• 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.

THD in tube phono preamp

The input tube's intrinsic signal to noise ratio is paramount.
Whether the playback RIAA curve is performed by either a passive network, or by a negative feedback loop, the rolloff rate of signal high frequencies is exactly the same (The Inverse RIAA curve is what all proper designs meet).
With passive network RIAA correction, the next stage tube's intrinsic signal to noise ratio is also very important (the passive network reduced the amplitude of the high frequency signals, but did not reduce the amplitude of the next tubes intrinsic noise).

There are lots of different kinds of noise sources.
Gaussian noise is random, and does not just cancel out because you use negative feedback.
Good Luck if you are counting on that.

OK, so which of you is going to use the super low noise 416A planar triode as your phono preamp input tube?

Just asking the question . . .
 
To amplify on this and earlier comments, several different kinds of noise matter in a phono stage. There's hum, which needs its own strategy, and can ideally but without too much drama be made completely insignificant.

Then there's Johnson/thermal/Nyquist noise, the noise caused by random field variations within resistors, including the resistances inherent in other components and including the reciprocal of transconductance in active devices including valves and semi-cons. In a well designed phono equalizer this noise should dominate, and essentially all should arise in the input. For high impedance magnetic sources like MM/MI cartridges it's common for the noise of the 47K loading resistor in parallel with the cartridge's source impedance to dominate
thermal noise.

Less well understood is excess noise, meaning everything else, but usually taken to mean 1/f noise. This is still a field of study (in many unrelated disciplines) but in general is somehow related to material junctions and is very process dependent. Because the RIAA correction boosts low frequencies so much, it's easily possible for 1/f to dominate total unweighted output noise. Very (first stage) device dependent.

The noise of a well designed phono stage will dominate any conventional attempt to measure distortion. Modern methods using a spectrum analyzer will get you quite a ways further down the road, and today a spectrum analyzer is an ADC and a laptop. A good signal source is a DAC and a laptop. Personally, I've gotten mighty tired of keeping old Sound Technology 1700s running (except as interfaces!) and am knocking on the door of the present. Soon.

All good fortune,
Chris
 
I must admit my 0.05 % was measured with a wideband meter, so it's actually THD+N.

Technically, valve white voltage noise is shot noise with space charge smoothing rather than thermal noise, but the equation for its density is the same as for thermal noise of the reciprocal of its transconductance except for some correction factor.

1/f noise is very important in valve audio designs. It tends to increase with anode current while the white noise decreases, so you get an optimum anode current that depends on the noise weighting. Merlin Blencowe wrote an interesting article about it about a decade ago.
 
Last edited:
If we take the output of a phono cartridge to be 5mV per channel, requiring 7.07mV horizontal, a very generous definition of 0VU for 5cm/sec (typical numbers might be 6 or 10dB lower) then 0,05% (-66dB) THD+N would be about 2.5uV unweighted. This is a very challenging number to get, and a testament to good design and excellent valve construction re: 1/f noise.

Noise measurements including the phono cartridge can only be worse than the effectively shorted input of a distortion test. Again, noise dominates THD+N, and sometimes (MM/MI) at the highest performance levels, limits further performance gains. The total plateaus.

This is all covered both better and in more detail in MarcelvdG's various articles in Linear Audio and here on diyAudio. Worth studying.

All good fortune,
Chris
 
Technically, valve white voltage noise is shot noise with space charge smoothing rather than thermal noise, but the equation for its density is the same as for thermal noise of the reciprocal of its transconductance except for some correction factor.

I used to think of this as "de Forest's Demon" or a transconductance "engine" operating at 1000k. But I'm straying from the subject. Nobody wants to know how things work anymore.

Much thanks, as always,
Chris
 
If we take the output of a phono cartridge to be 5mV per channel, requiring 7.07mV horizontal, a very generous definition of 0VU for 5cm/sec (typical numbers might be 6 or 10dB lower)
Fortunately, you're the wrong way up. Typical equalised programme on vinyl peaks to 12dB above 5cm/s (I've measured it). But 1/f noise is definitely a problem for valves. We really shouldn't count hum in as noise as it's a coherent man-made signal rather than the outcome of random processes.

7N7 (diyAudio tag) tried 416D but was disappointed. Given the intended use for them, I suspect any we find on land were rejects.
 
Consider the thermal noise of a resistor at 20 degrees C, that does not have any DC flowing through the resistor.
That thermal noise follows KTB. It is -174dBM / Hz.

A 1 Meg Ohm resistor has much more thermal noise voltage than the thermal noise voltage of a 47k Ohm resistor.
But the thermal noise power is still -174 dBM / Hz.
The same noise power is in any value of resistance, 50 Ohms, 1 k Ohms, etc.

Just remember, if the unweighted 2.5uV noise in post # 24 covers from 20Hz to 20kHz, the noise at each 1 Hz bandwidth is far less than it is in full bandwidth 20 - 20k.
But a 5mV 1kHz test tone that is measured with a 1 Hz wide spectrum analyzer filter, stands very tall, way out of the nose floor that is measured with the same 1Hz wide spectrum analyzer filter.
That allows us to easily see not just the 1kHz tone; but also to easily see the 2kHz 2nd harmonic, and the 3kHz 3rd harmonic of the phono cartridge.
And those distortions are rather large, versus a good phono preamp.

It should be noted that the THD + Noise measurements using classical THD + Noise measurement equipment is not the best way to measure the harmonic distortion of the phono cartridge.

Instead, the best spectrum analyzers, like the 7L5, changes the measurement from THD + N, to THD.
Unless the diamond is banging around in the walls of the vinyl, the only harmonics of significant amplitude, are the 2nd and 3rd harmonics.
And who wants to listen to music on vinyl, if the diamond is banging around in the walls?
Done.
 
Last edited:
  • Like
Reactions: Chris Hornbeck
0VU in analog days was referenced to a "working level" on a meter with well defined ballistics, as opposed to the modern definition of "running out of bits". Signal well above 0VU was expected, and a Shure survey of records available at the time (1970s) showed rare but possible peaks of +30VU. Phono cartridge sensitivity specs are given for a nominal 0VU of 5cm/sec although whether this is lateral or per channel (diagonal) is usually not stated and probably often fudged for a 3dB advantage.

I'd suspect that many folk have been disappointed trying to use very high Gm valves meant for long-line telephone/data, thinking that they would be quiet but finding high 1/f noise (unimportant for their intended use) and microphony caused by their required tricky constructions. A possible exception may be the little hot-pants Russian pentodes, 6Zh9Pi, 6Zh11Pi, etc. Some really good reports, but you may need to have already bought a supply to try. Maybe some in Ukraine.

All good fortune,
Chris
 
  • Like
Reactions: tubetvr
At this link there is a project where the EF184 in triode was used. I found some 184 TFK, Mullard , Sylvania ( I have 100 pcs9
The results are very fine

https://www.diyaudio.com/community/threads/the-phono-stage-single-child.380977/


The overload for 1% THD for thre main frequencies

1% THD con
1 kHz 280 mv 33 vout pre enfasi
20 Hz 220 mV 27 vout pre enfasi
20 kHz 85 mv 10 Vout pre enfasi

1 Khz 280 mv con 33 vout no preenfasi
20 hz 25 mV 28 Vout
20 khz 800 mv 11 Vout

The ratios between two test is 1:10 / 20dB theory. I am close

Walter