There's some unfortunately irrelevant stuff in this thread. Look up "anode follower" for the answer to the OP question.
NFB will improve linearity, even for a 12AU7.
6CG7/6FQ7 has a different heater pinout than 12AU7, so it's not a drop-in replacement. Besides, a 12AU7's linearity is not that bad in the first place...
C17 and C18 should not be necessary. I would remove C17 and replace C18 with a proper cathode bypass capacitor of 47uF or so (high quality electrolytic rated at 25V DC or higher). Or you can leave C18 out as well, which will reduce the gain from V1B.
The second stage is a cathode follower. The output signal comes from the cathode of the cathode follower, not its plate.
"Yes, both channels hissing." - That might be caused by the lack of any grid leak resistor. As has been pointed out, either you need a grid leak resistor (V1B grid to ground) or you can make use the steps above to make this circuit an 'anode follower' buffered with a cathode follower and the NFB will create a virtual ground, fixing the problem of the floating grid on V1B.
Incidentally, the DC characteristics of a 12AU7 are quite close to those of a 6SN7. A circuit designed around 6SN7 should at least bias up close to correctly with a 12AU7 instead.
NFB will improve linearity, even for a 12AU7.
6CG7/6FQ7 has a different heater pinout than 12AU7, so it's not a drop-in replacement. Besides, a 12AU7's linearity is not that bad in the first place...
C17 and C18 should not be necessary. I would remove C17 and replace C18 with a proper cathode bypass capacitor of 47uF or so (high quality electrolytic rated at 25V DC or higher). Or you can leave C18 out as well, which will reduce the gain from V1B.
The second stage is a cathode follower. The output signal comes from the cathode of the cathode follower, not its plate.
- Replace R23 with a 100k ohm resistor.
- Take a new 470k ohm resistor. Connect one lead to the junction of R24 and R26. Connect the other end of the resistor to the grid of V1B (the first stage 12AU7). The gain of V1B is not very high (probably about 12X) so the 470k/100k voltage divider will likely result in an overall closed loop gain (with NFB) of only about 2.5X gain. You could increase the value of the added 470k feedback resistor to 560k or 680k to increase the overall closed loop gain, if needed. Adjust à ton goût.
- R24 is a 'build out' resistor which can suppress oscillation when circuit is used to drive an interconnect cable and load input with capacitance. I'd leave it there. It might do some good, and it doesn't do any harm (it's only 50 ohms. not enough to raise the output impedance in any meaningful way).
"Yes, both channels hissing." - That might be caused by the lack of any grid leak resistor. As has been pointed out, either you need a grid leak resistor (V1B grid to ground) or you can make use the steps above to make this circuit an 'anode follower' buffered with a cathode follower and the NFB will create a virtual ground, fixing the problem of the floating grid on V1B.
Incidentally, the DC characteristics of a 12AU7 are quite close to those of a 6SN7. A circuit designed around 6SN7 should at least bias up close to correctly with a 12AU7 instead.
The 12AU7 needs around 8 mA to have a reasonable Thd
And a decent sound
The anode and cathode resistor can be calculated with curves
Is not possible to set the bias at 1-2 mA
Then the FB as shown for me is not the right way
Having a 100 kohm in input in series with grid ( signal) is not a smart idea
But it is my opinion
And a decent sound
The anode and cathode resistor can be calculated with curves
Is not possible to set the bias at 1-2 mA
Then the FB as shown for me is not the right way
Having a 100 kohm in input in series with grid ( signal) is not a smart idea
But it is my opinion
Everything is a compromise.
A 100k resistor allows use of a 470k feedback resistor, which should not present an onerously heavy load on the 12AU7 cathode follower.
If adding a 100k resistor doesn't agree with you, you can use a 51k resistor instead, and a 240k NFB resistor. This will place a heavier load on the cathode follower, but the high frequency bandwidth will be a bit wider and a 51k resistor will make a little less noise than a 100k resistor. However, adding these parts will at least make a circuit that should work reasonably well, and would accomplish the goal of roughly 3X gain closed loop.
According to testing done by MerlinB for his High Fidelity Preamps book, 12AU7 THD is better at low plate current around 2 or 3mA than at 8mA or so. V1B has a very light load into V2A (the cathode follower) so a lot of plate current isn't necessary. However, I will agree that the operating point and choice of parts values for V1B might not be the most optimal. However, without drawing load lines, etc., it looks like what's there should at least work.
There might be a pot on the input, but there might not be. I can't tell from the supplied schematic. You know what they say... "When you assume, you make..."
Indeed it does!
We don't know what the DC voltages are in the OP's preamp, so we can't say for certain, but it looks like that Elekit circuit is very close to what the OP's circuit would be with shunt NFB added to make an 'anode follower' with cathode follower buffer.
He said that is a preamp so I think there is a pot
About Ibias I confirm my opinion on optimal value
Walter
I think you're getting mixed up with noise. Distortion improves with current, as for any triode.
But you're right that the 12AU7 has perfectly acceptable distortion; saying it has 'bad linearity' is a myth.
Probably one decent solution is to insert a pot of, p.e., 22 kohm, in output instead to use the feedback as shown
But it is my opinion
Walter
But it is my opinion
Walter
The distortion profile depends on the use case. A pre-amp deals with pretty small signals, while a PI deals with 10s of volts. The voltage swing in a pre-amp usually stays in a linear region. Adding nFB makes the tube linearity even less relevant
Putting a pot on the output of a gain stage makes the gain stage worse.
The stage will be getting the full amplitude signal at its input, and then you put a voltage divider at its output.
Because the stage will be getting the full amplitude signal at its input, all the time, it will have higher distortion in its output (THD is proportional to level in a triode stage).
With the pot at the input, you're reducing the signal level to the input of the gain stage, so signal levels are lower in general. Being that THD is proportional to level, there will then be less THD in the output.
Do you mean put the pot (volume control) after the cathode follower? Like this?
That amounts to the same thing.
Let's say the signal source is a phono preamp. It will be playing with its full output. That could be up to 1V rms.
V1B will amplify that by about 12X, so there will be 12V rms at the grid of V2A and something like 11.5V rms at its cathode.
You will then use the 22k pot on the output of V2A to attenuate the signal back down to 1V rms or less.
However, V1B will still be putting out 12V rms, no matter what position the 22k pot is set to. That's a lot of signal, so there will be a lot of THD.
If you make the 100k grid leak on V1B a pot, now 1V into the pot will be attenuated to 1/12th of a volt (83.3mV), amplified back up to 1V at the plate of V1B. That's a much smaller voltage swing for V1B for the same amplitude at the circuit's output. V1B will be operating with 1/12 of the THD compared to having the volume control on the output of V2A.
That amounts to the same thing.
Let's say the signal source is a phono preamp. It will be playing with its full output. That could be up to 1V rms.
V1B will amplify that by about 12X, so there will be 12V rms at the grid of V2A and something like 11.5V rms at its cathode.
You will then use the 22k pot on the output of V2A to attenuate the signal back down to 1V rms or less.
However, V1B will still be putting out 12V rms, no matter what position the 22k pot is set to. That's a lot of signal, so there will be a lot of THD.
If you make the 100k grid leak on V1B a pot, now 1V into the pot will be attenuated to 1/12th of a volt (83.3mV), amplified back up to 1V at the plate of V1B. That's a much smaller voltage swing for V1B for the same amplitude at the circuit's output. V1B will be operating with 1/12 of the THD compared to having the volume control on the output of V2A.
Last edited:
The input has a 50k shunt relay attenuator. Heater voltage is 6.3V.
So I could potentially subsituste for 6sn7. And I can add a feedback resistor with no extra grid stop? Or maybe like 27k that I have laying around?
So I could potentially subsituste for 6sn7. And I can add a feedback resistor with no extra grid stop? Or maybe like 27k that I have laying around?
Sounds like you do not need a grid leak resistor if you have a 50k resistance shunting V1B's grid.
6SN7 is an octal based tube with a completely different pinout from 12AX7.
The pinout for 6CG7 or 6FQ7 allows replacing one of those for your 12AU7 but you will absolutely need to change the heater wiring.
For 12AU7 with a 6.3V heater, pins 4 and 5 are shorted together and the 6.3V connects to pins 4+5 and pin 9.
For 6CG7/6FQ7, the 6.3V is connected to pin 4 and pin 5, and pin 9 is left disconnected or grounded to 0V.
As for the negative feedback...
1) What kind of power amplifier is this preamp circuit driving? One of these new-fangled Class D things? If so, those have input impedance of around 10k ohms.
2) You can't use really small values of resistance for the NFB voltage divider, because those appear in parallel with the load. You risk loading down the cathode follower pretty badly, which might not sound so good. If the NFB series resistor is 27k ohms and you want about 3X gain, the NFB parallel resistor will probably need to be about 120k ohms. That's going to be in parallel with the load impedance (which might be 10k ohms) and the NFB series resistor, so 27k // 120k // 10k = 6.5k ohms. That's going to be a pretty heavy load for a 12AU7 or 6CG7 cathode follower.
If you use the 100k and 470k I was suggesting, you'd get 100k // 470k // 10k = 8.9k ohms.
Maybe that's not a big enough difference to worry about. Neither are optimal.
This will work better if your power amp is a tube amp with high input impedance, or a solid state amp with 47k ohm input impedance. But if your power amp is a contemporary design with 5k ohm or 10k ohm input impedance, this circuit is going to struggle.
--
6SN7 is an octal based tube with a completely different pinout from 12AX7.
The pinout for 6CG7 or 6FQ7 allows replacing one of those for your 12AU7 but you will absolutely need to change the heater wiring.
For 12AU7 with a 6.3V heater, pins 4 and 5 are shorted together and the 6.3V connects to pins 4+5 and pin 9.
For 6CG7/6FQ7, the 6.3V is connected to pin 4 and pin 5, and pin 9 is left disconnected or grounded to 0V.
As for the negative feedback...
1) What kind of power amplifier is this preamp circuit driving? One of these new-fangled Class D things? If so, those have input impedance of around 10k ohms.
2) You can't use really small values of resistance for the NFB voltage divider, because those appear in parallel with the load. You risk loading down the cathode follower pretty badly, which might not sound so good. If the NFB series resistor is 27k ohms and you want about 3X gain, the NFB parallel resistor will probably need to be about 120k ohms. That's going to be in parallel with the load impedance (which might be 10k ohms) and the NFB series resistor, so 27k // 120k // 10k = 6.5k ohms. That's going to be a pretty heavy load for a 12AU7 or 6CG7 cathode follower.
If you use the 100k and 470k I was suggesting, you'd get 100k // 470k // 10k = 8.9k ohms.
Maybe that's not a big enough difference to worry about. Neither are optimal.
This will work better if your power amp is a tube amp with high input impedance, or a solid state amp with 47k ohm input impedance. But if your power amp is a contemporary design with 5k ohm or 10k ohm input impedance, this circuit is going to struggle.
--
The power amp is the Aspen Amplifier MAYA. There's some info about it here on this forum. 29dB gain transistor class AB amp. Input impedance about 30kohms.
Maybe I'll try with 50k/220k.
I was thinking to try with some 12au7 to 6sn7 adapters. Should work in this circuit. But with the feedback maybe it doesn't make any difference
Maybe I'll try with 50k/220k.
I was thinking to try with some 12au7 to 6sn7 adapters. Should work in this circuit. But with the feedback maybe it doesn't make any difference
I just noticed this...
Here we go again.
Why do you even want a preamp in this system? Your power amp has something like 30X voltage gain, so 1V in would swing 30V out.
Your DAC puts out 1V rms (1.414V peak) out at 0dBFS, I'm sure.
Your speakers are extremely sensitive. What are they? Big horns?
It makes no sense to add 20dB or more of gain to the 29dB you already have, and use an attenuator to take that extra gain away. You've just amplified the signal a whole bunch and turned it back down again for no good reason, which will inevitably add noise and distortion.
Perhaps all you need or want is a unity gain buffer.
If you really need to stick a tube in the signal path, you could take those two 12AU7s, parallel them, and make nice cathode followers out of them.
https://www.valvewizard.co.uk/accf.html
https://wtfamps.com/cathode-follower/
Here we go again.
Why do you even want a preamp in this system? Your power amp has something like 30X voltage gain, so 1V in would swing 30V out.
Your DAC puts out 1V rms (1.414V peak) out at 0dBFS, I'm sure.
Your speakers are extremely sensitive. What are they? Big horns?
It makes no sense to add 20dB or more of gain to the 29dB you already have, and use an attenuator to take that extra gain away. You've just amplified the signal a whole bunch and turned it back down again for no good reason, which will inevitably add noise and distortion.
Perhaps all you need or want is a unity gain buffer.
If you really need to stick a tube in the signal path, you could take those two 12AU7s, parallel them, and make nice cathode followers out of them.
https://www.valvewizard.co.uk/accf.html
https://wtfamps.com/cathode-follower/
My dac puts out maybe 0.3V. I used to match it with a passive volume control. Makes the sound very clear but somewhat flat. Probably an inmpedance matching problem.
The tube preamp makes it sound more punchy and alive. A unity gain buffer is probably more sensible.
The tube preamp makes it sound more punchy and alive. A unity gain buffer is probably more sensible.
I apologize for my little outburst there. I'm kind of a hot head, and I should watch that.
In my situation, I've found that a so-called passive preamp based on one of Dave Slagle's autoformer volume controls improved my system to the point where I got rid of any line stage. I designed my phono preamp to have a bit more gain than the usual 35dB, to get its output pretty close to the output from my DAC (which is the usual kind that puts out 2V rms at 0dBFS). I have a selector switch box into the AVC into the power amp driving the speakers.
Yes, your situation makes more sense if your DAC only puts out 0.3V at 0dBFS. (That's very low, btw.)
I found a spec sheet for the MAYA amplifier. It says: Input sensitivity: 1.51 RMS (2.14Vp) for 203 watts into 8 ohms.
Looking at the numbers in isolation, it looks like your speakers get to 110dB SPL with 1W input. Is that correct? If so, they go crazy loud on just a watt or two. That's a powerful amplifier, so it probably takes only millivolts of signal input to get the speakers playing really loudly. Is your listening room cavernous?
Assuming your DAC puts out 0.3V rms for 0dBFS, then you may want up to 6.5X gain from a preamp to bring the maximum output level of your DAC up to the usual 2V rms at 0dBFS. In that case, your 12AU7 line stage with gain begins to make sense. However, with a powerful amp and such high sensitivity speakers, it's no wonder you hear hiss from the preamp. Tubes run without NFB are going to be noisy. I'm surprised you haven't noticed an increase in hum as well. You'll want to make a very quiet preamp.
Do you also play vinyl records? Do you have any worries about matching the level from your DAC with the level from your records?
In my situation, I've found that a so-called passive preamp based on one of Dave Slagle's autoformer volume controls improved my system to the point where I got rid of any line stage. I designed my phono preamp to have a bit more gain than the usual 35dB, to get its output pretty close to the output from my DAC (which is the usual kind that puts out 2V rms at 0dBFS). I have a selector switch box into the AVC into the power amp driving the speakers.
Yes, your situation makes more sense if your DAC only puts out 0.3V at 0dBFS. (That's very low, btw.)
I found a spec sheet for the MAYA amplifier. It says: Input sensitivity: 1.51 RMS (2.14Vp) for 203 watts into 8 ohms.
Looking at the numbers in isolation, it looks like your speakers get to 110dB SPL with 1W input. Is that correct? If so, they go crazy loud on just a watt or two. That's a powerful amplifier, so it probably takes only millivolts of signal input to get the speakers playing really loudly. Is your listening room cavernous?
Assuming your DAC puts out 0.3V rms for 0dBFS, then you may want up to 6.5X gain from a preamp to bring the maximum output level of your DAC up to the usual 2V rms at 0dBFS. In that case, your 12AU7 line stage with gain begins to make sense. However, with a powerful amp and such high sensitivity speakers, it's no wonder you hear hiss from the preamp. Tubes run without NFB are going to be noisy. I'm surprised you haven't noticed an increase in hum as well. You'll want to make a very quiet preamp.
Do you also play vinyl records? Do you have any worries about matching the level from your DAC with the level from your records?
Last edited: