• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Tube Signal to Noise Ratio / Dynamic Range

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I would like to toss out a topic for discussion.

What is the upper limit for Signal to Noise Ratio for tube equipment or out another way what can we do as DIY practitioners do to minimize hiss and hum from our systems?

I have read of practitioners limiting the output voltage swing of a line stage as a distortion reduction technique. Practitioners use high transconductance (gm) tubes to reduce noise. High gm usually comes with high mu and gain that is burned off in a volume pot. I like high mu high gm tubes in tube buffers (less to be burned off in the volume pot). This is just a little musing to let you know where my question is coming from.

Please jump in and share your thoughts. If you have an opinion please also share your best reasoned thought of why your opinion may be correct and perhaps things to improve a circuits’ performance (please include your reasoning).

For Fun!

DT
 
What is the upper limit for Signal to Noise Ratio for tube equipment

You did not determine what tube equipment you mean, but I quess audio tube amplifiers. This is essential to determine since there are very different enviroment in RF and radio circuits.

One key factor is bandwidth, but in audio systems we can assume it is some 20 kHz.
Good tube audio preamps have (termal) noise level below 2 µV at the input.
Such amplifiers can handle input signal voltages, say 500 mV.
So signal-to-noise ratios over 100 dB are not be impossible for audio tube amplifiers.
 
SNR depends on the signal level. Very different for an MC preamp versus a line stage driven by a CD player. The main source of noise in a well-designed circuit will be flicker noise in the first valve. Unlike the shot noise which RF people have to think about, this might not be particularly well correlated with gm.

If you can ignore flicker noise (but you can't for audio) and if you are only interested in third-order distortion being below the noise level (but you are not for audio) then you can use a hand-waving argument to show that valves will give you a spurious-free dynamic range somewhere around 90dB.
 
I am speaking of the audio noise (hiss) that I hear from my “stereo” or head phones when I wake up 10 minutes after the turntable has clicked off.
I built a headphone amplifier using 6BQ6GTB’s strapped as triodes. The power amplifier is capable of better than 1.5 watts output at more than a fair amount of distortion if set up for speakers. With the 7000:300 ohm headphone output transformers this amplifier operates well below its’ potential output power, also a lot of db’s below its possible dynamic range. The way I read things is typically the distortion at the output is proportional to the power output for a given power amplifier. I built this amplifier thinking that the distortion level at 0.1 watt maximum would be pretty clean. I regulated both the B+ and H+ power supplies. This all worked out according to plan, no hum or hiss with the Sennheiser HD600’s on my head. I have not measured distortion or noise to date.

Another way of putting this, say a fictitious 10 watt, @ 5% THD, 300B SET power amplifier has a SNR of 90 db at full output. With the volume control set to run the amplifier at 1 watt output the distortion will be reduced to 0.5% THD and the noise floor will be unchanged. There is also a corresponding effective reduction of SNR as the output power is reduced.

I do not want to hear hum or hiss from the speakers. I want there to be enough head room not to hear audible distortion and or hard clipping at the loud parts. Did someone say that vinyl is pressed with a dynamic range of 45db? CD’s and DVDA can b e higher.

DT
 
High gm usually comes with high mu and gain that is burned off in a volume pot.

resistive based pots are not a good approach for this concept. resistive based pots add heat, increase distortion and reduce dampening factor. better approach is transformer volume control based on an autoformer. the turns ratio are connected to steps...typically 24 to 33 steps...full open is 1:1, as you change the position of the stepped autoformer your turns ratio will change, resistance increase, inductance increases, distortion actually decreases and dampening factor will increase. Basicaally, you are lowering the noise floor by lowering the volume level. In a resistive pot approach....noise floor is constent.
 
Artosalo,
I agree that distortion can be and in the big picture should be considered as noise. I am thinking that tube noise (like Shot and 1/f) and distortion are separate sources of noise. They can be addressed and minimized separately. While we are at it power supply noise is also a component of the overall noise. My main concern for the moment is the tube source of noise that we can hear at the speaker when there is no signal applied.

Speakerfritz,
Is the autoformer you are speaking of between stages or at the output of the power amplifier? I thought along similar lines when selecting the turn ratio (7000:300 ohms) in the in the headphone amplifier mentioned in an earlier post. When the output is reduced due to the higher turn ratio both signal and noise (floor) are reduced equally. If the output is reduced by a voltage divider at the input to the power stage only the signal is reduced, not the noise (floor) at the output.

Anything, like a crossover network, between the output transformer and speaker (driver) will reduce the dampening factor.
DT
 
Last edited:
Another approach for shifting the dynamic range window would be an output transformer with more step-down taps, which would lower the humm and hiss along with the maximum volume. So if you have super sensitive speakers or headphones you could select a greater step-down ratio. Of course the output impedance would also be lower for the lower voltage taps, which changes the sound a little depending on the speakers.

Limits?

Hum can be eliminated for all practical purposes using good power supply isolation and filtering.

Hiss can be reduced to about -120 dBu EIN (about one microvolt) without getting too heroic, as long as the source impedance is down around a few hundred ohms.

If the amplifier has say 100 watts output power with .775 volt RMS signal input, then the output noise power will be approximately 120 dB below 100 watts, which is 100 dB below 1 watt.

If your speaker has a sensitivity of 100 dB for 1 Watt, then the noise level at the reference position will be about 0 dB which is vewwwy quiet, as E.F. would say. Below the limit of audibility in a quiet room.

So it's theoretically possible to build a 100W tube amp for which output noise is not audible on 100 dB speakers. I don't have sensitivity of headphones handy but it could be worked the same way.
 
Michael,

Thanks for the shout. You are a real EE. I play with this audio stuff for fun. My Dad was a WWII radio electronics camera repairman. I have played with audio since I was a kid, capacitors were condensers to me for a long time. I know your neighborhood and Mad River skunk weed. My sister in-law lived at Dinsmore outback off the road. They all moved on years ago.

In my experience the biggest heartburn is getting the power supply quiet. Yes you can hear the power supply; rectifiers, regulators, pass transistors, droops, sags, spikes and all.

I was musing about the noise and distortion performance of the “active” part of the amplifier. I will keep digging.

For the fun!
DT
 
Dr. Raymond M. Warner Jr. Do You Know him or his work?

Dr. Raymond M. Warner Jr.

After dredging through old thought regarding Tube / valve noise I recalled that Sy had recommended a book from the 1940’s, that reported amplifier technology used to develop Atomic Energy, “Electronics Experimental Techniques” by Elmore and Sands. WWII was the peak of Vacuum tube technology. I bought the book for $20.00 from Abebooks.com. I am still digesting the content and will get back with impressions regarding noise and signal to noise ratio.

The really interesting thing is the name, university and date hand written by the previous owner inside the cover, Raymond M. Warner Jr.? Case Institute of Technology April 14, 1952. I googled him and found that he got his PhD. From CIT in 1952, worked at several brand name high tech places, originated several patents, wrote papers & books and was a professor of physics and electronics at the University of Minnesota. Really a cool guy that contributed in his life.

Was anyone a colleague, student or did you Know Dr. Raymond M. Warner Jr.? I am glad to have his book.

DT
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.