Hi Friends,
I'm currently finding a new way to design my new integrate amp with zero negative feedback.
I heard some audio brands were design their amp like this.
What do you think about this kind of amp?
Have you ever listen the sound from this kind of amp.
I'm only afraid the base quality, because of high amplitude signal might not so well in case of without feedback.
Please feel free to share your think here.
I'm currently finding a new way to design my new integrate amp with zero negative feedback.
I heard some audio brands were design their amp like this.
What do you think about this kind of amp?
Have you ever listen the sound from this kind of amp.
I'm only afraid the base quality, because of high amplitude signal might not so well in case of without feedback.
Please feel free to share your think here.
Hi Friends,
I'm currently finding a new way to design my new integrate amp with zero negative feedback.
I heard some audio brands were design their amp like this.
What do you think about this kind of amp?
Have you ever listen the sound from this kind of amp.
I'm only afraid the base quality, because of high amplitude signal might not so well in case of without feedback.
Please feel free to share your think here.
"Zero negative feedback" have several different implications and does not tell the real truth as a statement without clarifications...
I believe the "zero negative feedback" term normally are referred to a setup without LOOP feedback, but with a lot of local negative feedback.
The local feedback factor are determined by the electrical parameters of the active device (Bipolar, JFET, valve) and the emitter / source impedance in some configurations and you may have a negative feedback factor of 10, 100 or 1000 etc..
The tradeoff with high local negative feedback (and thus low distortion) are noise etc..
I have since the 70ties been using (my own designs) bipolar and JFET amplifiers with zero loop feedback.
Distortion below 0.005% and so low noise that you cannot hear it
Did a test in the early 80ties at a audio dealer:
Placed two identical Infinity loudspeaker sets side by side.
One set was not connected and one set connected to the amplifier with volume control set to maximum.
No music was played = silence.
All people did believe it was the un-connected speaker set that was connected to the amplifier as the noise level was highest.
The un-connected speaker set picked up room noise due to it was un-connected, but the connected speaker set was "shorted" by the noisefree and low impedant amplifier.
I have since the 70ties been using (my own designs) bipolar and JFET amplifiers with zero loop feedback.
Distortion below 0.005% and so low noise that you cannot hear it
Did a test in the early 80ties at a audio dealer:
Placed two identical Infinity loudspeaker sets side by side.
One set was not connected and one set connected to the amplifier with volume control set to maximum.
No music was played = silence.
All people did believe it was the un-connected speaker set that was connected to the amplifier as the noise level was highest.
The un-connected speaker set picked up room noise due to it was un-connected, but the connected speaker set was "shorted" by the noisefree and low impedant amplifier.
Oh Impressive.
Very good designed, interesting.
So what about feeling when you listen the music compare with NF design.
Have you published your designed somewhere?
I have built and listened to various GNFB and NoGNFB amps, and can share some observations.
The middle and high frequencies ranges are usually better with NoGNFB designs, especially if they have short signal path.
As for the bottom end, NoGNFB amps are a bit more difficult to match with speakers, irrespective of what output impedance they have. One gets impression of a bit not enough articulated bass. I guess, the reason can be in the preferences of speakers manufacturers, they usually use deep GNFB amps during speaker adjustments. Some kind of speakers will be perfect with NoGNFB amps, but there are not so much of them.
In the designs, having short signal path and low level and carefully arranged GNFB, the last does not make something bad for the sound, while the amp becomes more universal, it is easier matched to almost any speaker.
The middle and high frequencies ranges are usually better with NoGNFB designs, especially if they have short signal path.
As for the bottom end, NoGNFB amps are a bit more difficult to match with speakers, irrespective of what output impedance they have. One gets impression of a bit not enough articulated bass. I guess, the reason can be in the preferences of speakers manufacturers, they usually use deep GNFB amps during speaker adjustments. Some kind of speakers will be perfect with NoGNFB amps, but there are not so much of them.
In the designs, having short signal path and low level and carefully arranged GNFB, the last does not make something bad for the sound, while the amp becomes more universal, it is easier matched to almost any speaker.
In this issue, there is something more than simply the output impedance. My last low-level GNFB design, with paralleled jFETs at the output stage, in spite that has a bit higher Zout than NoGNFB disign, gives more articulated bass. It seems that NoGNFB designs, even with mass paralleled power BJTs, have some specific bass signature (I do not say that weak bass, just specific sound signature, a bit less sharply defined), that can be removed only by matching with definite kind of speakers.
Last edited:
Low level circuits are a whole new ball game as they are not driving 8 or 4 ohm speakers.
I tried mass paralleled power BJTs or mosfets as this is the route taken by designers such as Charles Hanson and recommended by John Curl, I ve come to agree with them that it helps.
Member
Joined 2009
Paid Member
Perhaps what would give a good combination is a bi-amping, where the mids+highs are without loop feedback and the bass is with loop feedback. This has been my thinking of late and I am planning a tube amplifier along these lines.
I do feel that the requirements of the mids+highs and bass are different, the speaker drivers are different with respect to their impedance and back emf, and the nature of music transients places some big demands on the bass that are not so common with the highs. Separate amplifiers seems to me a good route for DIY - gives us a reason to build more amplifiers
I do feel that the requirements of the mids+highs and bass are different, the speaker drivers are different with respect to their impedance and back emf, and the nature of music transients places some big demands on the bass that are not so common with the highs. Separate amplifiers seems to me a good route for DIY - gives us a reason to build more amplifiers
Perhaps what would give a good combination is a bi-amping, where the mids+highs are without loop feedback and the bass is with loop feedback. This has been my thinking of late and I am planning a tube amplifier along these lines.
I do feel that the requirements of the mids+highs and bass are different, the speaker drivers are different with respect to their impedance and back emf, and the nature of music transients places some big demands on the bass that are not so common with the highs. Separate amplifiers seems to me a good route for DIY - gives us a reason to build more amplifiers
Thats what Ive been doing lately.
My tests showed the same results. Through the reconnection of various jumpers to realize both or additional topologies of one power amplifier board we have the royal way. Who can prepare such an amplifier PCB ??
What do you think of my approach to have two(actually three) different amps, one with very high GNFB over 70dB at 20kHz and other with lower GNFB of 30dB at 20kHz?
Both are using the same PCB.
http://www.diyaudio.com/forums/solid-state/182554-thermaltrak-tmc-amp-3.html#post2778836
dado
Local feedback with minimum possible phase shift
is an entirely different animal from the global loop.
I don't think global loop is automatically a problem,
if the distance is short, and the coupling direct...
Thoughltlessly executed loops can stick, or ring, or
integrate unfixable errors, or slap the output against
the rails and slew rate like brick walls. But that's not
automatically true of every global loop.
After you are obligated to throw a dominant pole at
stability with less than 96KHz bandwidth, everything
is compromised.
A loop with no give that tries too hard to fix something
unfixable (like clipping), or remembers an error and tries
to fix it after the fact. These behaviors are unwanted.
is an entirely different animal from the global loop.
I don't think global loop is automatically a problem,
if the distance is short, and the coupling direct...
Thoughltlessly executed loops can stick, or ring, or
integrate unfixable errors, or slap the output against
the rails and slew rate like brick walls. But that's not
automatically true of every global loop.
After you are obligated to throw a dominant pole at
stability with less than 96KHz bandwidth, everything
is compromised.
A loop with no give that tries too hard to fix something
unfixable (like clipping), or remembers an error and tries
to fix it after the fact. These behaviors are unwanted.
Last edited:
>if the distance is short, and the coupling direct...
Which is the case of an emitter follower 'in optima forma'. Even this kind of (hidden) NFB might cause oscillations. NFB isn't bad. Incompetent designers have given it a bad reputation. It's all a matter of sufficient phase margin (and of course sufficient bandwidth).
Which is the case of an emitter follower 'in optima forma'. Even this kind of (hidden) NFB might cause oscillations. NFB isn't bad. Incompetent designers have given it a bad reputation. It's all a matter of sufficient phase margin (and of course sufficient bandwidth).
I have a great deal of respect for Charles Hansen's audio design philosophy for Ayre products.
1) Amplifiers with only local feedback sound different than amplifiers with global feedback. Charles believes they sound more musical.
--Especially if JFETS are used for the input stage and voltage gain stage.
2) High output dampening on a non-global-feedback amplifier is desirable.
--Can be achievied with a large number of output transistors.
--Charles prefers using more parallel outputs over the Hawksford type EC.
3) If you are going to use global feedback, it is best to use a lot of feedback.
---Global feedback generates higher harmonics, especially odd harmonics, and high global feedback is required to substantially push their absolute value down.
Charles Hansen from Stereophile interview:
"I have some beliefs that approach religious faith: FETs are better than bipolars, zero feedback is better than feedback, balanced is better than single-ended, and the simpler a circuit, the better. But here's the thing: Just because I believe something doesn't make it true, and designing the MX-R taught me some lessons.
"The first is that, in parts of the circuit, bipolar devices sound better than FETs, so that's what we used. (the Arye MX-R uses 16 ThermaTrak bipolar output transistors). And then there's the EquiLock circuit, which violates my 'simpler is better' dictum. EquiLock is kind of like creating a cascode by combining two triodes. You could say we're joining together two transistors to act like one transistor that has a really stable operating point. Adding a second transistor to the signal path seems like it deviates from my belief that simpler is better, except that it works better."
Key to Ayre's development is what they call EquiLock circuitry for the MX-R monoblock. "In a conventional circuit, the gain transistor has a load, usually either a resistor or a current source," Hansen said. "When the current through the gain transistor changes, then the voltage across the load also changes, which, in turn, means that the voltage across the gain transistor is changing. In fact, all of the parameters (transconductance, capacitance, etc.) vary when the voltage across the transistor varies.
"The EquiLock circuit adds another transistor between the gain transistor and the load. (In our case, the load is actually a current mirror.) This extra transistor holds the voltage of the gain transistor at a fixed level while still transmitting the changes in current to the load (the current mirror). By stabilizing the voltage across the gain transistor, all of the parameters of the gain transistor are also stabilized. The circuit is very similar to a cascode circuit, which has been used by other manufacturers, but EquiLock is an improvement over a conventional cascode circuit."
1) Amplifiers with only local feedback sound different than amplifiers with global feedback. Charles believes they sound more musical.
--Especially if JFETS are used for the input stage and voltage gain stage.
2) High output dampening on a non-global-feedback amplifier is desirable.
--Can be achievied with a large number of output transistors.
--Charles prefers using more parallel outputs over the Hawksford type EC.
3) If you are going to use global feedback, it is best to use a lot of feedback.
---Global feedback generates higher harmonics, especially odd harmonics, and high global feedback is required to substantially push their absolute value down.
Charles Hansen from Stereophile interview:
"I have some beliefs that approach religious faith: FETs are better than bipolars, zero feedback is better than feedback, balanced is better than single-ended, and the simpler a circuit, the better. But here's the thing: Just because I believe something doesn't make it true, and designing the MX-R taught me some lessons.
"The first is that, in parts of the circuit, bipolar devices sound better than FETs, so that's what we used. (the Arye MX-R uses 16 ThermaTrak bipolar output transistors). And then there's the EquiLock circuit, which violates my 'simpler is better' dictum. EquiLock is kind of like creating a cascode by combining two triodes. You could say we're joining together two transistors to act like one transistor that has a really stable operating point. Adding a second transistor to the signal path seems like it deviates from my belief that simpler is better, except that it works better."
Key to Ayre's development is what they call EquiLock circuitry for the MX-R monoblock. "In a conventional circuit, the gain transistor has a load, usually either a resistor or a current source," Hansen said. "When the current through the gain transistor changes, then the voltage across the load also changes, which, in turn, means that the voltage across the gain transistor is changing. In fact, all of the parameters (transconductance, capacitance, etc.) vary when the voltage across the transistor varies.
"The EquiLock circuit adds another transistor between the gain transistor and the load. (In our case, the load is actually a current mirror.) This extra transistor holds the voltage of the gain transistor at a fixed level while still transmitting the changes in current to the load (the current mirror). By stabilizing the voltage across the gain transistor, all of the parameters of the gain transistor are also stabilized. The circuit is very similar to a cascode circuit, which has been used by other manufacturers, but EquiLock is an improvement over a conventional cascode circuit."
Attachments
Edmond I completely agree with you. Just I never had a chance to compare high GNFG to the low GNFG amps. All my amps were high GNFG type. Here I sow possibility to use the same PCB layout and try both types with very similar circuit configuration. You can't deny that some people think low GNFG sounds better. Did you ever had a chance to compare?
Damir
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.