By the way, dadod, your preamp is a must. A unique occasion to get an idea of the subjective differences between two of my favorite designs concerning feedbacks. Current feedback VS no global feedback. In the best conditions we can imagine, because most of the components stay the sames in both situations.
So, i think it would be VERY interesting that all the people witch had build this piece of cake report detailed descriptions of their feelings in the two modes, and the differences they can feel, if any. (Oh, btw, we have to chose carefully the best quality parts we can find for R19, R46 ;-)
It is a pity i cannot try-it in my digital multiway active system (too much work, not enough room in the boxes etc.).
So, i think it would be VERY interesting that all the people witch had build this piece of cake report detailed descriptions of their feelings in the two modes, and the differences they can feel, if any. (Oh, btw, we have to chose carefully the best quality parts we can find for R19, R46 ;-)
It is a pity i cannot try-it in my digital multiway active system (too much work, not enough room in the boxes etc.).
Last edited:
By the way, dadod, your preamp is a must. A unique occasion to get an idea of the subjective differences between two of my favorite designs concerning feedbacks. Current feedback VS no global feedback. In the best conditions we can imagine, because most of the components stay the sames in both situations.
So, i think it would be VERY interesting that all the people witch had build this piece of cake report detailed descriptions of their feelings in the two modes, and the differences they can feel, if any. (Oh, btw, we have to chose carefully the best quality parts we can find for R19, R46 ;-)
It is a pity i cannot try-it in my digital multiway active system (too much work, not enough room in the boxes etc.).
Thank you Christophe, I would like to here those opinions too, but not to much respond up to know. I hope to get more soon as 20 boards where sold.
I am not the one who can judge this preamp as author and owner of a pair of old ears.
Damir
Tss tss. As all the designers, i think you are able to judge your own work. If not, how can-you make choices, improve and optimize your designs ?I am not the one who can judge this preamp as author and owner of a pair of old ears.
Please, just replace "old" by "experienced" ;-)
This said, as you are the one which probably had spend the most time listening to the two versions, i'm curious to read your reports.
Tss tss. As all the designers, i think you are able to judge your own work. If not, how can-you make choices, improve and optimize your designs ?
Please, just replace "old" by "experienced" ;-)
This said, as you are the one which probably had spend the most time listening to the two versions, i'm curious to read your reports.
When I started to design this preamp, idea was to design non GNFB amp. As the configuration allows very simple introduction of CFA I thought that I should add that possibility too.
I can confess that mostly I am listening in the CFA mode, as I use it with my VFA TT power amp and not to often with the headphones (strange I designed it as a headphone amp??) . 200W CFA VMOSFET is not ready yet(only one channel is working for now). I like non GNFB mode too and I m not sure why I use CFA mostly. Could be because I know THD is lower and I am to much an engineer. I hope I will get more responses from the builder soon.
The so-called Baxandall super pair, in fact invented and patented by the late Frank Boxall in about 1956 (!), is an example of control-electrode current recycling, following the terminology of Wurcer. At Tektronix it was again re-invented by Larsen and known within the company as a super-alpha pair.So the baxandall super pairs are a variant of the darlington or CFP... they form a super transistor.... In this case are they used as followers ?
The greatest advantage of the construction is low effective output capacitance compared to most other circuits. The collector-base capacitance of the primary device, which creates a component of the collector current when the collector voltage changes, is itself variable with this voltage and amounts to a distortion generator. In the Boxall topology this displacement current is mostly pumped back into the primary device's emitter, provided that the impedance at the emitter is small compared to the rest of that node's. In the case of the amplifier under discussion the current source biasing assures that this obtains. R18 could be replaced with additional current sources but this would have a small effect.
So, since the displacement current of Ccb (say of Q6) is recycled, with about the only loss due to the finite alpha of the auxiliary transistor (say Q9), the effective output capacitance at the collector of the main device is practically zero, and variations in it have a correspondingly small effect.
Another way to achieve a comparable result was advanced by Hawksford, and once again predates his particular cascode construction by many years, having been disclosed by Aldridge. With the latter's circuit the cascode device has its control electrode current recycled to the emitting electrode of the lower device. It has a slightly higher voltage burden than Boxall but has other advantages.
One issue with these schemes is that oscillation can occur if there is not a little lumped capacitance in the output, as at high frequencies the Boxall or Aldridge has a slightly negative output capacitance. This capacitance can still be very small to stabilize things.
There is a contribution from Walt Jung somewhere within this site that details the history of these and related circuits. I would stress that I am sure none of the re-inventors knew of Boxall or Aldridge, or other related inventors (Csanky did a version with JFETs). Boxall and Aldridge were working on telcom repeater amplifiers in the early days of transistors.
Brad
In the two builds I see in this thread, both decided to connect gnd wire from output rca to the shunt star gnd.
I believe this is an opportunity to enlarge loop area because signal wire and gnd wire are soldered quite far apart in the board.
Can I solder the gnd from the output rca directly to the gnd shunt wire near the board outputs ?
I believe this is an opportunity to enlarge loop area because signal wire and gnd wire are soldered quite far apart in the board.
Can I solder the gnd from the output rca directly to the gnd shunt wire near the board outputs ?
The so-called Baxandall super pair, in fact invented and patented by the late Frank Boxall in about 1956 (!), is an example of control-electrode current recycling, following the terminology of Wurcer. At Tektronix it was again re-invented by Larsen and known within the company as a super-alpha pair.
The greatest advantage of the construction is low effective output capacitance compared to most other circuits. The collector-base capacitance of the primary device, which creates a component of the collector current when the collector voltage changes, is itself variable with this voltage and amounts to a distortion generator. In the Boxall topology this displacement current is mostly pumped back into the primary device's emitter, provided that the impedance at the emitter is small compared to the rest of that node's. In the case of the amplifier under discussion the current source biasing assures that this obtains. R18 could be replaced with additional current sources but this would have a small effect.
So, since the displacement current of Ccb (say of Q6) is recycled, with about the only loss due to the finite alpha of the auxiliary transistor (say Q9), the effective output capacitance at the collector of the main device is practically zero, and variations in it have a correspondingly small effect.
Another way to achieve a comparable result was advanced by Hawksford, and once again predates his particular cascode construction by many years, having been disclosed by Aldridge. With the latter's circuit the cascode device has its control electrode current recycled to the emitting electrode of the lower device. It has a slightly higher voltage burden than Boxall but has other advantages.
One issue with these schemes is that oscillation can occur if there is not a little lumped capacitance in the output, as at high frequencies the Boxall or Aldridge has a slightly negative output capacitance. This capacitance can still be very small to stabilize things.
There is a contribution from Walt Jung somewhere within this site that details the history of these and related circuits. I would stress that I am sure none of the re-inventors knew of Boxall or Aldridge, or other related inventors (Csanky did a version with JFETs). Boxall and Aldridge were working on telcom repeater amplifiers in the early days of transistors.
Brad
Thank you Brad for very detailed explanation. I've seen Walter Jung contribution.
In the two builds I see in this thread, both decided to connect gnd wire from output rca to the shunt star gnd.
I believe this is an opportunity to enlarge loop area because signal wire and gnd wire are soldered quite far apart in the board.
Can I solder the gnd from the output rca directly to the gnd shunt wire near the board outputs ?
You can experiment what you think to help. I did not have any hum or noise problem with the ground connections as I suggested.
Because the 'expansive" behavior of the CFA adds some dynamic during fast transients that compensate all the little losses everywhere else ? Did you found the usual character of the CFA (more details separation, more fluid trebles, easy listening) as a difference between the two versions ?I like non GNFB mode too and I m not sure why I use CFA mostly.
Brad, you're kind of a guy, as usual ;-)
Last edited:
Brad, you're kind of a guy, as usual ;-)
Well last time I checked, yes. 😀
Because the 'expansive" behavior of the CFA adds some dynamic during fast transients that compensate all the little losses everywhere else ? Did you found the usual character of the CFA (more details separation, more fluid trebles, easy listening) as a difference between the two versions ?
Brad, you're kind of a guy, as usual ;-)
If I could describe what I hear in that kind of vocabulary I would work for some audiophile magazine. For me an amp sounds good if it come close to the sound of the live classic music concert, and I would not use such words as expansive , fluid trebles, but yes easy listening I would.
My opinion is that in both technic CFA or VFA god amp is good amp. I don't have the CFA one ready and the VFA in the same level of power, the one I am building now 200W CFA is much more powerful then my 100W VFA and there is no fair comparison.
"Expansive" (© Richard Marsh) is a word reflecting the *technical* current behavior of an CFA, while VFA is "Compressive".... for some audiophile magazine....and I would not use such words as expansive

Other people talk about "Current on demand".
Last edited:
"Expansive" (© Richard Marsh) is a word reflecting the *technical* current behavior of an CFA, while VFA is "Compressive".
![]()
Other people talk about "Current on demand".
As well, there are topologies invented by oscilloscope designers for extending the range of differential inputs while preserving linearity. As far as I know they have not been applied to audio as yet. They are described in Dennis Feucht's books, such as his Handbook of Analog Circuit Design.
Brad
As well, there are topologies invented by oscilloscope designers for extending the range of differential inputs while preserving linearity. As far as I know they have not been applied to audio as yet. They are described in Dennis Feucht's books, such as his Handbook of Analog Circuit Design.
Brad
Pat Quinn's cascomp in Jim Williams book.
https://books.google.com/books?id=K...EwAA#v=onepage&q=cascomp jim williams&f=false
Pat Quinn's cascomp in Jim Williams book.
https://books.google.com/books?id=K...EwAA#v=onepage&q=cascomp jim williams&f=false
Yes, from John Addis' great contribution to that Williams book. I cited him some years ago as a circuit design great in some online poll, and it provoked him to actually acquire my unpublished phone number and call me with thanks! We've remained friends since, and he has given me a lot of inside information about things at Tektronix and thereafter, including some uses of base current recycling and how to stabilize such structures.
Another part of his article I cherish is his insistence that it is time to slay the Time-to-Market God. He uses the 2467 and its microchannel-plate-intensified CRT as a case in point.
Brad
First, quite crude impression is allready outstanding.
I am listening to Dave Brubeck's piano with a DL103 through my own riaa preamp into the gainwire ending with manley 300B monoblocks.
Gain is way too much (during build procedure I mixed the schematics and built the 7dB instead of the 3.5dB thing).
Dynamics are really impressive as is detail and overall tone. Everything is "in place" and with just the right amount of size.... outstanding.
I am using the No Global Negative feedback connection right now ... must listen a lot more to be objective, but I believe you got a real winner here 🙂
I am listening to Dave Brubeck's piano with a DL103 through my own riaa preamp into the gainwire ending with manley 300B monoblocks.
Gain is way too much (during build procedure I mixed the schematics and built the 7dB instead of the 3.5dB thing).
Dynamics are really impressive as is detail and overall tone. Everything is "in place" and with just the right amount of size.... outstanding.
I am using the No Global Negative feedback connection right now ... must listen a lot more to be objective, but I believe you got a real winner here 🙂
First, quite crude impression is allready outstanding.
I am listening to Dave Brubeck's piano with a DL103 through my own riaa preamp into the gainwire ending with manley 300B monoblocks.
Gain is way too much (during build procedure I mixed the schematics and built the 7dB instead of the 3.5dB thing).
Dynamics are really impressive as is detail and overall tone. Everything is "in place" and with just the right amount of size.... outstanding.
I am using the No Global Negative feedback connection right now ... must listen a lot more to be objective, but I believe you got a real winner here 🙂
Thank you Ricardo,
How about CFA mode? Do you hear any difference there, did you listen with a headphones?
I did not have the opportunity yet to test it with headphones.
Must finish some connections first.
After I have everything ready I will post my subjective opinion on both modes 🙂
Must finish some connections first.
After I have everything ready I will post my subjective opinion on both modes 🙂
Today I connected the gainwire to my singleton lateral mosfet amp "Assemblage" and was up to some serious revelations.
The Assemblage is quite fast and unforgiving so I could easily detect the differences between preamps.
After listening to the AQ2001 tube pre that I have in repair mode, I switched to my "old" Pedja buffer and got some improvements in speed and detail.... when I switched to the gainwire I was really impressed. It is much faster and more detailed than the previous ones but always musical.... not analytical at all.
Focusing on your design, I listened repeatedly to some reference recordings using NGNF and Current feedback and everytime I much preferred the NGNF. In fact the feedback tightens things up but immediately looses poise and high freq detail.
Using feedback this amp sounds slightly better than an opamp but not that much better IMO.
Feedback seems to erase fluidity and spatiality.
Without feedback this amp is something special. It has almost the same stage width than the DCB1 with more detail and slam.
The longer I listen to it the better it becomes..... Now I must really convert it to 3.5dB by reducing R17 to 1k5 but I have a doubt... should I also increase C2 to 220p ?
Why do we need to increase capacitance if having less gain ?
The Assemblage is quite fast and unforgiving so I could easily detect the differences between preamps.
After listening to the AQ2001 tube pre that I have in repair mode, I switched to my "old" Pedja buffer and got some improvements in speed and detail.... when I switched to the gainwire I was really impressed. It is much faster and more detailed than the previous ones but always musical.... not analytical at all.
Focusing on your design, I listened repeatedly to some reference recordings using NGNF and Current feedback and everytime I much preferred the NGNF. In fact the feedback tightens things up but immediately looses poise and high freq detail.
Using feedback this amp sounds slightly better than an opamp but not that much better IMO.
Feedback seems to erase fluidity and spatiality.
Without feedback this amp is something special. It has almost the same stage width than the DCB1 with more detail and slam.
The longer I listen to it the better it becomes..... Now I must really convert it to 3.5dB by reducing R17 to 1k5 but I have a doubt... should I also increase C2 to 220p ?
Why do we need to increase capacitance if having less gain ?
- Status
- Not open for further replies.
- Home
- Amplifiers
- Solid State
- No NFB line amp (GainWire mk2)