jcx,
Thanks for that, someone had to do it
...
I agree with your conclusion which I understand as follows: distortion, whatever it is, ultimately manifests itself in unwanted spectrum components. Feedback can make those unwanted components inaudible under the best of circumstances, let alone in music material. Therefore, the exact source of the distortion becomes irrelevant.
What is not clear to me what happens with the original spectral component (carrier) in some of these exotic distortion types. Could there not be a distortion type that translates a wanted freq into an unwanted? Like, we feed it 1kHz and out comes 2kHz, and the 1kHz is gone or heavily supressed?
Jan Didden
Thanks for that, someone had to do it
...I agree with your conclusion which I understand as follows: distortion, whatever it is, ultimately manifests itself in unwanted spectrum components. Feedback can make those unwanted components inaudible under the best of circumstances, let alone in music material. Therefore, the exact source of the distortion becomes irrelevant.
What is not clear to me what happens with the original spectral component (carrier) in some of these exotic distortion types. Could there not be a distortion type that translates a wanted freq into an unwanted? Like, we feed it 1kHz and out comes 2kHz, and the 1kHz is gone or heavily supressed?
Jan Didden
On the other hand, I strongly doubt about testing of audio amplifiers only by steady-state, periodical, repetitive, one or little bit more fundamental harmonics signals . This is very good when we evaluate our products by eyes (visualized results) , probably not so good in correlation with what we hear .
. This is very good when we evaluate our products by eyes (visualized results) , probably not so good in correlation with what we hear .
traderbam said:
I am hardly surprised the open-loop performance of your amplifier was somewhat superior to it's closed-loop characteristics.
The reason is quite.....elementary my dear Watson...
If, as you claim, you were able to listen to your design open-loop, it can only mean the circuit had a foward-path gain no greater than 35dB!!!!
(This assumes of course, that the design was operating from approximately 50V supplies.....???
Otherwise, for lower rails, the foward gain would have to be even less!!!)
So....when you closed the global loop, you couldn't possibly have had more than 3dB of feedback!
Such a low feedback factor has been shown to increase distortion...not reduce it.....(Baxandal)
Your experiment therefore, does not demonstrate that feedback was inherently 'bad' for your design's health, but rather all you were able to demonstrate was the woefully inadequate foward path gain of your amp., and that indeed too little feedback is not good practice.
I made another experiment. The same amp with very linear class A output stage (Av=+1), THD < 0.01%, damping factor = 80 (for the output stage itself). Feedback was once taken before the output stage and second time behind the output stage. I made a lot of listening tests and audience splitted into 2 halves - 1st prefered output stage outside global NFB, 2nd prefered output stage inside global NFB. There was the only 1 correlation - people who prefered output stage inside NFB were more experienced listeners and listening often to "complex" music. People who prefered output stage outside NFB were listening to folk and jazz preferably .
.
http://www.pha.inecnet.cz/macura/pma1_en.html
http://web.telecom.cz/macura/pma2_en.html
1st one with output stage outside global NFB, 2nd inside.
http://web.telecom.cz/macura/pma2_en.html
1st one with output stage outside global NFB, 2nd inside.
Basic problem in comparative tests of amps is incoming signal, 'cos " source " is in 99 % cases digital recording with all his disadvantages, which reliable erase and distorte listening results. I recomend to all, which can to know how is truth, use more for testing " direct signal ", incoming mainly from some realy good mic preamp. I mean, that with this method will be most of people very quickly " healed " from their " love " to " musical sounding " amps .
.
Nice suggestion, but what does the microphone record ?
Birds singing ? Cars driving ? Me singing ?
At least this discards the question, is the recording bad or my amp ?
Then its the question, is the microphone bad or my amp ?
As the amp is typically fed with CD, it should sound good with this.
But still, i want a good sounding amp and a neutral amp, is this too much ?
Mike
Birds singing ? Cars driving ? Me singing ?
At least this discards the question, is the recording bad or my amp ?
Then its the question, is the microphone bad or my amp ?
As the amp is typically fed with CD, it should sound good with this.
But still, i want a good sounding amp and a neutral amp, is this too much ?
Mike
I have one question here that I do not think have been covered in the 30+ pages of this discussion...
When someone claims an amplifier to be "without feedback" - just how strict is that statement? Take a constant-current source for example (like the one in the original Zen amp, and this is just an example - no critizism intended for NP ). This is indeed a circuit with feedback - and lots of it too. Clearly it is not GFB in the strict sense and most people would probably think of it as LFB. Which in fact, even an emitter degeneration resistor is.
But personally I can not see how any amplifier using this topology or similar ones could be said to be "without feedback". Can someone please enlighten me?
/Magnus
When someone claims an amplifier to be "without feedback" - just how strict is that statement? Take a constant-current source for example (like the one in the original Zen amp, and this is just an example - no critizism intended for NP
). This is indeed a circuit with feedback - and lots of it too. Clearly it is not GFB in the strict sense and most people would probably think of it as LFB. Which in fact, even an emitter degeneration resistor is.But personally I can not see how any amplifier using this topology or similar ones could be said to be "without feedback". Can someone please enlighten me?
/Magnus
Mike, I know what you can - the same as I can , but in the end you will have realy " fidelity amp " and you will hear "all " - and I guarantee to you, that this " all " will be not so pleasant 
. What you will be doing in that time ? Do you will be searching for some " musical one ", which make all now ?
, but in the end you will have realy " fidelity amp " and you will hear "all " - and I guarantee to you, that this " all " will be not so pleasant . What you will be doing in that time ? Do you will be searching for some " musical one ", which make all now ?
Swedish Chef said:But personally I can not see how any amplifier using this topology or similar ones could be said to be "without feedback". Can someone please enlighten me?
Hi Magnus,
There have been a few massive flame wars about this in the past
.One of them is here: http://www.diyaudio.com/forums/showthread.php?s=&threadid=28448&highlight=
A spinoff from that one is here: http://www.diyaudio.com/forums/showthread.php?s=&threadid=28764&highlight=
john curl said:
I guess it’s hard to see with the smoke from burning 20 yr old straw men in your eyes – the title of the thread is “What’s your reasoning”
The smoke certainly isn’t shedding any light on your rational for treating PIM or “FM” intermodulation distortion products as special compared to “AM” type intermodulation products
To justify creating a high open loop bandwidth by shunting a VAS/gm stage output with a load resistance – throwing away low frequency loop gain, you have to make the case that PIM is orders of magnitude worse than AM distortion products – when the difference is merely the relative phase of the IM products
I have recently looked at Otala’s AES PIM preprint, Walt’s “Op amp Audio” series, Gilbert’s “Are Op Amps Really Linear” article (resurrected from my Win/temp internet folder by a content search) and Cherry’s JAES papers from the 80’s, please point out the missing piece(s) of your rational for us because I certainly missed the part about how it has been established that PIM deserves to be treated differently by audio amplifier designers
Let's just say that I believe that PIM or something similar is why many professional audio designers don't like to use negative feedback any more than necessary.
I spoke to Walt Jung about it today. We talked about local and loop feedback on input stages. Walt agreed that this would linearize the input stage and greatly reduce PIM. However, he also stated that that would still not make global negative feedback next to perfect, as far as he was concerned.
Walt and I, through experience, have found that op amps with high open loop bandwidth almost always sound better that op amps with low open loop bandwidth. PIM is a good candidate for why this is our experience.
However, it is possible that acceptable designs exist that use large amounts of global negative feedback with low open loop bandwidth. So far, some candidates for this are much more expensive than my power amp designs and comparatively priced with my best preamp designs.
I spoke to Walt Jung about it today. We talked about local and loop feedback on input stages. Walt agreed that this would linearize the input stage and greatly reduce PIM. However, he also stated that that would still not make global negative feedback next to perfect, as far as he was concerned.
Walt and I, through experience, have found that op amps with high open loop bandwidth almost always sound better that op amps with low open loop bandwidth. PIM is a good candidate for why this is our experience.
However, it is possible that acceptable designs exist that use large amounts of global negative feedback with low open loop bandwidth. So far, some candidates for this are much more expensive than my power amp designs and comparatively priced with my best preamp designs.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.