Hi I know a push pull will cancel some noise but, what is the lowest noise tube pair I can use for a push pull in a hifi amp (standard transformer not UL output transformer).
Right know I have kt66 but since i am designing the power section, i have a clean slate to start, I would be happy with 40w of power.
Thank you.
Right know I have kt66 but since i am designing the power section, i have a clean slate to start, I would be happy with 40w of power.
Thank you.
Last edited:
The tubes first in the input will contribute the greatest amount of noise since their output will see the most gain in subsequent stages. Each following stage will contribute less.
This is why small signal tubes sometimes list noise and monophonic sensitivity but output tubes do not.
Output tubes should not be a major concern.
This is why small signal tubes sometimes list noise and monophonic sensitivity but output tubes do not.
Output tubes should not be a major concern.
Choose the output tubes which will provide sufficient power for your speaker sensitivity and listening preferences based on the output topology you wish to use.
Then work from output to input developing the gain structure necessary to provide that output from your signal sources.
Account for loss of gain due to NFB if it is to be utilized.
Then optimize each individual stage.
Then work from output to input developing the gain structure necessary to provide that output from your signal sources.
Account for loss of gain due to NFB if it is to be utilized.
Then optimize each individual stage.
It's doubtful whether power output tubes have ever been characterized for noise.
Higher gm would tend to have lower noise.
Higher gm would tend to have lower noise.
Last edited:
The only output tubes that have noticeable noise are broken or bad output tubes. Turn them in for a refund or replacement.
If good output tubes have a noise problem, then you have very sensitive speakers, and you have a much quieter listening room than most people's listening space (and a concert hall).
I want to hear yout system.
Before you take a dirty onion out of wet ground, peel the outer layers off first, before you cut out the onion core.
If good output tubes have a noise problem, then you have very sensitive speakers, and you have a much quieter listening room than most people's listening space (and a concert hall).
I want to hear yout system.
Before you take a dirty onion out of wet ground, peel the outer layers off first, before you cut out the onion core.
Last edited:
And lower drive voltage??
If Gm is doubled, drive voltage is half, and hiss resistance is half. Hiss voltage is 0.707. S/N is 3dB worse. No? (I'm on drugs so I may be all full of it.)
This is different from the case where you have to take the weak signal as you get it. In a power amp you adjust gain and drive to suit the power tube.
To the OP: you are not God, you can not "optimize every single point".
Good point. Amp design is all about compromise and balance, working with what you have, tweaking the design to get the best outcome. The OPT is the component that requires the biggest tradeoff, in terms of frequency .
Andy.
Maybe TS means power supply ripple cancelation? For that I would think the power tubes have to be as closely matched as possible.
Are there any noisy output tubes? At that stage the signal level is so high the noise is completely buried...
If you have noise from the tube's internal self noise, the tubes are bad. There is some shot noise, popcorn noise, and thermal noise in all tubes, but with the signal levels present at the output stage, that noise is not to worry about.
That said, it would seem logical that low mu tubes have the greatest signal to noise ratio, just b/c they need the largest signal to provide the output power.
But if you have noise at the output it is not b/c of the output tubes, look elsewhere.
If you have noise from the tube's internal self noise, the tubes are bad. There is some shot noise, popcorn noise, and thermal noise in all tubes, but with the signal levels present at the output stage, that noise is not to worry about.
That said, it would seem logical that low mu tubes have the greatest signal to noise ratio, just b/c they need the largest signal to provide the output power.
But if you have noise at the output it is not b/c of the output tubes, look elsewhere.
State of the art answer: measure them yourself.
Build the end of the Audio chain only: power tubes and OT
Plug a couple tubes you may wish to use. short grids to ground (through capacitors), apply proper bias, measure output noise across a load.
Different levels possible:
* ´plain RMS noise
* scope to see waveforms, noise , hum, etc.
* some kind of FFT analyzer to have even deeper detail.
Replace tubes and repeat.
Maybe different tubes same type and brand, or change brand or even tube types if same sockets (6L6 - EL34 - KT88 - etc.) or even wire different sockets in parallel and test octals - novals - whatever.
* put data in a table in orderly fashion, draw conclusions if you find anything interesting, and YOU publish the definitive study.
What´s not to like?
EDIT: almost simulpost with very similar ideas .... maybe there´s some merit to it 😎
Build the end of the Audio chain only: power tubes and OT
Plug a couple tubes you may wish to use. short grids to ground (through capacitors), apply proper bias, measure output noise across a load.
Different levels possible:
* ´plain RMS noise
* scope to see waveforms, noise , hum, etc.
* some kind of FFT analyzer to have even deeper detail.
Replace tubes and repeat.
Maybe different tubes same type and brand, or change brand or even tube types if same sockets (6L6 - EL34 - KT88 - etc.) or even wire different sockets in parallel and test octals - novals - whatever.
* put data in a table in orderly fashion, draw conclusions if you find anything interesting, and YOU publish the definitive study.
What´s not to like?
EDIT: almost simulpost with very similar ideas .... maybe there´s some merit to it 😎
Last edited:
And LOWER gain not higher. Assuming you could get less ripple injected from the driver stage (possible if it’s supply is regulated).
Thank you very much guys, it seems no one had done measurements on power tubes. I think i will do it, can you point me to a methodology and ready made uln boards. I never did these kind of very controlled testing, so i need your help 😉
As for equipments i only have one digital oscilloscope and many multimeter, and i can build a psu for testing.
As for equipments i only have one digital oscilloscope and many multimeter, and i can build a psu for testing.
Last edited:
What do you define as 'noise'?
Probably the only source of noise in an output tube that is inherent in its design is partition noise ...
partition noise
... which is caused by the electron beams being affected by grids between cathode and anode.
Then there is distortion and noise that is due to interference, power supply ripple or magnetic fields, which are all design considerations.
Probably the only source of noise in an output tube that is inherent in its design is partition noise ...
partition noise
... which is caused by the electron beams being affected by grids between cathode and anode.
Then there is distortion and noise that is due to interference, power supply ripple or magnetic fields, which are all design considerations.
No it will not!, noise which is by nature random doesn't cancel in a push-pull amplifier but even order distortion will cancel to a degree which depends on how well the tubes are matched and balanced.
If you are really worried about noise in power tubes it will be masked by the noise from the input and driver, noise for series connected amplifiers can be calculated as N1 + (N2-1)/A1 + (N3-1)/A2*A1... (known as Friis formula), where N1 is noise figure for first stage, A1 is gain of first stage aso. If you have 20dB gain or more in the first stage the following stages noise contribution can be almost completely ignored.
I suggested some basic tests, you might follow one or more, or any other proposal here.
* Test #2: you have an oscilloscope, that´s a powerful tool
You can see and differentiate between different kinds of noise, most scopes are very sensitive so they will show even a few mV noise, you can visually measure peak to peak voltages, that alone is over 50% of what you want to know
# test #1: your multimeter probably will have a minimum 200mV scale, kinda on the limit, but you can easily make (protoboard / perfboard) a switchable 10X/100X Op Amp based preamp,to boost its sensitivity.
So now you will have a 20mV and 2 mV full scale meter, how´s that?
meter should be true RMS so readings are meaningful
*test #3: you can inject that noise into your PC and with proper software analyze it.
In a nutshell: you can do all tests needed, spending peanuts.
Then you publish Power Tube Noise. The Definitive Study here 🙂
Not kidding.
* Test #2: you have an oscilloscope, that´s a powerful tool
You can see and differentiate between different kinds of noise, most scopes are very sensitive so they will show even a few mV noise, you can visually measure peak to peak voltages, that alone is over 50% of what you want to know
# test #1: your multimeter probably will have a minimum 200mV scale, kinda on the limit, but you can easily make (protoboard / perfboard) a switchable 10X/100X Op Amp based preamp,to boost its sensitivity.
So now you will have a 20mV and 2 mV full scale meter, how´s that?
meter should be true RMS so readings are meaningful
*test #3: you can inject that noise into your PC and with proper software analyze it.
In a nutshell: you can do all tests needed, spending peanuts.
Then you publish Power Tube Noise. The Definitive Study here 🙂
Not kidding.
Last edited:
Just pull out the driver tubes and see if you can hear any remaining noise.
Probably there will be just a little hum.
Probably there will be just a little hum.