Jan,
you misunderstood me, I did mean your criticism of my post in the other thread.
Otala revealed the mechanism of dynamic distortions, but he does not necessarily need to be an excellent designer. It`s never good to base a design on just one principle. I personally dislike the idea of two voltage amplifier stages. An absence of dynamic distortions does not mean linear behavior in all aspects, distortions have many sources.
Feedback tries to correct insufficiencies, but such a thing as a free lunch does not exist in physics. Distortions can only change forms and cannot be swept out like trash, GNF transforms harmonics into high frequency noise doing harm in its way, you may like the sound better that way or you may not.
you misunderstood me, I did mean your criticism of my post in the other thread.
You did indeed and my view is not at all different. The need for more feedback indicates problems.
Otala revealed the mechanism of dynamic distortions, but he does not necessarily need to be an excellent designer. It`s never good to base a design on just one principle. I personally dislike the idea of two voltage amplifier stages. An absence of dynamic distortions does not mean linear behavior in all aspects, distortions have many sources.
Feedback tries to correct insufficiencies, but such a thing as a free lunch does not exist in physics. Distortions can only change forms and cannot be swept out like trash, GNF transforms harmonics into high frequency noise doing harm in its way, you may like the sound better that way or you may not.
Jan, I got Mark Levinson 'on the map', so to speak, with my designs. If you don't believe that, then why am I as well known as I am throughout the world? Mark Levinson did me a professional favor when he first started, by including my name as the designer, with the products. This is why you know of me today, not for my work with the Grateful Dead, my years at Ampex, or my working on two major rock films in the early 70's, including my year in Hollywood, working part time on the studios there. This all happened before I met Mark Levinson. Ever heard of my contributions in this context? No!
Mark Levinson came to regret giving me credit, and wanted it all for himself, even though he was not paying me directly, up front, for my circuit designs, and only a minimal royalty (candy money really) was asked for by me. By 1976, Mark reneged on our handshake deal, first by assigning credit for the JC-3, to someone else, then refusing to pay me further royalties on the JC-2. This is what caused our separation. I never did work FOR him in the first place. All this happened when I was working for Gale in London. Ever hear of that amp effort? How about my designs for a Paris recording studio that I made in Switzerland, or my time as a faculty member of IHEM, or the Institute for Advanced Musical Studies, located in Montreux, Switzerland, for several years?
You see, it was Mark Levinson, who put me on the map, so to speak, and I thank him for it. I had to keep my reputation going on my OWN merits, independently, so to speak, after 1976, but I never advertised, except for a little ad in the back of 'Audio' magazine once in a while, in order to get a little business.
It is true, like many companies, such as General Motors, or Ford, that once you establish some sort of reputation, the inertia of marketing will keep people buying your products. Still, the real Mark Levinson, makes no more electronics, and only a faceless group of electronic engineers, who work behind the scene, making well 'engineered' designs, that sometimes work well, and others, not so well, as they, like you, believe in measurement, and in their case, they can throw money at the product. (sometimes a good thing).
In later years, doing research with Matti, writing my own paper, for the IEEE, when working with HK as a consultant, and giving it at a number of locations across the world kept my reputation going. That is the fact of the matter.
Mark Levinson came to regret giving me credit, and wanted it all for himself, even though he was not paying me directly, up front, for my circuit designs, and only a minimal royalty (candy money really) was asked for by me. By 1976, Mark reneged on our handshake deal, first by assigning credit for the JC-3, to someone else, then refusing to pay me further royalties on the JC-2. This is what caused our separation. I never did work FOR him in the first place. All this happened when I was working for Gale in London. Ever hear of that amp effort? How about my designs for a Paris recording studio that I made in Switzerland, or my time as a faculty member of IHEM, or the Institute for Advanced Musical Studies, located in Montreux, Switzerland, for several years?
You see, it was Mark Levinson, who put me on the map, so to speak, and I thank him for it. I had to keep my reputation going on my OWN merits, independently, so to speak, after 1976, but I never advertised, except for a little ad in the back of 'Audio' magazine once in a while, in order to get a little business.
It is true, like many companies, such as General Motors, or Ford, that once you establish some sort of reputation, the inertia of marketing will keep people buying your products. Still, the real Mark Levinson, makes no more electronics, and only a faceless group of electronic engineers, who work behind the scene, making well 'engineered' designs, that sometimes work well, and others, not so well, as they, like you, believe in measurement, and in their case, they can throw money at the product. (sometimes a good thing).
In later years, doing research with Matti, writing my own paper, for the IEEE, when working with HK as a consultant, and giving it at a number of locations across the world kept my reputation going. That is the fact of the matter.
Lumba Ogir said:
OK, I see what you mean. But still I can't do much with it. I agree that Otala revealed this mechanism, but it is not correct that it can only be avoided with his type of amp (low OL gain, low FB factor), as has been shown for instance by Cordell, together with clear measurements.
Secondly, Electrocompagniet stated (in my quote) that increasing the FB by 10dB significantly improved the sound. And since this amp enjoyed a lot of success, you can't just blame a confused engineer here.
You may dislike 2 voltage amp stages, but again, so what (and I don't mean that denigratingly)? Should we compare what we dislike to decide what is the best amp topology?
Lumba Ogir said:
Again, these are some generalities 'distortion has many sources', yes, and? And that last sentence, as a generality yes, more NFB CAN lead to higher harmonics, but it can also decrease higher harmonics. Depends, as shown by Baxandall.
I think one of the problems in these type of discussions is that people try to make nice, comfortable black-and-white statements. 'NFB is bad', 'high OL BW is good'. Well, if you look around you with your eyes open for more than a few seconds it becomes abundantly clear that reality and nature is a lot of things, but not black and white.
Trying to force things in B&W decreases your ability to increase your understanding. (And yes, this is a 'grey' statement). 😉
Jan Didden
Jan, please have a look on schematics. The amplifier is inverting one (the noninverting input connected to ground), feedback resistors are 39K and 6K8, summing point connected to ground trough series 470 Ohm and 220 pF. You can understand that because of this feedback network the peak error voltage will be more than 10 times lower than input step, and the first stage wouldn't be overloaded. It is not low OL gain, low FB, this is clever feedback network. The input filter will do the same.
dimitri said:
OK, I don't have that schematic handy, I believe you on your blue eyes 😉 .
So where does that leave us? We still are facing the 'black-and-white' statement that feedback is bad and that the Otala amp is so good because it has so little of it. Even if that is factually incorrect, that's the statements flying around.
Jan Didden
Jan, it has little to do with the ACTUAL amount of feedback. It is the 20KHz open loop bandwidth requirement (originally set by Daugherty and Greiner in 1966, and the LIMITED gain bandwidth, due to the 4MH ft of the output transistors, that sets the AMOUNT of feedback. You have only 1/2 the picture in your head.
'Some Design Objectives for Audio Power Amplifiers' IEEE TRANSACTIONS IN AUDIO AND ELECTROACOUSTICS VOL. AU-14, No 1 March 1966
'Some Design Objectives for Audio Power Amplifiers' IEEE TRANSACTIONS IN AUDIO AND ELECTROACOUSTICS VOL. AU-14, No 1 March 1966
dimitri said:
And your point is...?
What's the relevance of the "peak error voltage"?
Why is this so "clever"?
Have you calculated or simulated the open loop gain, the loop gain and ULG frequency?
john curl said:
Sure. The effective feedback at a particular frequency is the difference between the OL gain and the CL gain at that frequency, no?
Jan Didden
john curl said:
I don't see this. Care to clarify how "the limited gain bandwidth sets the amount of feedback"?
EDIT: Is Jan correct in his assumption, this is what you are talking about?
Wavebourn said:
Defenitly worth experimenting with, and the path that the later electrocompaniets have taken.
Lumba Ogir said:
Yet these amps with two voltage stages recieve acolades from owners and the audio press, You can add the pioneers of yester year to that list as well as it seems that its during those years that subjective critics regard pioneer as worthy of being called audiophile amps.
Otala didn't design open loop amps, and low OL gain, low FB factor is marketing BS
Because in the amp with conventional fb network peak error voltage is equal to peak input step -> slewing (hard TIM)
because this particular design is free from hard TIM
Yes, I built this with many variations in early 80s
dimitri said:
But Dimitri, that is only when you drive the input with a signal faster than the amps' slew rate. It's not difficult to upset an audioamp by driving it with sub-microsecond rise times, but why would you do that?
That's the whole discussion isn't it: if you have an 100W amp (40V peak) that can slew that 40V in 1uS, it will never slew-rate limit at any reasonable audio input. If you want belt and suspenders, make it a 100V/uS amp and put a 200kHz input low-pass in front of it. Done. What's next.
Jan Didden
Hi everybody
I just talked with Per Abrahamsen on the phone and told him about this discussion.
He was on holiday now and had a bad internet connection, but he will register to the forum.
I just talked with Per Abrahamsen on the phone and told him about this discussion.
He was on holiday now and had a bad internet connection, but he will register to the forum.
stinius said:
Great! Looking forward to his views!
Moderators please note: zip this guy through moderation period fast!
Jan Didden
dimitri said:
- Did anybody say Otala designed open loop amps?
- Nope. No sane engineer would design an amp to slew under the input signal max slew rate. What is the max slew rate in audio? Should I quote Walt Jung's article about?
- It could be, but not because of any low feedback.
- And the OLG LG and ULG freq results are...?
according to Baxandall, 0.5V/us 😀
about 100kHz, 22dB, 3MHz
I see that the words "low feedback" act as a red cloak for you. And loudspeaker doesn't have feedback, and it works.
I *think* I remember that rule of thumb, with some headroom, wanted slew rate is about 1V/uS per V amplitude at 20kHz, eg for a 50V pk-pk signal at 20kHz you'd want 50V/uS.
Don't remember where I read or heard it though.
Jan Didden
Don't remember where I read or heard it though.
Jan Didden
Walt Jung, and I agree with him. Actually, it is 0.5V/us/Vpp. So a 200 Vpp swing would require 100V/us or more. Check out the spec. of the JC-1 power amp.
john curl said:
Hi John,
So therefore you agree that an amplifier with 40 dB of NFB at 20 kHz will sound as good or better than a comparable amplifier with 20 dB of NFB at 20 kHz, as long as they both have 20 kHz of open-loop bandwidth?
BTW, I find it important, when describing an amplifier as having high or low negative feedback, to be clear about whether we are talking about amount of NFB at 20 kHz or amount of NFB at low frequencies, like 100 Hz. These are of course very different numbers for amplifiers with low open loop bandwidth (but which may have just as much NFB at 20 kHz, just as much slew rate, and just as much closed-loop bandwidth).
Given that TIM is not increased by higher NFB at low frequencies (and hence lower open-loop bandwidth), then why specifically do you assert that wide open-loop bandwidth is necessary? It is certainly not apparent from the 1966 Greiner paper you quoted.
Cheers,
Bob