How to make a dead quiet PS for tube ?!?

[fikri]

Hi !

I always have trouble with hums when designing a power supply for tubes. And yes, i have read the articles about proper grounding (star grounding) but it seems that it always cant get enough to make this little demon quiet ! could anyone share experiences with me ?

Thanks

[The Tone God]

Alittle more information would be nice. Just off the top of my head I'd say your not filtering the power supply enough. If you are using AC for the filiments then you have to make sure you ground the center tap (or use a pair 100ohm resistors to ground from each leg) and watch the wire routing to not go near audio lines. I guess the routing applies to any AC lines.

Andrew

[fikri]

Thanks andrew, but i forgot to mention, my fillament is alredy DC and it is grounded. Im using shielded cable too for my input and output jacks.

[Johan]

with tubes, its not only the powersupply for the circuit, but also the power for the filaments...unless you run the heaters with DC, you need to make sure it's running in a way that canceles its own artifacts ( ballanced ). if you run it with 6,3v make sure it has a centre tap to ground ( 3,15-0-3,15 volt ) or make a virtual center with two resistors. if you run the heaters at 12,6volt, I would rectifie it to make sure it doesnt leak into the uadio...


johan

[The Tone God]

Then there might be a filter issue with either the DC of your filaments or your B+. Are you powering the box from an AC source then making the DC change ? Do you have a schematic of what your doing ?

Andrew

[cd]

Use a high voltage regulator.

[R.G.]

Grounding a tube amp for hum free operation is an art. The input impedances of the tubes are so  high that one wire flexing an inch out of place can give you hum.

To do it right:
(1) have one and only one ground connection between signal ground and chassis; I use isolating jacks and connect the chassis to signal ground explicitly with a single wire and verify that unhooking the wire opens them. No, don't connect the speaker jack to chassis unless that is the single point or your circuit is not using feedback to the amp from the secondary.
(2) Unless you are very experienced with tube amp grounding, use star grounding.
(3) Twist all AC-carrying wires together at about 3 twists per inch. The AREA between two AC carrying wires is proportional to how well that loop transmits hum. AC power lines, wires to rectifiers, and all heater wires need this treatment. If you're after a dead-quiet studio amp, use shielded twisted pair wire for all heater wires.
(4) Route all AC carrying wires right against the chassis.
(5) Insulate any heater centertap, and make an artificial centertap across the heater supply from two 47 ohm 1/2W resistors in series. Ground the center of the two resistors. If you have more than one heater supply do this for each one.
(6) Put the AC power at one end, the input at the other and keep them as separate as you can.
(7) Orient your power transformer, output transformer, and choke so that the tongue of each is at right angles to the other two. Put the power transformer by the AC switch, then the choke and finally the output transformer in the direction of the input jack(s).
( Use three wire safety ground, and verify by ohmmeter that it connects to the chassis and to signal ground.
(9) This is critical: Connect the rectifier side of the power supply to the negative side of the first filter cap with a single wire, preferably the centertap of the B+ winding itself. Then connect ONE wire from the negative side of that filter cap to your star ground point. Otherwise the current pulses from rectification will make intractible hum.
(10)Now you are ready to start worrying about whether your filter cap is big enough. Over 47uF for a first filter is usually wasted in a push-pull amp.

All this won't guarantee a hum free amp, but it will help a lot.

[lovekraft0]

Another simple trick with AC heaters is to reference the heater voltage to a potential above ground - in a cathode-biased power amp, for instance, connecting the heater secondary's center tap (or the voltage divider, if it doesn't have a center tap) directly to one of the power tube cathodes will do the trick. That in combination with tightly twisted heater wires, as RG mentioned, will usually eliminate any problems with noise from the filaments. I've found it to perform better than a lot of regulated DC filament supplies, at least in terms of noise.

[MetalGuy]

I always run heaters at 12.6V regulated if possible. Regulating the HV supply is not always necessary but sometimes can help. Try also moving wires around and make sure your PT is not too close to the preamp tubes especially if you have a high gain preamp. Sometimes half an inch from a wire to a sensitive spot makes big difference.

Good luck!

[cd]

I always run heaters at 12.6V regulated if possible. Regulating the HV supply is not always necessary but sometimes can help.

There's another benefit to HV regulation, but I'll leave that to you guys to figure out

[Dai H.]

hi, here is an article I translated from a Japanese site which might be helpful:

http://www.maximoaudio.com/forum/viewtopic.php?t=220

[puretube]

GREAT translation of the well-respected Kimura`s excellent explanations!

Thank you, Dai!

[Dai H.]

glad to be of service!   Hope the article de-mystifies grounding a bit. I found it very helpful for me.

[geertjacobs]

Use the Power supply Designer II software from duncanamps. It's free and lets you simulate your power supply setup.

http://www.duncanamps.com/psud2/index.html

[Jason Stout]

While reading this thread I recalled my own problems with the grounding of my senior design project. When I was in the midst of my problems a respected professor of mine said to me "Ground is not ground the whole way around." Voltage drops across a ground bus can cause all kinds of bizarre behavior! This is what star grounding tries to eliminate. I took a glance at the translated article, thanks so much Dai. And thank you R.G. for the clear and concise explanation as well!

[TryingToDo613]

RG,

Does that mean that the Torres suggestion of adding a filter cap stage decoupled from the first OT stage is a waste of time? I was just going to try building one into my firefly. He also suggests beefing up standard cap arrangements past the choke so popping the recto isn't an issue. I bought a whole case of 220uf/350 caps on ebay figuring I'd make cap farms on my amps. Please tell me if it is one of those "dubious" innacuracies I've heard about this book. -ph

[R.G.]

Here's what's going on inside a tube amp with the filter caps.

The first filter cap does all of the hard work in terms of filtering. It takes short pulses of DC from the rectifiers and holds up the DC to the rest of the amp between the pulses. This is why it has to be wired specially on the ground line - those big current pulses will be heard if any of your preamp tubes share the wire that contains the rectifier pulses.

When vacuum tubes did the rectification, you could not put huge filter caps in. A tube rectifier has a maximum pulse current, and the bigger the cap, the bigger the rectified current pulses. So tube rectifiers actually have a maximum capacitor value listed in their specification for the first filter cap. It's usually something like 22uF to 47uF, not huge at all.

The second filter cap is where all the rectification ripple vanishes. It's separated from the first filter cap by a choke, and so it forms a really effective ripple filter. You might tens of volts of ripple on the first cap, but its usually a fraction of a volt at the second filter cap. The ripple reduction is ohm's law again, but for AC. The tens of volts of ripple on the first cap go through a voltage divider of the inductor and the second cap. The inductor's AC impedance is high, the second cap's AC impedance is low, so the ripple is divided by that ratio. The choke's DC resistance is low and the second cap's DC resistance is high, so the DC is not divided.  This now-pretty-pure DC is fed to the screens, which are actually secondary signal inputs, so the ripple reduction is necessary.

The resistor-capacitor filter sections for the third through Nth sections are for decoupling. They keep the preamp sections from interacting through the DC power supply. More is better here, but the law of diminishing returns sets in pretty quickly. There is some ripple filtering in the third stage, but it's a minor function.

You can bulk up any filter cap except the first one with a tube rectifier, and all of them with solid state rectifiers without damaging the amp.

Bulking up the first one is nice if you can, but remember that push pull OT stages are relatively immune to ripple. That's one reason they were designed - they guys in the Golden Age couldn't fix the tube rectifier current problem, so they designed the circuit not to care very much. Bigger caps here reduce ripple all throughout the amp, but amps perform pretty darned well with caps of up to 47uF, so you have to decide whether you want to spend the money and time on this or not. It's already into diminishing returns.

As to the second through Nth caps - sure, go wild. It won't hurt any amp. But it's entirely possible to make a tube amp be substantially hum-free and interference-free with caps of 22uF there. I just did that.  If you hadn't already bought the caps, you could have saved some money, but you got them, so have fun.

Star wiring the B+ out to preamp stages is another little-used way to beat multistage interference that minimizes capacitor size that you won't need.

Another alternative is using a resistor-filter cap on each tube stage. That one has some real advantages in letting you be sloppier with other wiring.

But you don't really need these far-out suggestions.

I am particularly fond of stepping sideways on this one. I don't recommend you do this, because it's a huge pain and you already have your caps. But I have this concept I call The Immortal Amp where everything that fails is made incredibly durable, and the parts that simply wear out are made easy to replace. In the Immortal Amp, I would use polypropylene motor run caps to replace the filter caps. They come in values up to 47uF at 370VAC and bigger in smaller voltages. They ...never... wear out or decay like electrolytics do. So you replace all the electros with polypro caps and now the amp will never need a cap job. Ever. Did you ever read the poem "The Wonderful One-Hoss Shay"?

[Eb7+9]

[edited ... ]

In older tube amps you've got multiple HUM producing mechanisms and the single-ended nature of the gain stages in the preamp means that there's some cancelling made possible through PSU feedthrough and resulting inversion-cancelling ... those who idealize PSU operation may fail to recognize this natural form of hum cancelling ...

in fact you can make use of this cancelling ability to make SE tube amps dead quiet without reverting to DC filament biasing as well - that's part of what I do for a living ... all you have to do is introduce a measured amount of hum out-of-phase at the right place ...

~jc

[TryingToDo613]

RG,

Torres uses a corrolation between cap size and stability, and bass response. He says that the higher the better as long as it doesn't motor boat, and references some high fi amps in the tens of thousands of uf.

He does note that you have limits on tube rectos before the choke and of course mentions the fact that the benefit is pronounced in single ended amps. He actually seems to be saying the same things you are, except your original comment that over 47uf is overkill. But something you also said makes alot of sense as why it would specifically be in that book. He does say to decouple stages as much as possible and I suspect it's because of the lead dressing freedom you speak of.  Not having to find that one centimeter sweet spot is sure important enough to me to spend the money on the caps. Saves the guy who moves the wire later having to figure out what he did to make the amp hum.

He also notes that if you have a ton of filtering up front it will give you the freedom to go down to next to nothing on high gain stages that you don't want to muddy up the sound. Specifically that decoupled with a 10w/10ohm resistor first stage before the OT. I was planning on doing that with a couple of my 220s (110) on my firefly, then going down to like 4uf on the cascode boost stage and see how it goes.

Also, I don't see those caps on mouser. Could you tell us where you get them? I assume you have to put four in series, parralell side by side with resistors in a big amp, but do you have to do that on a single ended amp with lower voltages?  Thanks as usual for your thoughtful posts. -ph

[R.G.]

Just out of respect for you J.C., would you like to have a chance to re-think and edit what you've written before I reply?