my last project

[padamiecki]

Hello,
please look at my last aplifier,
sounds very good, but I got a feeling that it starts to distoring the sound too fast. I believe that its potential is greater, I built identical similar one before.
I attached measured voltages.
could you give me some advice hoe to made some regulations to improve the power?

[ray_moth]

Hi, padamiecki.

Some thoughts on how I would improve the amp circuit:

* Reduce the input grid resistor to 100k, to minimise the Miller capacitance, which could attenuate the higher frequencies.

* Add a 10k grid stopper resistor, between the input tube grid and the grid resistor, to prevent parasitic oscillations from occurring; this needs to be very close to the grid pin to be effective.

* Change the ECC83 to a long-tail pair splitter, instead of the paraphase you have now. This will give you superior blance and lower distortion. Use a solid-state constant current sink (CCS) in the tail of the splitter (i.e. in place of R25) and add a 470 ohm resistor in the cathode of each half of the ECC83, i.e. between each cathode and the CCS, to give some current feedback and to improve DC balance, since you have DC coupling to the ECC82

* For the CCS, you need a low negative voltage supply. If you have an existing separate winding of around 12v to 30v on your power transformer, which you could use for this purpose, it would be the best solution. If not, you could probably achieve it with a voltage doubler from the 6.3v heater supply but you would must disconnect the existing heater circuit from ground, so that it is grounded only via the positive end of the voltage doubler.

* Change the ECC82 to a differential amplifier instead of cathode follower (CF). There is no need to use a CF to feed the EL4s and the differential arrangement will give more gain. Use plate load resistors of around 33k, 2W and a tail resitor of 27k, 5w. Also, add 1Meg feedback resistors between the plates of the ECC82 and ECC83, which will improve balance and distortion still further.

* Bypass the cathode resistors of the EL34s (R30 and R31) with 470uF 50v capacitors, which will improve gain and available power.

* Finally, if you find that the overall gain is more than you require, you can reduce it by adding feedback resistors (say, 2.2Meg) cross-coupled between the plates of the EL34s and ECC83 (this has to be cross-coupled for the feedback to be negative). You can experiment with different values of feedback resistors to obtain the gain, damping and sound quality that you need.

[Yvesm]

Wouah,

Full balanced feedback both for the whole amp and another in the push pull cathodes !
Check for possible parasitics with such scheme !

Just a question, why do not use a commun cathode resistor at the secondary CT for the EL34 ?
Quick looking, this seems increase the PP internal impedance and degrades damping factor.

Yves

[ray_moth]

Just a question, why do not use a commun cathode resistor at the secondary CT for the EL34 ? Quick looking, this seems increase the PP internal impedance and degrades damping factor.

Yes, you're right about the need to get rid of the current NFB in the PP pair but a common bias resistor would be unsuitable for class AB1. I suggested bypassing the existing resistors instead to improve the application of self-bias.

Better still, though, would be to change to fixed bias, which would provide control over the quiescent current and make it possible to get more power. This would require a negative supply of at least -45v and a couple of pots to adjust and balance the OP tube currents.

[padamiecki]

Thank you all for your comments,
they were much appreciated

Originally posted by ray_moth
Hi, padamiecki.

* Change the ECC83 to a long-tail pair splitter, instead of the paraphase you have now. This will give you superior blance and lower distortion. Use a solid-state constant current sink (CCS) in the tail of the splitter (i.e. in place of R25) and add a 470 ohm resistor in the cathode of each half of the ECC83, i.e. between each cathode and the CCS, to give some current feedback and to improve DC balance, since you have DC coupling to the ECC82

? I thought that paraphase introduced local feedback and gives this stage more linear?
Next time I will try to update the schematic with your comments.

Originally posted by ray_moth
* Change the ECC82 to a differential amplifier instead of cathode follower (CF). There is no need to use a CF to feed the EL4s and the differential arrangement will give more gain. Use plate load resistors of around 33k, 2W and a tail resitor of 27k, 5w. Also, add 1Meg feedback resistors between the plates of the ECC82 and ECC83, which will improve balance and distortion still further.

* Bypass the cathode resistors of the EL34s (R30 and R31) with 470uF 50v capacitors, which will improve gain and available power.

* Finally, if you find that the overall gain is more than you require, you can reduce it by adding feedback resistors (say, 2.2Meg) cross-coupled between the plates of the EL34s and ECC83 (this has to be cross-coupled for the feedback to be negative). You can experiment with different values of feedback resistors to obtain the gain, damping and sound quality that you need.

well, when designed I would minimise gain to get full power on maximum pot setting. I am afraid that after resigned fro CF I will get too much gain and will have to use more nfb, which I would not to do.

Similar thing is with use decoupling cathode resistors. I would not use them.

Anyway all of your comments were very interesting, tommorow I will send updated schematic

[padamiecki]

... but should work!
Ray_Moth, could you please check if feedback loops are right? I changed the values proposed by you.

Anyway from simple schematic it became hard to pass like the maze!

I would like to point once again that schematic in the first post is working, and the sound is excellent, with very deep scene.

But who knows, if this last proposition is possible to build, maybe I will try?

[padamiecki]

the picture of working devices

[padamiecki]

next pic
probably too small?

[Yvesm]

Originally posted by padamiecki
... but should work!
Ray_Moth, could you please check if feedback loops are right? I changed the values proposed by you.

Anyway from simple schematic it became hard to pass like the maze!

I would like to point once again that schematic in the first post is working, and the sound is excellent, with very deep scene.

But who knows, if this last proposition is possible to build, maybe I will try?

My preference goes to the first one
With just a question !
I do like the CF to isolates the phase splitter from the power stage, but why are stoppers in the EL34 grids so large ?

Yves.

Here is my new born babe !

[SY]

why are stoppers in the EL34 grids so large ?

4k7 is large?

In UL, the input capacitance will be something like 40pF, maybe a bit less. 4k7 feeding 40pF creates a pole at around 850kHz, easily a decade higher than the output transformer, but low enough to be suppressing oscillation. It's a pretty reasonable choice, I think.