• 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.

Linearity

You can get a rough idea from the graphs. Bear in mind that these are intended for choosing biassing, not calculating distortion. They are just an average of several samples, so any particular valve will not follow them exactly but won't be off by much more than about 20%. The only way to find distortion accurately is build it and measure it.
 

artosalo

Member
2010-02-16 9:00 am
Based on my practical experience I agree with DF96.

You can get an idea of the tube's linearity from the graphs, but just an idea.

Let's take a pentode as an example. The curves are usually given with one, two or max. three fixed g2-voltages. So we do not know the values between those. Or below or above.

In case of triode, by fine-tuning the cathode resistor you have big effect to distortion of the stage.

I usually do so that after I have built the amplifier or some part of it, I optimize the stages one by one. I use audio generator to feed the planned input level and analyze the output with a distortion analyzer. I search for the minimum distortion ( = optimum linearity) by fine-tuning the component values obtained with graphical planning methods.

This method is superior compared to design a stage just using given graphs.

Actually I see that a good HIFI-amplifier can not be built without such procedure. ( except those guys who can do all this by just listening ;)) .
 
Looking at the 2 tubes I am thinking about, any initial guesses whether they will work ok for audio?

Based on my practical experience I agree with DF96.

You can get an idea of the tube's linearity from the graphs, but just an idea.

Let's take a pentode as an example. The curves are usually given with one, two or max. three fixed g2-voltages. So we do not know the values between those. Or below or above.

In case of triode, by fine-tuning the cathode resistor you have big effect to distortion of the stage.

I usually do so that after I have built the amplifier or some part of it, I optimize the stages one by one. I use audio generator to feed the planned input level and analyze the output with a distortion analyzer. I search for the minimum distortion ( = optimum linearity) by fine-tuning the component values obtained with graphical planning methods.

This method is superior compared to design a stage just using given graphs.

Actually I see that a good HIFI-amplifier can not be built without such procedure. ( except those guys who can do all this by just listening ;)) .
 
From the graphs you can get a rough estimate of 2nd, and an even rougher estimate of 3rd. You get the best estimate for the one you are probably least interested in.

You are sometimes given explicit graphs of 2nd and 3rd order products for remote cutoff RF valves (in Philips/Mullard datasheets), in terms of maximum input for 1% distortion, but you would not normally use these for audio except in a compressor.
 
The triodes in those (6AF11, 6BD11) certainly have nice flat Mu curves, which is a good indication of linearity. 6AS11 is another pin compatible tube with the same triodes, but with a 6CX8 pentode included. Have heard good comments before from others on these triodes.

Depends also on how you are going to use them. A flat Mu curve should be good for a SE class A stage.

In P-P mode, class A, and a CCS tail, a different type of triode may work well. Here you want the sum of the gm's versus cathode current to sum to a constant. This can be approximated with tubes having a linear ramp of gm versus plate current. Take a look at the triode gm curves for the 6JN8 tube just posted on another thread (bottom of page 4):

http://www.mif.pg.gda.pl/homepages/frank/sheets/135/1/19JN8.pdf

Many other tubes have this weird triode curve set like 12AT7, 6LQ8, 6S4, 6LB8. You may have noticed a lot of positive comments for 12AT7 in P-P driver stages. The huge 2nd Harmonic that would show up in SE use cancels out in P-P, and the flat gm ramp largely knocks out the 3rd harmonic due to constant total gm.

As mentioned above, an analyzer, operating point, and tuning component values is the final arbiter. The datasheet curves are only a guide, and certainly not accurate enough to determine higher order distortions. A lot of data sheets seem to be "French curve" designs too. And then there is the significant variability between different "identical" tubes too. Leave some room for variability: auto bias or bias pots, gain balance pots, screen voltage adjust pots ...
 
Last edited:
The end goal of all this is to build a pre-amp to drive the red board. I have a turntable that has a magnetic catridge, so some form of amplification is necessary. So, given that these are dual triode, pentode tubes, how would you proceed signal wise, triode->triode->pentode?


The triodes in those (6AF11, 6BD11) certainly have nice flat Mu curves, which is a good indication of linearity. 6AS11 is another pin compatible tube with the same triodes, but with a 6CX8 pentode included. Have heard good comments before from others on these triodes.

Depends also on how you are going to use them. A flat Mu curve should be good for a SE class A stage.

In P-P mode, class A, and a CCS tail, a different type of triode may work well. Here you want the sum of the gm's versus cathode current to sum to a constant. This can be approximated with tubes having a linear ramp of gm versus plate current. Take a look at the triode gm curves for the 6JN8 tube just posted on another thread (bottom of page 4):

http://www.mif.pg.gda.pl/homepages/frank/sheets/135/1/19JN8.pdf

Many other tubes have this weird triode curve set like 12AT7, 6LQ8, 6S4, 6LB8. You may have noticed a lot of positive comments for 12AT7 in P-P driver stages. The huge 2nd Harmonic that would show up in SE use cancels out in P-P, and the flat gm ramp largely knocks out the 3rd harmonic due to constant total gm.

As mentioned above, an analyzer, operating point, and tuning component values is the final arbiter. The datasheet curves are only a guide, and certainly not accurate enough to determine higher order distortions. A lot of data sheets seem to be "French curve" designs too. And then there is the significant variability between different "identical" tubes too. Leave some room for variability: auto bias or bias pots, gain balance pots, screen voltage adjust pots ...
 
I haven't worked on any Phono preamps but Morgan Jones's book has a good section on this. I think the requirements on the 1st stage are quite stringent and conflicted. Low input capacitance, high gm, high gain, low noise, low microphonics, low Rp for a triode. A lot of preamps seem to end up with a JFET for the 1st stage. If it weren't for the noise, the pentode would be a good fit, and if it weren't for the low input capacitance req. and high Mu req. the pentode in triode mode would fit. (pentode Mu is around 25) Leaving, maybe triode section 1 with the current max'd out. Since section 1 is on the end and section 2 is in the middle, one might try upping the max Watts on section 1 to 2 W and dropping it on section two (in the middle) to 1 W. Keeping the pentode below the 5 W spec some also.

Note: the diagram on the GE data sheet does not show the sections visually in the same positions as they are mechanically. Check the pins on a real tube and you will see that section 1 is in the outside position.
 
Last edited:
Here is the best MM phono corrector I've ever seen:

bf6be9d96f7b.jpg
 

artosalo

Member
2010-02-16 9:00 am
After studying the data of both tubes, I prefer 6BD11 for phono amplifier.

Below is the result of quick (bias) calculations.

[IMGDEAD]http://kuva.termiitti.com/image/14570.gif[/IMGDEAD]



Pentode section (as triode connected) is selected for 1st amplifier because it has highest gm and thus lowest noise. It is biased to relatively high anode current ( some 19 mA)

Triode 2 is selected for output stage due to it's lower Rp compared to Triode 1. Triode 1 is biased to 3,5 mA and Triode 2 to 6,0 mA respectively.
Triode 1 and 2 form a feedback amplifier and it's gain can be adjusted with feedback resistor (22k...27k in schematic). The total gain of the phono amplifier should be some 40...50 dB.

As can be seen the bias determining components are not standard sizes. The reason is that those are just calculated. The final values can be determined by optimizing the circuitry as a typical desing work.
 
Since the 6AF11.. etc pentodes have such high gm (10,000), it might be worth considering a little trickery to get one into the 1st gain stage position. The pentode screen grid current can be bypassed to the plate using a small HV Mosfet or bipolar transistor.

Basically: Mosfet Source to the screen grid through a stopper resistor. Mosfet gate to a quiet screen voltage reference thru a 1K resistor. Zener, 15V, between source and gate for protection of the Mosfet at power up, and the Mosfet Drain to the plate. Now there is no screen partition noise (since it is recombined with the plate current, this might even lower microphonics some), and the low input capacitance of a pentode, with high gm. The 1st stage only has small signals on the plate, so it won't plunge below the screen voltage here.

You might want to consider something like a 6LY8 (20,000 gm pentode) plus 100 Mu triode, or other 9DX based tubes with a lower Mu triode if the pentode is "tricked" in for 1st base.

These are no doubt way outside the mainstream RIAA pre-amp design conventions.
-----------------------
edit: Ahh, I see Artosalo has gone with the triode'd pentode in 1st position. High gm rules for low noise.
 
Last edited: