Hi Charles,
Indeed. I doubt that emitter (or source) follower character would be considered bad design just a discrepancy to be rendered insignificant. There are a number of ways to do this -
Complex output stages that linearise the transition.
Class A - wasteful of heatsink capacity and power. inelegant.
Error correction again complex and critical.
NFB simple effective and elegant. Borrowed from nature.
In a multistage amplifier you would need to separately and effectively (read with extra cost/complexity) linearise each and every stage to a similar standard to meet the error budget, albeit usually using local NFB.
Whereas with global NFB one can have 3 stages each with 1% THD apply 40 dB NFB and have 0.01%.
Then there are issues of output damping factor and PSRR that are poorly addressed without great complexity, without global NFB. There appears no good reason NOT to use global NFB so why?
Please enlighten me.
Cheers,
Greg
Indeed. I doubt that emitter (or source) follower character would be considered bad design just a discrepancy to be rendered insignificant. There are a number of ways to do this -
Complex output stages that linearise the transition.
Class A - wasteful of heatsink capacity and power. inelegant.
Error correction again complex and critical.
NFB simple effective and elegant. Borrowed from nature.
In a multistage amplifier you would need to separately and effectively (read with extra cost/complexity) linearise each and every stage to a similar standard to meet the error budget, albeit usually using local NFB.
Whereas with global NFB one can have 3 stages each with 1% THD apply 40 dB NFB and have 0.01%.
Then there are issues of output damping factor and PSRR that are poorly addressed without great complexity, without global NFB. There appears no good reason NOT to use global NFB so why?
Please enlighten me.
Cheers,
Greg
amplifierguru said:
To many (including myself), a circuit with negative feedback sounds "electronic" and "artificial". That is an excellent reason not to use feedback.
I am also aware that some people prefer to listen to equipment that uses feedback. (I suppose that there is no accounting for taste.) So it is good that there are companies that build products to satisfy these customers.
amplifierguru said:
There are other ways also.
amplifierguru said:
Yes, if your goal is to minimize steady-state distortion of a test tone.
amplifierguru said:
This is incorrect. The output impedance of an emitter follower is roughly 26 divided by the idle current in mA. It doesn't take much idle current to create a damping factor of 50 or more without using feedback, and without "great complexity". (Going much beyond a damping factor of 50 is pretty much a waste of time due to the impedance of the speaker cables and/or crossover components.)
Adding feedback gives you 3 things:
- Lower distortion.
- Wider bandwidth.
- Lower output impedance.
If that's all there were to the subject of amplifier design, we could all retire now, as Halcro has already made the ultimate amplifier. But in the real world, there are plenty of people that don't like the sound of the Halcro. Why? Because of the colorations introduced by feedback.
Another approach is to build a circuit with low distortion, wide bandwidth, and low output impedance without resorting to the use of feedback. It can be done, and without resorting to circuits of "great complexity". Try it some time and see what you think of the sonic results.
Hi Charles
i see you're still peddling your particular brand of "no feedback" bs ( bs stands for "back story" of course )
in a world of amplifying devices that are well modeled as controlled current sources there is no practical means of obtaining low output impedance without negative feedback
you seem to have a strong marketing interest in denying the existence of local negative feedback in emitter/source followers and seem to think degeneration doesn’t count as local negative feedback either – not a position that gains any respect with competent engineers however well it plays with your customers
care to show us the means used “build a circuit with low distortion, wide bandwidth, and low output impedance without resorting to the use of feedback” ?
(I suggest 2 “other” means beside negative feedback are well known and used extensively by all amplifier designers: even order distortion cancellation in diff pair/push-pull circuits and increasing bias to signal ratio in a stage – generalized predistortion linearization is not “simple” and not robust without negative feedback to stabilize operating point)
as for subjective evaluation you again come to the self serving conclusion about feedback in the face of at least sufficient evidence to consider an alternative:
“because some people like the colorations and distortions that are difficult to remove without global negative feedback”
tellingly: “Going much beyond a damping factor of 50 is pretty much a waste of time …” allows amplifier/loudspeaker interaction response differences over the 0.1 dB established as a just noticeable difference in double blind testing – amplifiers designed with damping factors this low do add audible amplifier coloration to the sound
i see you're still peddling your particular brand of "no feedback" bs ( bs stands for "back story" of course )
in a world of amplifying devices that are well modeled as controlled current sources there is no practical means of obtaining low output impedance without negative feedback
you seem to have a strong marketing interest in denying the existence of local negative feedback in emitter/source followers and seem to think degeneration doesn’t count as local negative feedback either – not a position that gains any respect with competent engineers however well it plays with your customers
care to show us the means used “build a circuit with low distortion, wide bandwidth, and low output impedance without resorting to the use of feedback” ?
(I suggest 2 “other” means beside negative feedback are well known and used extensively by all amplifier designers: even order distortion cancellation in diff pair/push-pull circuits and increasing bias to signal ratio in a stage – generalized predistortion linearization is not “simple” and not robust without negative feedback to stabilize operating point)
as for subjective evaluation you again come to the self serving conclusion about feedback in the face of at least sufficient evidence to consider an alternative:
“because some people like the colorations and distortions that are difficult to remove without global negative feedback”
tellingly: “Going much beyond a damping factor of 50 is pretty much a waste of time …” allows amplifier/loudspeaker interaction response differences over the 0.1 dB established as a just noticeable difference in double blind testing – amplifiers designed with damping factors this low do add audible amplifier coloration to the sound
jcx said:
For the benefit of Ampguru, I will condense an on-going argument that has wasted quite a bit of time of those who read this forum:
On one side are those who say that local degeneration (as found, for example, in an emitter follower) is a form of feedback. People on this side say that even with a common-emitter circuit with *no* emitter resistor, there is still "feedback" due to the intrinsic emitter resistance. According to this side, there is no such thing as a "zero feedback" circuit.
On the other side are people that recognize that a word may have a slightly different meaning depending on the context in which it's used, and that some common sense should be applied when using words. (For example, my 8-year old son tells me that there is no such thing as black and white, only shades of grey. Of course, he is literally correct, and at the same time we all know what is meant when I tell my neighbor that I bought a "black" car.) When speaking of complete amplifier circuits, it is generally recognized that there are three broad categories:
a) Circuits with an overall feedback loop that compares the output signal with the input signal; eg, nearly all modern audio amplifiers.
b) Circuits with shorter feedback loops that do not contain the entire circuit;, eg, "error correction" topologies and complementary feedback pairs.
c) Circuits with no feedback loop at all, only local degeneration; eg, an emitter-follower.
This last category is typically referred to as an "open loop" or "zero feedback" design, not only by most prominent audio designers, but also by major companies such as Analog Devices, Burr-Brown, and Maxim. As noted in previous threads, I am unwilling to discuss the semantics of this topic any further. I am summarizing it here solely for the edification of AmpGuru. I am pretty sure that he knows what I mean when I say a "zero feedback" amp, even if JCX doesn't.
As far as a "marketing interest" goes, I really don't make much money sharing design techniques here on DIY Audio.
jcx said:
That's fine. If you believe that a damping factor of 100 or 200 (or whatever) is required to achieve correct amplifier performance, then that is what you should design for. It's simple enough to do without resorting to the use of feedback. As I suggested to AmpGuru, why don't you build two amplifiers that achieve your performance goal, one with feedback and one without. Then listen to them and see what you think.
If you can't tell the difference, or you prefer the one with feedback, that's fine by me. I can tell the difference and prefer the sound without feedback. Many people agree with me. The reason I bring this topic up at all is to introduce new ideas and new possibilities to the DIY enthusiast. Several people have tried these ideas and reported positive results here on DIY Audio.
Term 'zero feedback' definitely needs definig or re-defininig (meybe by IEEE 
). You know what I feel when I look at (quotation) 'COMPLETELY NO FEED BACK EVEN LOCALLY' like this ?
BTW in my opinion we'd better stick to what thread starter asks, I mean if he wants to build a no fb amp than help him to build a no fb amp even if you see no point in it. Tolerance? Open-mind?
regards
). You know what I feel when I look at (quotation) 'COMPLETELY NO FEED BACK EVEN LOCALLY' like this ?BTW in my opinion we'd better stick to what thread starter asks, I mean if he wants to build a no fb amp than help him to build a no fb amp even if you see no point in it. Tolerance? Open-mind?
regards
Charles Hansen said:
The output impedance of an emitter follower is the output impedance of the previous stage divided by the instantaneous current gain, plus that 26/Ie component due to Vbe changes.
In practice, several stages are required in order to go from the classic 47Kohm input impedance to the said 0.16 ohms 50x damping factor output.
PS: I's amazing to see how much global feedback loops operate inside the human body, and they work fine... I'm sure that our auditive system has plenty of them. Sometimes they are even unstable (for example: female hormonal cycles involve a negative feedback loop that suddenly turns into positive feedback for several days, until the system saturates causing ovulation, then it goes back to negative feedback). Global feedback is inspired by nature.
Charles Hansen said:
Here is the real point of this discussion.
Home audio is not a science - it's an adquired taste.
Some prefer open loop amps (AKA zero feedback), other prefer closed loop (AKA high feedback).
Yet another group abhor solid state, whatever it may be the topology.
Speakers may be closed box, vented boxes, TL or dipoles...
Etc, etc, etc.
It would be nice if, for the sake of beginners, posters would make it clear - 'that's my personal taste' - and not the absolute truth.
Hi Charles,
Indeed I have, in the course of my work, done the very audition/test you ask to compare global NFB with NO global NFB. This was on a proudly manufactured and honed 200W/ch bipolar output NO NFB amplifier by a company in Thailand. I was asked to see if I could improve this, their flagship product and I did so in a number of stages each time auditioned by 3 company directors/engineers, myself and Mr. Hans the Dutch loudspeaker engineer.
The first stage was to release as much local feedback as possible in each stage while still ensuring stability of operatingpoint and good linearity. I then put in a global feedback loop and we auditioned. 5 voted overwhelmingly in favour of NFB.
The second stage was to finesse each stage of the amplifier by device optimisation , improving cascode biassing, tuning each stage currents and output voltages. 5 in favour again. A/B/C comparison.
I then ripped half the parts out and put in a chip drive and nested output stage loop and once again all 5 could hear it further improved over A/B/C.
They settled on C and put it into production, as they considered D to have not enough parts in it to be credible in the marketplace.
This product, and the folded cascode preamp I designed in conjuction with a staff engineer/director ( remote control function), gained them international repute and attracted a suitor and the product was rebadged ( and a power filter included) for sale in the US. It won TAS 'Golden Ear'.
Cheers,
Greg
Indeed I have, in the course of my work, done the very audition/test you ask to compare global NFB with NO global NFB. This was on a proudly manufactured and honed 200W/ch bipolar output NO NFB amplifier by a company in Thailand. I was asked to see if I could improve this, their flagship product and I did so in a number of stages each time auditioned by 3 company directors/engineers, myself and Mr. Hans the Dutch loudspeaker engineer.
The first stage was to release as much local feedback as possible in each stage while still ensuring stability of operatingpoint and good linearity. I then put in a global feedback loop and we auditioned. 5 voted overwhelmingly in favour of NFB.
The second stage was to finesse each stage of the amplifier by device optimisation , improving cascode biassing, tuning each stage currents and output voltages. 5 in favour again. A/B/C comparison.
I then ripped half the parts out and put in a chip drive and nested output stage loop and once again all 5 could hear it further improved over A/B/C.
They settled on C and put it into production, as they considered D to have not enough parts in it to be credible in the marketplace.
This product, and the folded cascode preamp I designed in conjuction with a staff engineer/director ( remote control function), gained them international repute and attracted a suitor and the product was rebadged ( and a power filter included) for sale in the US. It won TAS 'Golden Ear'.
Cheers,
Greg
Eva said:
Yes. Thank you for clarifiying this point.
Eva said:
Again, yes. I take this for granted, but this is where many, many audio amplifiers fall on their face. An emitter follower double output stage does not have enough current gain to avoid creating high levels of distortion, yet this is what most audio power amps use. Then the designer relies on negative feedback (as a crutch) to clean up the mess he has made with his bad circuit design. (Using a CFP in the output stage does not solve this problem.)
Eva said:
This is a bad and misleading analogy:
1) There is no "feedback" mechanism in the human body that attempts to operate in real time. Instead these "feedback" mechanisms have a time constant much longer than the forward operating mechanism. To take one example, the level adjustment offered by the three bones in the middle ear don't operate at audio frequencies. If they did, the result would be a gross distortion of the sound we hear.
Instead, they operate at an infra-sonic frequency. In electronic terms, this would be more comparable to a "servo-loop" where the correction happens much more slowly than the operation of the desired signal.
2) It's a red herring to even compare an amplifier to a living organism. Why not compare it to a stone or a star? It's simply a made up construct designed to appeal to one's emotions and has nothing to do with amplifiers.
If you were to make up some silly construct and compare amplifiers to something, why not compare them to musical instruments? A violin has no feedback mechanism. How good do you think a violin would sound if there were an electronic feedback mechanism to try and ensure that the spruce top vibrated in synchronism with the strings?
amplifierguru said:
Hi Greg,
I think you like the "CHIP" front-end very much in your amplifiers....could i ask what's the reason behind it.....apart from that it provides large OL gain and consumes very less space on PCB....Kindly tell us about your Passion towards the CHIP Frontends.....
Kanwar
amplifierguru said:
An interesting experiment, no doubt, but difficult to draw any conclusions from, as there are many unidentified variables. I have had the opposite results with similar experiments myself. My first experiment with a zero feedback solid state amplifier was about 20 years ago. I had a Hafler XL-280 amp which I modified to remove the feedback. I and three others who auditioned it strongly preferred it after the modifications. I haven't looked back since.
amplifierguru said:
If I'm reading your post correctly, you continued to improve upon this circuit after you added the feedback loop. This would strongly imply that the orignal circuit was none too keen. I wonder what would have happened if you had optimized the circuit first, and then added the feedback later. Perhaps the feedback helped the sound of the mediocre circuit, but wouldn't have had the same sonic benefit on an optimized circuit.
amplifierguru said:
As have my zero-feedback designs (most recently for the V-5xe power amp in the February 2005 issue, I think).
Hi Charles,
The amplifier in my example was honed for it's no NFB topology before I worked on it.
Hi Kanwar,
Yes I am partial to chip front ends. Best way to buy a monolithic T tracking FET cascode front end with defined performance AND 100+ dB of defined gain/phase attached, fully spec'd for PSRR,CMRR,S/R....
I have one coming with my R2R output.
Cheers,
Greg
The amplifier in my example was honed for it's no NFB topology before I worked on it.
Hi Kanwar,
Yes I am partial to chip front ends. Best way to buy a monolithic T tracking FET cascode front end with defined performance AND 100+ dB of defined gain/phase attached, fully spec'd for PSRR,CMRR,S/R....
I have one coming with my R2R output.
Cheers,
Greg
Charles, you are truly amazing; to deny negative feedback has anything to do with emitter followers and in your next post use/acknowlege Blackman’s formula for the input impedance of a emitter follower (in a feedback amplifier in/out impedances become functions of the feedback gain)
“feedback is so fundamentally important that analog engineers who don’t understand it should be legally barred from circuit design.” – James K Roberge
“feedback is so fundamentally important that analog engineers who don’t understand it should be legally barred from circuit design.” – James K Roberge
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.