GainWire-NGNFB-classB-PowerAmp

Tanks to smms73 I can show here novel Error Correction circuit used in Non GNFB class B power amp.
I tried before with classic Hawksford Error Correction, but with not so good result as with th is novel EC. I Think that smms73 came to this idea by himself, an latter has found that this is also already presented by Hawksford.
The THD can be seen on the schematic.
This is not ready to be used as a working amp, as it needs to be more simulated and improved in some areas.
BR Damir
 

Attachments

  • GainWire-ClassB-EC-power-amp-sch.jpg
    GainWire-ClassB-EC-power-amp-sch.jpg
    149.6 KB · Views: 2,606
  • GainWire-ClassB-EC-power-amp-GAIN.jpg
    GainWire-ClassB-EC-power-amp-GAIN.jpg
    265.4 KB · Views: 2,485
Here are some FFTs for different power levels and different frequencies.
Damir
 

Attachments

  • GainWire-ClassB-EC-power-amp-FFT20k-50W.jpg
    GainWire-ClassB-EC-power-amp-FFT20k-50W.jpg
    242.4 KB · Views: 423
  • GainWire-ClassB-EC-power-amp-FFT20k-1W.jpg
    GainWire-ClassB-EC-power-amp-FFT20k-1W.jpg
    230.8 KB · Views: 1,866
  • GainWire-ClassB-EC-power-amp-FFTk-50W.jpg
    GainWire-ClassB-EC-power-amp-FFTk-50W.jpg
    237.3 KB · Views: 1,981
  • GainWire-ClassB-EC-power-amp-FFT1k-1W.jpg
    GainWire-ClassB-EC-power-amp-FFT1k-1W.jpg
    236.1 KB · Views: 2,140
This simulation shows how this EC is very efficient, 35 dB lower THD1k, and 25 dB lower THD20 when EC is active.
"error_resistor" is the place where the cancelation is happening, to defeat it, this resistor was shorted.
Damir
 

Attachments

  • GainWire-ClassB-EC-power-amp-EC efficiency-THD1k.jpg
    GainWire-ClassB-EC-power-amp-EC efficiency-THD1k.jpg
    241.3 KB · Views: 520
  • GainWire-ClassB-EC-power-amp-EC efficiency-THD20k.jpg
    GainWire-ClassB-EC-power-amp-EC efficiency-THD20k.jpg
    229.8 KB · Views: 450
outstanding performance even for class ab gnfb!

Thank you Pawel, you are very kind.

With this power amp, if used as integrated amp, volume control could be positioned on the input or of the current conveyor output, Rx1 to be variable.

For both cases there are one advantage and one disadvantage.
If potentiometer is at the input, the noise is higher, but distortion is lower at the low level.
If Rx1 is used as the volume, then distortion is higher and it's coming mostly from the input stage and in that way does not change with volume level, but the ratio signal to noise is the same for all volume settings, thus very low at low volume.
Damir
 
I must say 17 ppm THD20 is exceptional, and even more impressive without global feedback....
However, it has to be laid out, built, and auditioned Damir, you have at least six months of work ahead to achieve this!!
How about asking AlexMM to lay it out - he is a genius in this work - then Borys for do the work, he is a very quick, clever builder.......

Sadly I'm too bloody old and slow now, I love this stuff but my hands are fall with my larger amp designs at present.

Cheers,

Hugh
 
Hi Damir. Looking good :) .
May I suggest to split the project in two pcb , one for the voltage amplifier and another for the output stage ? That way will be easier to make changes in the future. If AlexMM would design the pcb as Hugh suggest it would be perfec , I have see some of his pcb designs and they are very good.
 
I must say 17 ppm THD20 is exceptional, and even more impressive without global feedback....
However, it has to be laid out, built, and auditioned Damir, you have at least six months of work ahead to achieve this!!
How about asking AlexMM to lay it out - he is a genius in this work - then Borys for do the work, he is a very quick, clever builder.......

Sadly I'm too bloody old and slow now, I love this stuff but my hands are fall with my larger amp designs at present.

Cheers,

Hugh

Thank you Hugh, and you are so right, I am old too and slow. So many project and so little finished. Somebody could ask why? Those simulations are very good brain exercise(to keep it young ha ha) and to build all that asks for agility and strength of youth.
Yes, of course a help from those PCB layout expert is very welcome. I think that I have to check the output, to see if a zobel or coil needed for capacitive load.
Cheers,
Damir
 
Hi Damir. Looking good :) .
May I suggest to split the project in two pcb , one for the voltage amplifier and another for the output stage ? That way will be easier to make changes in the future. If AlexMM would design the pcb as Hugh suggest it would be perfec , I have see some of his pcb designs and they are very good.

Sergio you suggestion to split it on two boards is good. I will divide schematic in two part.
What do you think of needs for zobel and coil at the output, I don't have practical experience with the non GNFB amps?
BR Damir
 
Hi Damir, Sergio, Hugh and all,

I would build this quickly for live testing - looking for a good power section for an amp with tube head-end and no GNFB or low GNGB (like 8 db loop gain) design.

The longest event will be waiting for ZVP3310/ZVN3310's to come from Mouser (takes about 1 month, can't buy them locally), all the rest may be done in a couple of weeks in parallel.

AlexMM - are you up to making a layout? Otherwise I do it myself, but you are definitely the man for this - agree with Hugh ;)

Damir, just noticed, some BC550/560 seem to be out of SOA (like Q34, Q35) - at high output swing they will face up to 80-something volts c-e. Am I right?

Cheers,
Valery
 
Like your enthusiasm Valery :) .

Let me just check the Damir circuit, as you say some transistors may be out of SOA and need to be modify.

It is possible to make the error correction part without the small mosfets, using transistors instead but the circuit gets more complex.
 
Yes, I love Valery and Sergio's enthusiasm, it's wonderful, we need more of it. We need builders, and committed designers.

Valery, I should have praised your circuit, too. However, I don't care too much about reducing H2, H3 and H4 distortion. I'm very keen to get H5 and beyond down to disappearing levels.

Think on tube sounds: they have very high levels of H2, H3 and H4. Do they sound bad?

Nevertheless, lovely design, very intelligent and refined.

Cheers,

Hugh
 
Hi Damir, Sergio, Hugh and all,

I would build this quickly for live testing - looking for a good power section for an amp with tube head-end and no GNFB or low GNGB (like 8 db loop gain) design.

The longest event will be waiting for ZVP3310/ZVN3310's to come from Mouser (takes about 1 month, can't buy them locally), all the rest may be done in a couple of weeks in parallel.

AlexMM - are you up to making a layout? Otherwise I do it myself, but you are definitely the man for this - agree with Hugh ;)

Damir, just noticed, some BC550/560 seem to be out of SOA (like Q34, Q35) - at high output swing they will face up to 80-something volts c-e. Am I right?

Cheers,
Valery

Valery thanks for your enthusiasm. As I said it's not final, needs some refinement, as the Q34, Q35, some more base resistor, and for what is more important, some more tweaking of the OPS, to be more robust with capacitive loads. So don't rush, just few days more and I think it's could be ready to start the layout.
BR Damir

ps. Sergio can help here a lot.
 
Last edited:
Valery thanks for your enthusiasm. As I said it's not final, needs some refinement, as the Q34, Q35, some more base resistor, and for what is more important, some more tweaking of the OPS, to be more robust with capacitive loads. So don't rush, just few days more and I think it's could be ready to start the layout.
BR Damir

ps. Sergio can help here a lot.

OK, cool, I have just ordered "long positions" so far, waiting for refinements before starting the layout work.

Cheers,
Valery
 
Yes, I love Valery and Sergio's enthusiasm, it's wonderful, we need more of it. We need builders, and committed designers.

Valery, I should have praised your circuit, too. However, I don't care too much about reducing H2, H3 and H4 distortion. I'm very keen to get H5 and beyond down to disappearing levels.

Think on tube sounds: they have very high levels of H2, H3 and H4. Do they sound bad?

Nevertheless, lovely design, very intelligent and refined.

Cheers,

Hugh

Thank you Hugh. I don't care about H2 and H4 too much as well, however I normally try to be careful with H3 and all the odd ones beyond...
After some experiments with the tubes at the input I really like their "character" - low noise, high linearity, rather soft reaction on overload - making them a good choice for low loop gain designs... Something I plan to "dig deeper". I really like the way they sound.

Cheers,
Valery
 
OK this one should be stable on capacitive load and the schematic shows separate IPS and OPS.
To make it stable on a capacitive I increased capacitance C5 and C6 and added in series resistors R63 and R64, increase base stoppers R52 and R53, added a coil paralled wit a resistor at the OPS output.
Q34 and Q35 changed to 2SA970 and 2SC2240.
All that changes increased THD20k to 28ppm at 50 W/8ohm.
Damir
 

Attachments

  • GainWire-ClassB-EC-power-amp-sep-IPS_OPS.jpg
    GainWire-ClassB-EC-power-amp-sep-IPS_OPS.jpg
    162.8 KB · Views: 535
OK this one should be stable on capacitive load and the schematic shows separate IPS and OPS.
To make it stable on a capacitive I increased capacitance C5 and C6 and added in series resistors R63 and R64, increase base stoppers R52 and R53, added a coil paralled wit a resistor at the OPS output.
Q34 and Q35 changed to 2SA970 and 2SC2240.
All that changes increased THD20k to 28ppm at 50 W/8ohm.
Damir

Damir, I would actually keep the source follower with the output section - it will be useful to have high input impedance and certain "isolation" there... what do you think?

Cheers,
Valery