A good route to a ECC82 preamp

[rongon]

Ah, good catch!

Yes, just the 22k (R2) in series with 100pF (C2) in parallel yields a Low-Pass Filter with F3 of 72kHz.

But that's still high enough that it should not be causing an F3 of 17kHz from the preamp. I suspect something else is going on here...

Perhaps it's the output impedance of the volume control? If it's a 100k pot, the output impedance could be as high as 25k ohms. The voltage divider formed by the pot has a series resistance which could perhaps be adding to the value of R2, and could also be interacting with the (unnumbered) 470k ohm grid leak resistor (which is unnecessary anyway).

PS - I just thought of something. The Cin of the 12AU7 is likely around 50pF. That's in parallel with C2 (100pF), so the 12AU7's Cin would now be 150pF.
If you calculate a theoretical LPF with Rseries = 47k (the vol put Zout + R2) and 150pF, you get an F3 of only 22.5kHz.

Could that be it?

 

[indaco]

Yes, I suspect something in the input stage due to the redundant R=470k (apart the Cg-k as a minor issue). I'll remove them all. The potentiomenter is a 100kA.

Meantime I ran the new test done with REW (here the RMS out at the preamp has been raised up to 1.2V)

 

[Kay Pirinha]

Don't forget the 1st triode's Miller capacitance. Did the simulation take it into account?

Best regards!

 

[indaco]

I haven't checked with the removed components yet. Maybe I'll do a new test.

Regarding the main circuit, I was looking just for comparison with the Audionote M7 which makes use of the 6072 (12AY7) in the line preamp section.

Any improvements possible using this type of tubes? As far as I know they are loved e.g. by guitar players but in hifi they are perhaps less popular

 

[rongon]

I haven't checked with the removed components yet. Maybe I'll do a new test.

Regarding the main circuit, I was looking just for comparison with the Audionote M7 which makes use of the 6072 (12AY7) in the line preamp section.

Any improvements possible using this type of tubes? As far as I know they are loved e.g. by guitar players but in hifi they are perhaps less popular

I believe this is the portion of the M7 circuit you'd be looking at, correct?

12AY7 was used as the input stage (only) in some mid to late 1950s Fender guitar amps, but Fender had switched to 12AX7 by about 1963 or so.
12AY7 has published mu of 40, as compared to 12AX7 with 100 and 12AT7 with 60 (compared to 12AU7 with 20).
12AY7 was also used in some microphone preamps, as it was designed for low noise.
The M7 circuit as above would have a gain of about 25X (28dB). You'd need to reduce that with a lot more NFB than you're using now.

Don't forget the 1st triode's Miller capacitance. Did the simulation take it into account?

Yes, Adrian Immler's 12AU7 model includes the interelectrode capacitances, so it should model the Miller effect, etc.
Remember that 12AU7 happens to have a quite low Cag compared to 12AT7, 12AX7, etc.
Also, since the NFB reduces the gain, shouldn't applying NFB also reduce the Miller effect?

The Cin of the 12AU7 is likely around 50pF. That's in parallel with C2 (100pF), so the 12AU7's Cin would now be 150pF.
If you calculate a theoretical LPF with Rseries = 47k (the vol put Zout + R2) and 150pF, you get an F3 of only 22.5kHz.

--

 

[ruffrecords]

Have you checked the loop back response of the interface? Some of the cheaper ones drop off well before 20KHz. I have a Behringer that does just that.

Cheers

Ian

 

[indaco]

Have you checked the loop back response of the interface? Some of the cheaper ones drop off well before 20KHz. I have a Behringer that does just that.

Cheers

Ian

Yes, I read the specs...they claim the soundcard is +/-1dB linear up to 22kHz but I this is a little bit optimistic. Both RMAA and REW show about the same behaviour at high frequencies.

I believe this is the portion of the M7 circuit you'd be looking at, correct?

View attachment 1053213

12AY7 was used as the input stage (only) in some mid to late 1950s Fender guitar amps, but Fender had switched to 12AX7 by about 1963 or so.
12AY7 has published mu of 40, as compared to 12AX7 with 100 and 12AT7 with 60 (compared to 12AU7 with 20).
12AY7 was also used in some microphone preamps, as it was designed for low noise.
The M7 circuit as above would have a gain of about 25X (28dB). You'd need to reduce that with a lot more NFB than you're using now.
--

Yes, that's right. I didn't know those uses for the 12AY7, anyway it's as you say...the doubled mu proves the way higher gain for the AN circuit than mine (having also a feedback loop).
By the way, I discovered others commercial circuits using mainly the CCDA. I initially thought it was a very "disregarded" configuration respect to more common ones.

 

[rongon]

I discovered others commercial circuits using mainly the CCDA. I initially thought it was a very "disregarded" configuration respect to more common ones.

Actually, that configuration is not necessarily a CCDA. (Constant Current Draw Amplifier)

It's only a CCDA if V1 and V2 each draw the exact same plate current using identical value Rp for V1 and Rk for V2.
It's extremely common to see a common cathode stage DC coupled to a cathode follower, but these are usually not done as a Constant Current Draw Amplifier. They're usually done with more plate current through V2 (the cathode follower) than through V1.
It's also common for this to be done using a single twin-triode (6DJ8, 6SN7, 6FQ7, 12AX7, 6SL7, 12AT7, 12AU7, 12BH7, etc. etc. etc.)

 

[Kay Pirinha]

...a CCDA. (Constant Current Draw Amplifier)

Thank you for the explanation! I've always been scratching my head about what this acronym might mean.

I'm well aware that this thread is about 12AU7's the TO wants to use. Anyway and in general, wouldn't it be preferable to use a 12DW7/7247 with the high µ triode as the voltage amplifier, optionally and with a stereophonic setup, a 12AX7 at the inputs and a 12AU7 as cathode followers?

Best regards!

 

[ruffrecords]

The problem with threads like this is that the OP basically wants a preamplifier with not much gain. As a specification it is pretty vague but presumably it also implies low distortion and noise (for some definition of low). Little is said of output impedance except it is again expected to be low when usually drive capability, which is not the same, is probably more important. Stability, PSRR and other important parameters never even get a mention. As a result, there are any number of possible solutions, but insufficient information to decide which one is 'preferable'.

Cheers

Ian

 

[indaco]

Thank you for the explanation! I've always been scratching my head about what this acronym might mean.

I'm well aware that this thread is about 12AU7's the TO wants to use. Anyway and in general, wouldn't it be preferable to use a 12DW7/7247 with the high µ triode as the voltage amplifier, optionally and with a stereophonic setup, a 12AX7 at the inputs and a 12AU7 as cathode followers?

Best regards!

That explanation was useful for me too, I get the CCDA can be definied as a particular case of the DC coupled cathode follower. Anyway I meant, for what I know, that maybe it was less treated and employed that the generic one...

Kaypirinha. Actually, the 12DW7 could be a good match in alternative to 12AU7, since the 12AX7 is better as first stage in preamp mainly. Tubecad mentioned a surprisingly good buffer with the 12DW7.

Ruffrecords, you are right but but I don't have all the necessary tools to carry out in-depth "ad hoc" circuit analyzes. I limit myself to a few fundamental measures that give me some indications on what I have performed (good or not good...). If there is anyone more expert than me interested in implementing the circuit and performing the measures mentioned, well I'd be happy too.

 

[rongon]

Actually, the 12DW7 could be a good match in alternative to 12AU7, since the 12AX7 is better as first stage in preamp mainly. Tubecad mentioned a surprisingly good buffer with the 12DW7.

Again, I think you have to define what you're trying to achieve here. Why would a 12DW7 be better than a 12AU7 in this application?

One possibility is that you want more gain, but that's not what I thought you had in mind. (12DW7 has one triode with mu = 100 and second triode with mu = 20) Also, this makes a true CCDA configuration impossible, since in a 12DW7 the high mu triode draws far less current than the med mu triode.

Another possibility is that you want the extra gain so you can apply more NFB and lower THD, etc. But again, that's not what I thought you had in mind.

As for PSRR, the argument for using a true CCDA (same current draw from both V1 and V2) is that since the first stage is inverting and the second is non-inverting, and if both stages use identical triodes with identical operating points, when V1 draws current in negative direction V2 draws current in positive direction, so the psu sees no net change in current draw (hence the "Constant Current Draw Amplifier"). This is supposed to make it easier to implement a clean and quiet enough power supply.

In the end, we have to ask: What exactly do you want to achieve with this?

 

[Kay Pirinha]

As this was my idea, not the OT's: I suggested to increase open loop gain, then increase NFB to lower THD, output impedance etc.

Best regards!

 

[ruffrecords]

Ruffrecords, you are right but but I don't have all the necessary tools to carry out in-depth "ad hoc" circuit analyzes. I limit myself to a few fundamental measures that give me some indications on what I have performed (good or not good...). If there is anyone more expert than me interested in implementing the circuit and performing the measures mentioned, well I'd be happy too.

I agree, I do the same myself, but I do consider all the other factors when designing.

Cheers

Ian

 

[triode_al]

Above post #67 there is an ECC82 design with positive feedback, Ra = 27K. I simulated the design, the positive feedback node at the cathode was 'in phase' with the input signal: not positive feedback.

Instead of that design I used Ra = 120K. Totally different. I have used an EAR-RIAA board (Zero-zone) for implementing the design complete with positive feedback :

1. leaving out the first tube and going into the summing node with 110K, with 330K feedback. I inserted the NFB resistor instead of the cap at the RIAA in the board. No parallel capacitor needed. Feedback is 12 dB. I have Ra = 120K
2. The positive feedback is 42K/1k2. I insert the PFB resistor from the output strip to the cathode resistor R18. Easy peasy.
3. In simulation the signal at the kathode is 'out of phase' and about -20 to -23dB dB ref the summing node input. No instability, I did not see any ringing tendency.
4. In simultation, blocking the neg feedback, the response stays flat. The pos feedback does not have a center frequency.
5. I can switch the PFB-effect in and out. The difference is about 2 dB increased with it applied open loop; closed just 0,5dB. So with the open loop amplification is + 24dB. But in itself the ECC82 has its mechanical and theoretical limits at around 22dB.
6. The bandwidth as measured is 5Hz-108kHz.
7. The square wave is very nice: no overshoot. I used PIO for out at first a few moths ago but yesterday inserted tin foil.

So looks good.

I also simulated the ECC83 variant. There the effect of PFB is bigger. At 100K/470 it is neg phase level almost the same level as the input. . . That is an idea is for those wanting more open loop at >35 dB and thus more neg feedback. But it also should be very dynamic! [I once borrowed an amplifier from Peter van Willenswaard. It blew my head off in dynamics and livelyneess. No, he didn't use a ECC82 ..]​

The ECC82 circuit without P-FB and a PIO output cap was not very exiting. Maybe the PIO have to settle, but I gave it no time to do. I like a warm sound. It had not that thing.
I did noticed the quality of the feedback and input resistor are crucial. As big as possible, I use 5W carbon.

I had the EAR board, so used it. There is also a board for a Conrad Johnsen ECC82 design, an ECC82 input and an ECC82 cathode follower, like the post 67. I have the power supply like in that board and the delay relais. But I precede the solid state bridge with a tube rectifier for slow start.

 

[Kay Pirinha]

A side note: Your paper in oil capacitors probably are decades old by now. Why should they »settle«? What do you expect to be altered within them during that process?

Best regards!

 

[triode_al]

The PIO have their mysteries. I don't really know. One tries this and then that if the result is not good and I do not know what the reason is.
I had PIO telephone output caps for years, they are very good.

 

[ruffrecords]

Above post #67 there is an ECC82 design with positive feedback, Ra = 27K. I simulated the design, the positive feedback node at the cathode was 'in phase' with the input signal: not positive feedback.

The input signal is inverted by the first triode but not by the CF. The PFB signal is therefore in antiphase to the input signal and is fed back to the first cathode and so it is therefore positive feedback. When you probe the cathode what you will see is the combined influence of the input to the grid and that from the PFB. The resulting signal will still appear to be in phase with the input signal but there is still PFB going on.

Cheers

Ian

 

[indaco]

Above post #67 there is an ECC82 design with positive feedback, Ra = 27K. I simulated the design, the positive feedback node at the cathode was 'in phase' with the input signal: not positive feedback.

Instead of that design I used Ra = 120K. Totally different. I have used an EAR-RIAA board (Zero-zone) for implementing the design complete with positive feedback :

1. leaving out the first tube and going into the summing node with 110K, with 330K feedback. I inserted the NFB resistor instead of the cap at the RIAA in the board. No parallel capacitor needed. Feedback is 12 dB. I have Ra = 120K
2. The positive feedback is 42K/1k2. I insert the PFB resistor from the output strip to the cathode resistor R18. Easy peasy.
3. In simulation the signal at the kathode is 'out of phase' and about -20 to -23dB dB ref the summing node input. No instability, I did not see any ringing tendency.
4. In simultation, blocking the neg feedback, the response stays flat. The pos feedback does not have a center frequency.
5. I can switch the PFB-effect in and out. The difference is about 2 dB increased with it applied open loop; closed just 0,5dB. So with the open loop amplification is + 24dB. But in itself the ECC82 has its mechanical and theoretical limits at around 22dB.
6. The bandwidth as measured is 5Hz-108kHz.
7. The square wave is very nice: no overshoot. I used PIO for out at first a few moths ago but yesterday inserted tin foil.

So looks good.

I also simulated the ECC83 variant. There the effect of PFB is bigger. At 100K/470 it is neg phase level almost the same level as the input. . . That is an idea is for those wanting more open loop at >35 dB and thus more neg feedback. But it also should be very dynamic! [I once borrowed an amplifier from Peter van Willenswaard. It blew my head off in dynamics and livelyneess. No, he didn't use a ECC82 ..]​

The ECC82 circuit without P-FB and a PIO output cap was not very exiting. Maybe the PIO have to settle, but I gave it no time to do. I like a warm sound. It had not that thing.
I did noticed the quality of the feedback and input resistor are crucial. As big as possible, I use 5W carbon.

I had the EAR board, so used it. There is also a board for a Conrad Johnsen ECC82 design, an ECC82 input and an ECC82 cathode follower, like the post 67. I have the power supply like in that board and the delay relais. But I precede the solid state bridge with a tube rectifier for slow start.

If you like you can insert the circuit diagram you modified from the original, so we all can look at.

The "zerozone" you mentioned is a pre-phono where the 12AX7 are used, with a high gain. Your measurements seem nice although I didn't fully understand all your modifications. The problem is the doubled amplification factor than mine, in fact I set myself a gain of about 4: 1 (+ 12dB).

Another matter is the Conrad Johnson clone, which seems more interesting. Incidentally, I had also noticed this in my research. Maybe it would be not bad as alternative...


PS: if someone has time and desire to keep up with it, I would like to propose to send my pcb boards for free (only paid for shipping, preferably in EU). They are 2 for the preamp 1 for the power supply.

 

[rongon]

That Zero-Zone PCB is based on a topology that used to be called a "Ring of Three".
What about a Ring of Three?

The first stage is a common cathode voltage amplifier.
That's RC-coupled to the second stage common cathode voltage amplifier.
The 2nd stage is DC-coupled to a cathode follower.

The CF drives the feedback loop with a lower impedance than would be available from the plate of the 2nd stage triode.

That circuit could be done using 12AU7 tubes and it could turn out to be pretty good.

If the purpose of this thread is to find the best route to an ECC82 preamp, then yes, I agree! A 'Ring of Three' using 12AU7 tubes is a good route to take. A stereo line stage will require three 12AU7s, or one could substitute a MOSFET follower for the 3rd stage cathode follower, which would be an improvement.