Similar post recently using a different method:
https://www.diyaudio.com/forums/tub...-acf-aikido-cathode-follower.html#post6706591
https://www.diyaudio.com/forums/tub...-acf-aikido-cathode-follower.html#post6706591
rongon:
I adhered to the schematic provided by Bas; see the last photo in the first post.
Regards,
Scott
I adhered to the schematic provided by Bas; see the last photo in the first post.
Regards,
Scott
If you have 6N1Ps in the first position and 6H30s in second position then simply swapping the two around could net you an 8dB decrease by itself. It's worth it as an experiment at least.
Roughly speaking, in the first stage of the Aikido circuit the 6N1P has a gain of around 18, while the 6H30 only 8ish.
Roughly speaking, in the first stage of the Aikido circuit the 6N1P has a gain of around 18, while the 6H30 only 8ish.
Last edited:
Here's the relevant portion from the manual.
The problem with this back-to-front swap as a permanent solution is that it kind of ruins the 'Aikido Mojo' by making the value of R10 sub-optimal.
If you have extra 6H30s then they could be placed in both input and output, and likely be fine without having to alter R10 or any cathode resistor values.
Someone please correct me if I'm wrong.
The problem with this back-to-front swap as a permanent solution is that it kind of ruins the 'Aikido Mojo' by making the value of R10 sub-optimal.
If you have extra 6H30s then they could be placed in both input and output, and likely be fine without having to alter R10 or any cathode resistor values.
Someone please correct me if I'm wrong.
Last edited:
Thanks, I guess I missed that.
I simulated the circuit in LTspice. Bear in mind that the simulations and real life can be different, but simulations are accurate enough to give you a starting point from which to work.
With a 220V B+ and a 6N1P in the first stage, I see 17.4X gain (24.8dB).
The 6N1P is drawing only 2.5mA plate current. The grid bias (cathode voltage) is only -1.1V. That's pretty close to 0V, which could induce grid current in some 6N1Ps. But that's probably OK.
If I put a 6CG7 (or 6FQ7) in the first stage, I see 9X gain (19dB gain).
The 6CG7 is drawing 5.6mA of plate current, with -2.4V grid bias. That's a quite reasonable operating point, and it's -6dB down from the gain with the 6N1P in circuit. That will work just fine. I'd try it. Just pop a couple of 6CG7/6FQ7 tubes in place of the 6N1P tubes. This looks like a perfectly good option, and it's easy too. (No changes to the circuit other than the tube swap.)
If I put a 6GU7 in the first stage, I see 8X gain (18dB gain).
The 6GU7 is drawing 7.8mA of plate current, which might be too much for your power supply. You'd need to check its ratings.
The grid bias is -3.35V, which is fine.
6GU7 would certainly work too, but the total current draw for the plate supply will go up from 25mA with 6N1P to 36mA with 6GU7. (it's 31mA with the 6CG7.)
If your B+ supply is capable of 40mA, then you will be fine with any of the above. The heater supply will need no changes.
--
PS -- If you have a couple extra 6N30P around (expensive buggers, those), you could put them in place of the 6N1P. Or you can use 6N6P, which is similar (basically the cheaper version).
If I put a 6N6P, I see 7.5X gain (17.5dB gain).
The 6N6P is drawing 8.1mA of plate current, with -3.5V grid bias. That's another reasonable operating point, and it's -7.5dB down from the gain with the 6N1P in circuit. That will work too, but now the plate supply will need to deliver 36mA. Also, both 6N6P and 6N30P draw a bit more heater current than the other options, so you'll need to make sure your heater supply is up to the task.
--
Last edited:
I was having similar thoughts to Rongon regarding filament current availability. I'd be tempted to use, if your filament supply can handle it, 6N6P tubes in all positions. The u is lower than the 6N1P so the gain will be lower, if that's what you want. Maybe 6N6P for the gain tube and 6N1P for the output tube to ease the filament current draw a bit?
The cool thing for a math-lexic like me is you can just swap out the Aikido pair cathode resistors to set the current flow to what you like (more or less).
Changing the output tube will necessitate a change in the "Aikido" noise cancelling resistor but Broskie provides a simple formula for that calculation. Even I can handle that one.
S.
The cool thing for a math-lexic like me is you can just swap out the Aikido pair cathode resistors to set the current flow to what you like (more or less).
Changing the output tube will necessitate a change in the "Aikido" noise cancelling resistor but Broskie provides a simple formula for that calculation. Even I can handle that one.
S.
Folks:
Thanks for the counsel! The first step will be to implement rayma's suggestion; I have plenty of Vishay 10k 1% resistors on hand and should be able to quickly determine if a 6 dB reduction is adequate. If more attenuation is needed, I'll likely buy a couple of 6CG7 tubes per rongon's advice and see if that does the trick.
Regards,
Scott
Thanks for the counsel! The first step will be to implement rayma's suggestion; I have plenty of Vishay 10k 1% resistors on hand and should be able to quickly determine if a 6 dB reduction is adequate. If more attenuation is needed, I'll likely buy a couple of 6CG7 tubes per rongon's advice and see if that does the trick.
Regards,
Scott
Just my 2 cents, the first Aikido line stage I built was all 6CG7s. I didn't like it, sounded too "tubey" and slow. It still sounded that way with 6CG7 input tubes and 6DJ8 outputs. I tried new Electro-Harmonix and NOS GE 6CG7 tubes, same slow and tubey with both. Eventually I switched to all 6DJ8s and it was fast and detailed, but not etched and fatiguing, which is what I prefer.
The 6CG7 and the 6N6P have roughly the same u so there's the option to try either without gain changing much.
S.
The 6CG7 and the 6N6P have roughly the same u so there's the option to try either without gain changing much.
S.
FWIW, I use an Aikido 6SN7 line stage, the Tubes4HiFi version. The gain is very high, which is good for my Williamson amplifiers, but the design includes an attenuator for each channel before the master volume control. This works very well for me.
The PSRR is probably calculated wrt to the gain.
Other ideas
Volume pot on the input
If it’s differential, a pot to allow partial signal cancellation.
If it has feedback you could increase it.
Other ideas
Volume pot on the input
If it’s differential, a pot to allow partial signal cancellation.
If it has feedback you could increase it.
I think my favourite first tube is 6N3... 6N6 output (for headphones) or another 6N3... Who'd have thought one of the lowest cost tubes sounds just fine? 🙂
I've been looking into ways of adding a low gain stage (approx 6dB) to the front of an Aikido cathode follower, rather than attempting to attentuate the gain that the full Aikido circuit already produces.
I'm not sure it fills that requirement but of interest is the anode follower, which Broskie himself has written about a few times and rongon has some experience in.
It appears to be an interesting solution. What I've gleaned from some initial reading is that by varying the feedback loop of the anode follower one can adjust the gain both negatively & positively as desired.
I have to admit a lot of what I'm reading goes over my head, but in this entry: Common-Cathode Amplifiers & Plate Followers Broskie provides an Aikido schematic incorporating the anode follower noting "...the Aikido amplifier accepts the feedback loop gracefully and adds its superb power supply noise rejection and low distortion and output impedance."
Here is the accompanying image. Surely it wouldn't be this simple? It's the Aikido circuit with an additional resistor.
The ratio of Rfb1 and Rfb2 defines the feedback loop and the associated gain correct?
---
PS, I can attest to the 6N6P being very nice in the Aikido.
I'm not sure it fills that requirement but of interest is the anode follower, which Broskie himself has written about a few times and rongon has some experience in.
It appears to be an interesting solution. What I've gleaned from some initial reading is that by varying the feedback loop of the anode follower one can adjust the gain both negatively & positively as desired.
I have to admit a lot of what I'm reading goes over my head, but in this entry: Common-Cathode Amplifiers & Plate Followers Broskie provides an Aikido schematic incorporating the anode follower noting "...the Aikido amplifier accepts the feedback loop gracefully and adds its superb power supply noise rejection and low distortion and output impedance."
Here is the accompanying image. Surely it wouldn't be this simple? It's the Aikido circuit with an additional resistor.
The ratio of Rfb1 and Rfb2 defines the feedback loop and the associated gain correct?
---
PS, I can attest to the 6N6P being very nice in the Aikido.
Last edited:
You will need a low output impedance source (<< Rfb1) to drive that circuit.
And the source must have an output blocking capacitor that is large enough
to not roll off the bass into the Rfb1, which is the circuit's input impedance.
Gain is: -Rfb2 / Rfb1
Input impedance is: Rfb1
And the source must have an output blocking capacitor that is large enough
to not roll off the bass into the Rfb1, which is the circuit's input impedance.
Gain is: -Rfb2 / Rfb1
Input impedance is: Rfb1
Last edited:
It is, or even simpler; everywhere in audio except cork-sniffer hi-fi where we need avocado-skin impedances and niobium conductors.
You'll still want a separate grid stop resistor. Rfb2 can be a variable and be your volume control. C P Boegli is the name to search.
All good fortune,
Chris
All good fortune,
Chris
I hope I'm not derailing SRMcGee's thread too wildly here.
Thanks Chris, I have found Boegli's anode follower article. It looks pretty dense but I'll give it a go.
The idea of making Rfb2 variable for volume control is intriguing. In essence, instead of using a pot strictly as an attenuator, it would in this case be used as a gain adjuster - increasing or reducing gain through the altering of the feedback loop?
I had thought, perhaps wrongly, that in a standard Aikido circuit (i.e. without Rfb2) that the input impedance was largely governed by the value of the grid leak resistor (1M). Is this no longer the case for Broskie's 'anode follower' version?
🙂
By simple, I suppose I mean without significant additional complication. Sometimes it feels like there's turtles all the way down.
Thanks Chris, I have found Boegli's anode follower article. It looks pretty dense but I'll give it a go.
The idea of making Rfb2 variable for volume control is intriguing. In essence, instead of using a pot strictly as an attenuator, it would in this case be used as a gain adjuster - increasing or reducing gain through the altering of the feedback loop?
I had thought, perhaps wrongly, that in a standard Aikido circuit (i.e. without Rfb2) that the input impedance was largely governed by the value of the grid leak resistor (1M). Is this no longer the case for Broskie's 'anode follower' version?
🙂
By simple, I suppose I mean without significant additional complication. Sometimes it feels like there's turtles all the way down.
Feedback can be either positive or negative, depending on polarity and timing ("phase"); it can be derived from either output voltage or output current; and it can be applied either in series with input signal or in parallel with it.
"Anode follower" feedback, now popularly called "Schade feedback", is negative feedback derived from output voltage and applied in parallel to input signal.
When signal and feedback are applied in parallel, the simplest model is a rough-and-ready assumption that open loop gain is high enough to be ignored, so signal current and feedback current into the summing junction (input grid, in this case) are equal and opposite, and therefor no signal / feedback voltage appears there - they balance each other exactly. Well, exactly enough, anyway.
Because there's negligible voltage at the summing junction, although signal / feedback currents are flying around it, the summing junction is said to be a short to ground. R=E/I. That's how input impedance becomes Rfb1.
Anode followers are useful in lots of places, but the drive current requirements need to be cranked into the design process. My favorite is a two stage RIAA amplifier with the first stage flat and the second stage with the EQ wrapped around it, "Schade style".
All good fortune,
Chris
"Anode follower" feedback, now popularly called "Schade feedback", is negative feedback derived from output voltage and applied in parallel to input signal.
When signal and feedback are applied in parallel, the simplest model is a rough-and-ready assumption that open loop gain is high enough to be ignored, so signal current and feedback current into the summing junction (input grid, in this case) are equal and opposite, and therefor no signal / feedback voltage appears there - they balance each other exactly. Well, exactly enough, anyway.
Because there's negligible voltage at the summing junction, although signal / feedback currents are flying around it, the summing junction is said to be a short to ground. R=E/I. That's how input impedance becomes Rfb1.
Anode followers are useful in lots of places, but the drive current requirements need to be cranked into the design process. My favorite is a two stage RIAA amplifier with the first stage flat and the second stage with the EQ wrapped around it, "Schade style".
All good fortune,
Chris
I was intrigued by the 'plate follower' or 'anode follower', and I needed a +6dB gain stage to boost the output from a phono stage up to the level from my DAC (peak output 2V rms), so I built one. I already had a little box with a +170VDC power supply and + 6VDC heater supply with a 6DJ8 in it, so I just put the feedback network in there. No cathode follower or anything. It worked out very well.
I did take care to fulfill a couple of requirements that allowed it to be successful.
1) The series input resistor in the anode follower is fairly large, I think 150k ohms, or maybe 120k ohms. As Chris explained in the previous post, this sets the input impedance of the anode follower stage. (Actually, the input impedance is a little lower than that, but not by much.)
2) The output impedance of the source driving the anode follower input is very low, <500 ohms. (The source is a Hagerman Bugle, an IC-based phono preamp.)
If your phono stage has a cathode follower or source follower on its output, it should be able to drive an anode follower well.
With the first stage configured as an anode follower, a higher gain triode in the first stage will yield better results than a lower gain triode. 6N1P will work a little better than 6N6P or 6FQ7 there. The output stage still should be a triode with high gm, so 6N6P or 6N30P will be the better choice for the second stage.
For a gain of 6dB (2X), you'll want Rfb2 to be somewhat more than 2X the resistance of Rfb1, to make up for the fact that the gain of the first stage triode is a lot less than infinite. (The ideal situation for an anode follower would be a device with infinite gain driving the feedback network. However, the approximately 30X gain from a 6N1P is far less than infinite.)
You'll want Rfb1 to be of a high enough value that it will present a reasonably high input impedance to the preceding stage (will load the source lightly).
On the other hand, you want Rfb1 to be of a low enough value that it doesn't cause too much of a treble roll-off due to the Miller capacitance of the anode follower triode.
I usually aim for Rnfb1 to be 47k to 150k ohms. Try 150k ohms.
I usually aim for Rfb2 to be as large as possible, so that it does not present too heavy a load to the anode follower (the load presented by the feedback loop is in parallel with the plate load resistor and rp of the first stage triode, along with the input impedance of the second stage triode).
Assuming an Rfb1 of 150k ohms, use a 330k to 390k resistor for Rfb2 and see what you get. It should come out with right around +6dB of gain.
You could make a two-position gain switch with resistor values for Rfb2 set for +6dB and +10dB gain. Or if you use a pot, put it in series with Rfb2 set for +6dB gain so that turning up the pot reduces the NFB and so increases the overall gain. If your Rfb2 is 330k, put perhaps a 250k pot in series with that resistor (not in series with Rfb1). I'm not sure I'd want a pot in the feedback loop at all times, so I'd go with the switch. But that's an individual choice.
(You could put a pot in series with Rfb1, but then increasing its resistance value would increase the amount of NFB, thus reducing the gain of the circuit. Increasing resistance in series with Rfb1 would also slightly decrease the high frequency bandwidth of the circuit.)
You do want the output cap to be a large value, so it doesn't roll off low frequencies and force the first stage to try to boost those missing low frequencies back up. I'd put something like 2.2uF as the output cap. This might be a physically large part. If you use a small value (like 0.47uF) for the output cap, you can put an input blocking cap (in series with the input, before the 1st stage grid leak) to filter out extreme lows that might upset the feedback loop. Then the 1st stage will only be boosting back musically useful low frequencies, not subsonic noise.
As you can see, there are lots of competing issues to attend to. There is no 100% ideal solution, but choosing good compromises will leave you with a well-enough-performing circuit.
I had limited room in my little box, so my output cap is a metallized polyester 2.7uF 250V Panasonic ECQ cap. Decidedly not fancy, but it seems to do the job just fine.
I did take care to fulfill a couple of requirements that allowed it to be successful.
1) The series input resistor in the anode follower is fairly large, I think 150k ohms, or maybe 120k ohms. As Chris explained in the previous post, this sets the input impedance of the anode follower stage. (Actually, the input impedance is a little lower than that, but not by much.)
2) The output impedance of the source driving the anode follower input is very low, <500 ohms. (The source is a Hagerman Bugle, an IC-based phono preamp.)
If your phono stage has a cathode follower or source follower on its output, it should be able to drive an anode follower well.
With the first stage configured as an anode follower, a higher gain triode in the first stage will yield better results than a lower gain triode. 6N1P will work a little better than 6N6P or 6FQ7 there. The output stage still should be a triode with high gm, so 6N6P or 6N30P will be the better choice for the second stage.
For a gain of 6dB (2X), you'll want Rfb2 to be somewhat more than 2X the resistance of Rfb1, to make up for the fact that the gain of the first stage triode is a lot less than infinite. (The ideal situation for an anode follower would be a device with infinite gain driving the feedback network. However, the approximately 30X gain from a 6N1P is far less than infinite.)
You'll want Rfb1 to be of a high enough value that it will present a reasonably high input impedance to the preceding stage (will load the source lightly).
On the other hand, you want Rfb1 to be of a low enough value that it doesn't cause too much of a treble roll-off due to the Miller capacitance of the anode follower triode.
I usually aim for Rnfb1 to be 47k to 150k ohms. Try 150k ohms.
I usually aim for Rfb2 to be as large as possible, so that it does not present too heavy a load to the anode follower (the load presented by the feedback loop is in parallel with the plate load resistor and rp of the first stage triode, along with the input impedance of the second stage triode).
Assuming an Rfb1 of 150k ohms, use a 330k to 390k resistor for Rfb2 and see what you get. It should come out with right around +6dB of gain.
You could make a two-position gain switch with resistor values for Rfb2 set for +6dB and +10dB gain. Or if you use a pot, put it in series with Rfb2 set for +6dB gain so that turning up the pot reduces the NFB and so increases the overall gain. If your Rfb2 is 330k, put perhaps a 250k pot in series with that resistor (not in series with Rfb1). I'm not sure I'd want a pot in the feedback loop at all times, so I'd go with the switch. But that's an individual choice.
(You could put a pot in series with Rfb1, but then increasing its resistance value would increase the amount of NFB, thus reducing the gain of the circuit. Increasing resistance in series with Rfb1 would also slightly decrease the high frequency bandwidth of the circuit.)
You do want the output cap to be a large value, so it doesn't roll off low frequencies and force the first stage to try to boost those missing low frequencies back up. I'd put something like 2.2uF as the output cap. This might be a physically large part. If you use a small value (like 0.47uF) for the output cap, you can put an input blocking cap (in series with the input, before the 1st stage grid leak) to filter out extreme lows that might upset the feedback loop. Then the 1st stage will only be boosting back musically useful low frequencies, not subsonic noise.
As you can see, there are lots of competing issues to attend to. There is no 100% ideal solution, but choosing good compromises will leave you with a well-enough-performing circuit.
I had limited room in my little box, so my output cap is a metallized polyester 2.7uF 250V Panasonic ECQ cap. Decidedly not fancy, but it seems to do the job just fine.
Attachments
Last edited: