Sullivan is pro. The guy who put them in my 100V was none other than Wes Schum himself, founder, president etc, of CE when he ran a little repair business after he left Zenith/CE but I'm not convinced it was such a good idea.  

The hum is most noticeable in AM and I like to run it in AM just for a change from the usual AM gear.  I have found one source. The audio input lead is routed tightly around the filter choke.  That is the stock wiring.  I re-routed it and took a bite out of the hum.  But there is more.  

I'm still puzzled.  Why ground that center tap?  If the wiring weren't buried under the filter choke, I'd be tempted to just do the experiment of lifting it and trying the Fender approach -- hum balance pots but on both sides.

[WD5JKO]

Jon,

   I did not realize that it was you on the CE reflector, and that we exchanged a few emails to boot.
    I'm hesitant to put more than 600v or so on a transmitting tube that has the plate run through the base. With a 6550, pin 3 is the plate, pin 2 is a filament pin, and pin 4 is the screen grid. Sure you can do it, but a flash over might do a whole lot of damage.

    As to why they did the 12vct thing, your guess is as good as mine. If you could separate the two windings from each other, then your dual pot idea would work. Just lifting the CT as it is, the whole filament string would float, and heater-cathode leakage would put a bias on the filament, and possibly noise.

    Is your HUM on am pure sinusoidal as seen on the scope? 60 hx, 120 hz, or a complex buzz that has harmonics of 60hz on up?

    Look at your 100v on a scope (external), and on SSB look at what is in the carrier when nulled (increase scope vertical gain). What does that look like?

    If you hear the buzz, it might be them diodes...

Jim
WD5JKO

[KE6DF]

"you mentioned about the virtue or vice of plug-in silicon rectifiers since they jump the voltages up.  Nick is con. "


This is only quasi related to this thread, but since it's my thread, I figured I could get away with posting it.

I'm not sure I agree with the general trend toward repacing tube rectifiers in old BA rigs with silicon diodes.

Seems like it causes more problems that it solves.

Tubes like 5R4's for transmitters, and small rectifiers for receivers are cheap and easy to find.

True there is more voltage sag, but then usually after someone posts about replacing the tubes with solid state rectifiers, someone else posts a suggestion to put a resistor is series with the silicon diodes to lower the voltage back down to spec. And then you are right back to voltage sag again.

Now, I understand the issue with MV rectifiers flashing over. But then, why not just replace 866's with 3B28's or 872s with 4b32s? -- both of these xenon types are cheap and reliable.

So is solid stating a BA rig really worth it?

If you really don't like tubes, wouldn't you be better off with a modern solid state rig anyway?

[WU2D]

You can run DC on the filaments of the low level stages and I have done this in desperation on 12AX7s, 6SJ7s etc... to get the hum out of the low level stages.

You can half wave rectify off the fills if you use a grounded scheme but as mentioned this is a hum maker to begin with. It is amazing how many radios are wired this way with one side of the fils grounded at each tube, and just the AC daisy chained around the chassis!

Usually it is best to rectify from a separate AC transformer or winding and bypass the diodes for hash. A 6.3 VAC or 12.6 VCT transformer winding will produce good DC with a bridge and a 4700 uF cap but the voltage will be high so you may need to attach all of your loads and see what kind of series drop you need using a variac and meter. A well bypassed low drop out regulator is a deluxe solution. But I have used a 7806 to good effect on a single or two tube because the 1.414 x E, voltage is high enough with a light load to guarantee regulation.

Mike Wu2D

[KE6DF]

"A 6.3 VAC or 12.6 VCT transformer winding will produce good DC with a bridge and a 4700 uF cap "

There may be more ripple than you think with this kind of thing.

I plugged this into the duncan amps powersupply simulator.

6.3 Vac transformer, diode bridge, capacitor filter of 4700uf, and a load of 3 ohms (resulting in a current output a little over 2 amps).

The resulting output voltage had about 2 volts peak-peak of ripple.

Seem pretty darn high for a 6.3 vdc supply.

You would either need a much larger capacitor, or a pi filter with a high current choke to get resonable ripple.

Dave

[K6JEK]

Of course 2 V of ripple is a lot less than before which was all ripple all the time.  Here's a schematic of a Marshall guitar amp that rectifies the 6.3V for the first stage filament.  It has a 1000 uF filter and not much else.  It floats but is tethered to ground by a little voltage divider.

http://www.dreamtone.org/images/Studio%2015%204001%20year1988.gif

Jim, dope that I am, I paid no attention to the carrier null pots on the 100V when operating AM. Adjusting them reduced the hum quite a bit but there is a still plenty to go around.