Thank you Circlomanen for the above posts, the design variations they show, and your explanations. Regarding the schematic of post#12, I thought the amp's 0.03% THD, and its better subjective performance may be further improved if the drain of the IRF9240 is at more negative voltage than common. With this proposed change, M1 now looks to me like a normally-biased amplifier. A 1 Vp-p ac voltage at its gate can now swing above and below ground, and thus M1 functions as a voltage variable current source. Still, your current results without this additional change are highly interesting.
http://www.firstwatt.com/pdf/art_zv5.pdf
Here is the complementary PCF Zen amp (a positive current feedback version of the Zen V5).
It simulates really good.
I like the combination of a single gainstage (I consider cascodes single gain stages) and positive current feedback.
Something for you all to play with.
I wish I had some complementary SIT devices for this......
Cheers,
Johannes
With 0,75 ohm sensing resistor (R13) the amp seems to have a very pronounced negative output impedance. It will give a 22,34 volt output into 8 ohm, 28,23 volts into 6 ohm and 40,6 volts into 4 ohm but with some visible rounding of the wave shape (almost clipping).
Not bad for 25+25 volts power supplies, one single gainstage and 1,5 amps of quiescent current.
This is not a final and optimized amp. It is just an example of what can be done with positive current feedback and a simple singe gain stage.
I would use a much smaller sensing resistor and less negative feedback.
Cheers,
Johannes
Not bad for 25+25 volts power supplies, one single gainstage and 1,5 amps of quiescent current.
This is not a final and optimized amp. It is just an example of what can be done with positive current feedback and a simple singe gain stage.
I would use a much smaller sensing resistor and less negative feedback.
Cheers,
Johannes
I have been exploring the different tonal character and general character of the amp driving my large BIB speaker with different ratios of negative voltage feedback to positive current feedback and the resulting output impedance and its effect on the speaker.
It is a very fun and useful tool to tune the amp-speaker cooperation. I can clearly hear when it starts exhibiting a negative output impedance. The gain starts to skyrocket and the midrange starts to shout because of the impedance dip in the crossover region of the bass driver and my Faital Pro 3Fe22 full range drivers. If pushed any more then that it starts to distort. I guess it is near to oscillation at that point.
With 500 ohm/10 kohm resistor negative feedback and 0,83 ohms sensing resistor for the positive feedback the amp sounds very large and alive. Not necessarily most neutral and clean, but very large and alive. The combination of the quite brutal IRLB3813 gain device (140 S transconductans and a hint of triod character at low currents) and positive feedback seems to turn the amp into a raging bull fueled by steroids and amphetamin. Once I get the sensing resistor down to about 0,6 ohms the amp sounds very clean and clear. Very dynamic and with a great control of the bass.
If I short out the positive feedback it just collapses into a very boring and dull lifeless electrical representation of recorded music. Nothing I would like to listen to.
I really want to urge everyone to test some positive current feedback. It is a great tool Nelson Pass has given us. I know it is not a new thing at all, but I have never seen it been used as a flavor or spice in simple class A amps before, just as a clever way to tune the output impedance towards infinity (or slightly beyond) and add some much needed gain, without resorting to multiple gain stages and hideous negative voltage feedback.
It is like Christmas here. I have been given a new tool to play with, and it seems that this tool is very powerful and has the capability of totally transform the way to design amps.
Cheers,
Johannes.
It is a very fun and useful tool to tune the amp-speaker cooperation. I can clearly hear when it starts exhibiting a negative output impedance. The gain starts to skyrocket and the midrange starts to shout because of the impedance dip in the crossover region of the bass driver and my Faital Pro 3Fe22 full range drivers. If pushed any more then that it starts to distort. I guess it is near to oscillation at that point.
With 500 ohm/10 kohm resistor negative feedback and 0,83 ohms sensing resistor for the positive feedback the amp sounds very large and alive. Not necessarily most neutral and clean, but very large and alive. The combination of the quite brutal IRLB3813 gain device (140 S transconductans and a hint of triod character at low currents) and positive feedback seems to turn the amp into a raging bull fueled by steroids and amphetamin. Once I get the sensing resistor down to about 0,6 ohms the amp sounds very clean and clear. Very dynamic and with a great control of the bass.
If I short out the positive feedback it just collapses into a very boring and dull lifeless electrical representation of recorded music. Nothing I would like to listen to.
I really want to urge everyone to test some positive current feedback. It is a great tool Nelson Pass has given us. I know it is not a new thing at all, but I have never seen it been used as a flavor or spice in simple class A amps before, just as a clever way to tune the output impedance towards infinity (or slightly beyond) and add some much needed gain, without resorting to multiple gain stages and hideous negative voltage feedback.
It is like Christmas here. I have been given a new tool to play with, and it seems that this tool is very powerful and has the capability of totally transform the way to design amps.
Cheers,
Johannes.
I am sorry for my rant here, but I am simply blown away by sheer size and importance of this all.
I am starting to zero in on a good ratio of NFB and PCF, and I am very surprised by the timbre and lifelike sense of scale presented for the first time from my BIB. Sure, they can sound large and intimidating with an effortless bass any owner of large high efficiency horns can recognize. But this is something else........
I am playing Ahmad Jamal Live at the Olympia and the very lifelike effortless sense of scale and natural timbre is astonishing. This is single ended class A amps at their absolute best, bar non..... Grand piano, cymbals and contrabass is effortlessly presented with a very lifelike presence.
If the positive feedback is made adjustable by a knob on the front of the amp, the knob should be marked Timbre and effortless sense of scale. With too little positive feedback everything sounds a little subdued and held back. Very clean, correct and polished, but a little small and un-engaging. With a little to much positive feedback everything sounds a little larger then life and very very intensely present. It is like an adjustable LSD trip to increase the positive feedback. Everything becomes very acutely "alive" and present, but a little to large and to in your face. It does not feel natural (if you are not on acid that is...).
I don't know how to explain this to all of you. I guess Papa is smiling with recognition when reading my explanation of the sonic attributes of positive feedback and simple class A amps....
A SIT in an amp like this...........
Papa!!! 😉😉
I am starting to zero in on a good ratio of NFB and PCF, and I am very surprised by the timbre and lifelike sense of scale presented for the first time from my BIB. Sure, they can sound large and intimidating with an effortless bass any owner of large high efficiency horns can recognize. But this is something else........
I am playing Ahmad Jamal Live at the Olympia and the very lifelike effortless sense of scale and natural timbre is astonishing. This is single ended class A amps at their absolute best, bar non..... Grand piano, cymbals and contrabass is effortlessly presented with a very lifelike presence.
If the positive feedback is made adjustable by a knob on the front of the amp, the knob should be marked Timbre and effortless sense of scale. With too little positive feedback everything sounds a little subdued and held back. Very clean, correct and polished, but a little small and un-engaging. With a little to much positive feedback everything sounds a little larger then life and very very intensely present. It is like an adjustable LSD trip to increase the positive feedback. Everything becomes very acutely "alive" and present, but a little to large and to in your face. It does not feel natural (if you are not on acid that is...).
I don't know how to explain this to all of you. I guess Papa is smiling with recognition when reading my explanation of the sonic attributes of positive feedback and simple class A amps....
A SIT in an amp like this...........
Papa!!! 😉😉
Last edited:
I am no designer or engineer. I will never publish a "final design" since this is just experiments I do because I love learning and playing with stuff.
This amp is not inherently great in any setup. The great thing is that positive current feedback is a great tool in the design of amps.
The reason I dare to write "bar non" above is that with the amp adjusted and optimized for my exact needs, preferences, loudspeakers etc, it does sound astonishingly great. This takes a couple of days of experiments, adjustments of the amp, a lot of listening and a lot of general meditation on the tradeofs and choices.
I know it can be improved, and probably a lot.
I will build a better amp with less rats nest/alligator clip leads design.
I just want to encourage you all to experiment with this tool.
I post this just to share my thoughts. I have learned a lot from all of you on this forum, and I hope I can contribute something back.
There is so much I would like to write about this amp, but I can't stop listening to music while my kids are at school. (They hate jazz).
Cheers,
Johannes
This is the amp as it is playing right now. This is not a simulation. I just used LTSpice to draw the schematic.
Remember that this is only what sounds really good on my speakers, in my room, with my crossovers and cables, according to my taste.
I have heard a lot of hifi during my 30 years of building speakers and amps, but this is something new to me....
I can not guarantee that this will sound good with any other devices, speakers or anything else.
When you use positive current feedback you make the amp somewhat "aware" of the speakers in a new way. This is why I urge some caution in using my exact schematic.
This complicates the interface between the amp and the speakers. Small changes can make a huge difference.
If I increase the positive feedback the loudspeaker gets a metallic "twang" when I dab my finger on the speaker. I guess it is an early onset of oscillation, exited by the sudden transient signal being fed into the output. I don't use that amount of positive feedback as it sounds horrible. It is just an example of the interface between the amp and the speaker.
Please see it as a basic recipe for your own experiments.
This is just an experiment, and not really the way I would recommend anyone to build an amp.
I used pretty much whatever I found in my scrapbox. It is old used parts from other projects.
I would use more then 15,55 volts if I could, but my power supply does not give more then 15,55 volts. 750 mA is the most I can use without overheating the power supply.
The IRLB3813 is a 30 volt part, so that limits the amount of power you can get out of this amp.
Cheers,
Johannes
Last edited:
I must add that very small adjustments of the sensing resistor (R6) makes a lot of difference to the character of the amp.
I have decreased the resistance from 0,825 to 0,817, and that made the difference I described above (LSD). I have been paralleling 100 ohm resistors over the 1,0 resistor.
A very minute change in the feedback will take this amp from hideous to wonderful, so you must have some patience. With 1,0 ohm sensing resistor it sounds horrible, with lots of distortion and a shouting midrange like a bad Lowther fullrangespeaker driven by a cheap solid state amp.
C2 is two 4700 uF 40 volts electrolytic capacitors with some smaller film capacitors thrown in. I don't know the exact total value, but it is close to 10000 uF.
Cheers,
Johannes
I have decreased the resistance from 0,825 to 0,817, and that made the difference I described above (LSD). I have been paralleling 100 ohm resistors over the 1,0 resistor.
A very minute change in the feedback will take this amp from hideous to wonderful, so you must have some patience. With 1,0 ohm sensing resistor it sounds horrible, with lots of distortion and a shouting midrange like a bad Lowther fullrangespeaker driven by a cheap solid state amp.
C2 is two 4700 uF 40 volts electrolytic capacitors with some smaller film capacitors thrown in. I don't know the exact total value, but it is close to 10000 uF.
Cheers,
Johannes
I want to share some of my ideas and my reasoning behind this amp.
First it is the use of two dissimilar devices. the use of one P-channel and on N-channel in a "sort of" cascode will tend to some extent to cancel out the distortion from each other.
Nelson Pass writes about this is his "Penultimate Zen" article.
One device does all the current gain and the other device all the voltage gain. It is similar to not having your welder also paint while he welds. If you separate the work, so the painter does the painting and the welder does the welding, you should get much better results, since each one does what he is best at. I know that even though I can paint and I know how to weld, the results would be horrible if I tried to do both things at the same time.
I don't want two separate gainstages though. A cascode is as long as I want to stretch it.
A common drain is much better at amplifying current with low distortion and large bandwidth and the common gate is much more linear and much faster with less phase-shifts when amplifying voltage, then the common source.
I guess my amp would love to broadcast AM transmissions if we used less gatestopper resistance.
When using positive feedback I consider it very important to linearize the feedback loop as much as possible, since any distortion, extra inductance and capacitance in the feedback loop will create an oscillator given any excuse to, regardless of how minute the excuse is.
Positive feedback is a very powerful concept. It tends to bring out the best (and the worst if given a reason to) of anything it comes into contact with.
Imagine a musician performing in front of a crowd. If the crowd is dead silent (no feedback) the gig will never be very entertaining.
If the crowd is enthusiastic and cheering the musician he will probably tend to push himself a little harder and play a little better (this would be positive feedback). If the crowd starts to point out every little mistake he did regardless of how minute and unimportant the mistake was, he would star focusing more on avoiding mistakes than actually entertain the crowd, the gig would be very technically correct but totally devoid of any joy - this would be analog to negative feedback.
Fascism is a social form of negative feedback. It creates resignation and apathy.
Encouragement and a positive collaboration is a social form of positive feedback. It tends to bring out the best from people.
Cheers,
Johannes
First it is the use of two dissimilar devices. the use of one P-channel and on N-channel in a "sort of" cascode will tend to some extent to cancel out the distortion from each other.
Nelson Pass writes about this is his "Penultimate Zen" article.
One device does all the current gain and the other device all the voltage gain. It is similar to not having your welder also paint while he welds. If you separate the work, so the painter does the painting and the welder does the welding, you should get much better results, since each one does what he is best at. I know that even though I can paint and I know how to weld, the results would be horrible if I tried to do both things at the same time.
I don't want two separate gainstages though. A cascode is as long as I want to stretch it.
A common drain is much better at amplifying current with low distortion and large bandwidth and the common gate is much more linear and much faster with less phase-shifts when amplifying voltage, then the common source.
I guess my amp would love to broadcast AM transmissions if we used less gatestopper resistance.
When using positive feedback I consider it very important to linearize the feedback loop as much as possible, since any distortion, extra inductance and capacitance in the feedback loop will create an oscillator given any excuse to, regardless of how minute the excuse is.
Positive feedback is a very powerful concept. It tends to bring out the best (and the worst if given a reason to) of anything it comes into contact with.
Imagine a musician performing in front of a crowd. If the crowd is dead silent (no feedback) the gig will never be very entertaining.
If the crowd is enthusiastic and cheering the musician he will probably tend to push himself a little harder and play a little better (this would be positive feedback). If the crowd starts to point out every little mistake he did regardless of how minute and unimportant the mistake was, he would star focusing more on avoiding mistakes than actually entertain the crowd, the gig would be very technically correct but totally devoid of any joy - this would be analog to negative feedback.
Fascism is a social form of negative feedback. It creates resignation and apathy.
Encouragement and a positive collaboration is a social form of positive feedback. It tends to bring out the best from people.
Cheers,
Johannes
Last edited:
Hi Johannes,
Thank you for sharing this with us. Looks like you had a lot of fun doing it.
Cheers,
Dennis
Thank you for sharing this with us. Looks like you had a lot of fun doing it.
Cheers,
Dennis
I must clarify this statement. With the right balance between positive and negative feedback the amp becomes "load invariant" as I think Nelson Pass write.
If I decrease the PCF it will be much more neutral and clean. I have increased the PCF to suit my taste and to work well with my speakers. With more PCF it will start to react to what the speakers and cables do (or do not do). Nelson Pass builds (semi) commercial amps that have to sound good with many different speakers and listers. I guess (based on my experience so far) that he has consciously moderated the amp a little to become more all around and well balanced. Remember that this is small adjustments in how you choose your load line (load line cancellation) and how you voice your amp.
I use an IRLB3813 so it should be obvious from start that I prefer a more aggressive and powerful reproduction. It is similar to preferring a Dodge Viper V10 to a Porsche. Both are great cars, but in very different ways.
Thanks! Yes, it is very fun developing amps and speakers. I love to gorge in creativity and experiments.
😀
Cheers,
Johannes
I am in the "post project analysis" phase, so you all have to excuse my steady stream of posts, thoughts and ideas...
Part of the success of my amp is found in Nelson Pass article The Sweet Spot.
http://www.firstwatt.com/pdf/art_sweet_spot.pdf
It is a great article (as usual from Nelson Pass).
The sweet spot is applicable to both electronics and speakers.
My large BIB back loaded horns are very bass heavy, and with an adjustable PCF and output impedance I use the amp to voice the speakers to my room and taste. The B&C 10PS26 bassdrivers I use in my BIBs lacks some Bl and has a little to high Qes. This is adjusted with the slightly negative output of the amp. I guess based on the sound quality and theory of using a negative output impedance that this also lowers distortion quit a bit.
This is partly why I want to encourage everyone to experiment with this technology, and it is why I am a bit hesitant to publish a final design. There can never be a final design from this technology, since it is always pertaining to the speakers, crossovers, cables, room, acoustics, taste etc.
A small amount of second order distortion with the right balance can to some extent cancel some of the second order distortion in a speaker. There is a lot of this sweetspots to seek, balance and perfect.
This is where the PCF and an widely adjustable output impedance is such a great tool and asset.
To be able to combine PCF with simple single stage low distortion, low feedback, class A amps is such a marriage in heaven that I am jumping with joy... It is like we all of a sudden could by Papas SiC SITs perfectly matched in both N and P channel devices for a few bucks each...
Cheers,
Johannes
Part of the success of my amp is found in Nelson Pass article The Sweet Spot.
http://www.firstwatt.com/pdf/art_sweet_spot.pdf
It is a great article (as usual from Nelson Pass).
The sweet spot is applicable to both electronics and speakers.
My large BIB back loaded horns are very bass heavy, and with an adjustable PCF and output impedance I use the amp to voice the speakers to my room and taste. The B&C 10PS26 bassdrivers I use in my BIBs lacks some Bl and has a little to high Qes. This is adjusted with the slightly negative output of the amp. I guess based on the sound quality and theory of using a negative output impedance that this also lowers distortion quit a bit.
This is partly why I want to encourage everyone to experiment with this technology, and it is why I am a bit hesitant to publish a final design. There can never be a final design from this technology, since it is always pertaining to the speakers, crossovers, cables, room, acoustics, taste etc.
A small amount of second order distortion with the right balance can to some extent cancel some of the second order distortion in a speaker. There is a lot of this sweetspots to seek, balance and perfect.
This is where the PCF and an widely adjustable output impedance is such a great tool and asset.
To be able to combine PCF with simple single stage low distortion, low feedback, class A amps is such a marriage in heaven that I am jumping with joy... It is like we all of a sudden could by Papas SiC SITs perfectly matched in both N and P channel devices for a few bucks each...
Cheers,
Johannes
Thanks TordNilsson!
The magic is gone!!! 😡
I have been testing the amp with different amounts of quiescent current, different amounts of feedback and different voltages, back and forth to test the limits and learn more about how it reacts to different operating conditions.
I can not however find my way back to the total magic I experienced the first day, which makes me think that it was more about my very very low expectations being surpassed by an rather ridiculous level of sound quality considering the extreme crudeness and puny power the amp has.
I don't listen to more the 2 volts peak over the speaker terminals and considering the 8 ohm nominal load it is only half a watt peak. My BIBs are very efficient.
It does however sound very good, and even more so considering the extreme crudeness, non optimal parts and low power. I have never heard my BIBs sound so exact, controlled, dynamic, clean and "big" before. One could not guess it is a 1 watt amp driving them. The sense of scale, the effortlessness and the natural reproduction of the timbre is still the best I have heard from my BIBs. I am still very impressed and very surprised by what is attainable with such a simple circuit and some careful tuning.
The IRFL3813 is a rather special part.
Most mosfets starts to loose transconductance below 1 volt, becoming very nonlinear. The IRLB3813 seems to function properly down to 0,1 volts d-s.
This is very important in an amp with only 15,55 volts power supplies.
Yesterday I bought a 520 x 460 x 15 mm piece of aluminum ( cheap heatsink) and some powerful clamps.
I will soon start to build a better amp with more voltage and current and what I hope is better and more suitable mosfets.
I still want to encourage everyone to play with the balance between a small amount of negative feedback, some positive current feedback in a simple single stage class A amp.
Considering the low power and crude build, the amp should not be able to do what it does. I am certain there is lots more to gain from some PCF in a suitable circuit.
Cheers,
Johannes
The magic is gone!!! 😡
I have been testing the amp with different amounts of quiescent current, different amounts of feedback and different voltages, back and forth to test the limits and learn more about how it reacts to different operating conditions.
I can not however find my way back to the total magic I experienced the first day, which makes me think that it was more about my very very low expectations being surpassed by an rather ridiculous level of sound quality considering the extreme crudeness and puny power the amp has.
I don't listen to more the 2 volts peak over the speaker terminals and considering the 8 ohm nominal load it is only half a watt peak. My BIBs are very efficient.
It does however sound very good, and even more so considering the extreme crudeness, non optimal parts and low power. I have never heard my BIBs sound so exact, controlled, dynamic, clean and "big" before. One could not guess it is a 1 watt amp driving them. The sense of scale, the effortlessness and the natural reproduction of the timbre is still the best I have heard from my BIBs. I am still very impressed and very surprised by what is attainable with such a simple circuit and some careful tuning.
The IRFL3813 is a rather special part.
Most mosfets starts to loose transconductance below 1 volt, becoming very nonlinear. The IRLB3813 seems to function properly down to 0,1 volts d-s.
This is very important in an amp with only 15,55 volts power supplies.
Yesterday I bought a 520 x 460 x 15 mm piece of aluminum ( cheap heatsink) and some powerful clamps.
I will soon start to build a better amp with more voltage and current and what I hope is better and more suitable mosfets.
I still want to encourage everyone to play with the balance between a small amount of negative feedback, some positive current feedback in a simple single stage class A amp.
Considering the low power and crude build, the amp should not be able to do what it does. I am certain there is lots more to gain from some PCF in a suitable circuit.
Cheers,
Johannes
I just came up with a with a simple explanation for the function of the bottom P-channel fet, the one that senses the current flowing through the speaker and couples that into the gain device source.
In the same way Nelson Pass seems to be trying to avoid using a source resistor to maximize the square law character of the fets. The bottom P-channel fet can be seen as a load-dependent negative source resistor, greatly increasing gain. It is literately like turbocharging the fet.
When used for positive current feedback it will create a negative source impedance (resistor) as a function of the current flowing through the speaker.
Based on the lack of interest in this thread, I guess you all understand this already, but for those not well versed in positive current feedback and negative output impedances it should be a quite simple explanation of the function of my simple circuit.
Cheers,
Johannes
In the same way Nelson Pass seems to be trying to avoid using a source resistor to maximize the square law character of the fets. The bottom P-channel fet can be seen as a load-dependent negative source resistor, greatly increasing gain. It is literately like turbocharging the fet.
When used for positive current feedback it will create a negative source impedance (resistor) as a function of the current flowing through the speaker.
Based on the lack of interest in this thread, I guess you all understand this already, but for those not well versed in positive current feedback and negative output impedances it should be a quite simple explanation of the function of my simple circuit.
Cheers,
Johannes
Last edited: