Noisy triode

[Kit]

Hi everybody.

What a great forum, wish i would have found out about it earlier.

I´ve got a question for the tube experts out there,since im more of a transistor kind of guy.
Recently i have been tampering with a 30w PP amp,
(my first tube project)
its basicly your standard ef86 driver, dc coupled into ecc83 LTP PI, into 2XEL34.
However, since i didnt have an ef86 at hand , i decided to use half an ecc83 and just drop the B+ so i would end up with about 100v on the plate.
Now here´s the thing, i get alot of noise from that single triode, not 50/60hz hum but noise/hiss.
My scope just broke down so im afraid i can´t tell you what it looks like.

The gain of the ecc83 driver is about 30 with a 100k plate resistor,
The gain of your standard triode strapped ef86 is about 25 right?
I can understand that my driver will have a little hard time driving the
miller capacitanse?(forgive me, i´m swedish) off the PI but other than that i should be fine right?´
Now what im wondering is, Are tubes this noisey or have i missed something?
Or is this some sort of wierd rf interference because its not in the B+ line.
I will try and post a schematic later on.
I would be very greatfull for any replys.


Kit.

 

[PRR]

It is not the tube. Unless it is sick, which is common.

Or mis-biased, but I can't begin to guess what you did from your description. How did you reduce the B+? Did you re-wire the socket correctly? It sounds like you know what you are doing, but the triode is hissing, so I can't be sure.

EF86 in a power amp would normally be run as pentode, not triode.

 

[Kit]

The plate voltage of the triode is a little lower than ideal (100V) but that shouldn´t be a cause of noise,i suppose?
At present there is no global NFB so gain is high but still, theres is to much noise. Other than that the amp functions like intended.

Kit.

 

[Kit]

Oops! :shock:

That didnt work out, anyway the B+ is dropped from the phase inverter´s 300v to 200v thru a 100K/22uF R/C filter, The driver is biased at about 1mA with a 100K plate resistor and a 560R cathode resistor so we end up with about 100V at the grids of the PI.
I have tried to change tubes but it doesn´t make a difference, they are all brand new tubes by the way.

Kit.

 

[Marik]

Ah,

With no NGF your gain is way too high and you just amplify the noise of the input stage. Close the NGF , and you are gonna be allright.

 

[Kit]

Thanks for the replys, PRR and Marik.

I will certainly try it with a closed loop, but i still think there is something more going on here, i read a post recently on this forum about a guy who had interference problems in his microphones and Bill Whitlock replys as follows:

"This is a classic case of RF interference in a microphone. The buzzing (which imitates the power line variety) is caused by RF interference from a TV transmitter. Peak power occurs during vertical blanking of the video - and this occurs at a frequency of 59.94 Hz for US television. Therefore the sound is virtually indistinguishable from power-line buzz. You can verify the TV origin by looking at the signal with an oscilloscope whose trigger is line-locked ... the waveform will slowly crawl horizontally. It would be stationary if the buzz source were the power line. "

Could it be something like this?, the reason im asking is because i happen to live very close to a TV transmitter.
The buzz im getting is not that low frequensy but it makes me wonder though...

Kit.

 

[Kit]

Ok. i dont know if anybody cares about this thread but ill guess i´ll write for therapeutic reasons :grin: Anyway, i closed the NFB loop and the noise floor went down, but we all knew that it would , however there was still to much noise present at idle, so in desperation i put a ferrite bead on the grid wire to the first stage and the darned thing went absolutly silent :shock:
I cant explain it, it shouldnt make a difference at those frequensis but it did.
Any thoughts from somebody who knows more about this kind of problem than i do?

Kit.

 

[Guest]

Just a guess, but since you describe your place as one with a high rf field your ferrite bead placement quieting the grid makes sense. You could also try a grid stopper resistor in similar situation. I think the bead is acting as such in this case.

 

[Kit]

Your guess does make sense, Thomas but what i dont understand is this:
i did have a 47K grid stopper on the input stage soldered directly at the grid pin, and while that killed the usual "picking up" of radio broadcast it didnt do away with the noise, while the bead made it all go away.
Maybe an even bigger grid stopper resistor would do the same thing but 47K seems large enough to me.

 

[bcarso]

47k is enormous for a grid stopper, whose purpose is to eliminate tube parasitic oscillations, not block external RF interference. A few hundred ohms is normal for this location. You don't want all that thermal noise there either.

Bear in mind that resistors have significant parasitic capacitance, so 47k doesn't look like that at sufficiently high frequencies.

Ferrites are lossy inductors---in fact they are often described as possessing a resistance at some radio frequency, and sometimes by how much current they can carry while preserving that impedance (I have some on a schematic called out as "120 ohms 800mA" for example). So they not only add inductance to a line but also dissipate some of the energy that might be induced.

Bill W.'s comment is probably bang on. I would verify by doing as he says---use power line sync for the 'scope and watch the retrace blanking noise slowly crawl across the screen to verify.

 

[Kit]

"47k is enormous for a grid stopper, whose purpose is to eliminate tube parasitic oscillations, not block external RF interference."

Can it not serve both purposes?

http://www.aikenamps.com/InputRes.htm

Kit.

 

[bcarso]

Too much noise in 47k unless you are already at nice high levels and aren't worried about such.

Also, it could slow things down considerably depending on the stage gain due to the Miller effect.

Aiken's examples refer to guitar amps, which are maybe about the most challenged in terms of having big nasty antennas on their inputs, and maybe the least stringent in terms of requiring wide bandwidth. But adding a bunch of extra lowpass using the tube input capacitance just smacks of sloppy to me.

For really frisky supertriodes like the 6C45 I've needed as much as 300 ohms to stop local parasitics, but a lot more (470k!!!) seems ridiculous overkill.

You are in a high RF level environment. Strange things go on. Sometimes a rusty rack screw will alter the behavior of equipment, and how it does so will depend on the temperature and how tight it is. I have some stories from Mt. Wilson Observatory....

 

[CJ]

You might have "fixed" something when you un-soldered-added bead-then re-soldered.

 

[Guest]

[quote author="Kit"]so in desperation i put a ferrite bead on the grid wire to the first stage and the darned thing went absolutly silent :shock:
[/quote]
Kurtz-Barkhausen oscillations.
If grid is grounded for HF, electrons between cathode and grid
oscilates at some VHF-UHF region, frequency depends on tube
construction.
Ways to taming it:
1] Ferite bead {as you did it}
2] small inductancy {some 0.1 - 1 microhenries} bridged by resistancy {0.1-1 kiloohms} in series to grid
3] small resistor {100 Ohms} in series to grid.

Variant 3 is usual, but degrade noise performance in audio frequency.
Variants 1,2 are more costly, but better from noise view.

xvlk

 

[Kit]

"Kurtz-Barkhausen oscillations.
If grid is grounded for HF, electrons between cathode and grid
oscilates at some VHF-UHF region, frequency depends on tube
construction."

xvlk, that is very interesting, i did not know this.
I will try and google it to see if i can find some info.

 

[Guest]

[quote author="Kit"]Kurtz-Barkhausen oscillations...
, that is very interesting,...[/quote]
Some old book on my bookshelf have great info on that oscillations:
Sarbacher, R.I.;Edson, W.A.: Hyper and Ultrahigh Frequency Engineering.,Willey, 1946, p.502-p551
But maybe A.T. Starr s book will be good, don t remember title.

xvlk