Hi,
I've read the 4 pdf's of audio opamps. The part where it talks about OLBandwith is in part 3. There are 3 suggestion for extending open loop bandwith:
1. Move the open loop corner by putting resistor to pin5 in AD829. The opamp has to have this compensation pin. NE5532 doesn't have this pin.
2. Use more linear device for input stage. Fet is preferable than Bipolar.
3. Simply pick up a device which has OLBandwith 20khz or more.
2 of these suggestion is "look for another opamp".
We cannot do nothing to extend OLBandwith of NE5532?
I've read the 4 pdf's of audio opamps. The part where it talks about OLBandwith is in part 3. There are 3 suggestion for extending open loop bandwith:
1. Move the open loop corner by putting resistor to pin5 in AD829. The opamp has to have this compensation pin. NE5532 doesn't have this pin.
2. Use more linear device for input stage. Fet is preferable than Bipolar.
3. Simply pick up a device which has OLBandwith 20khz or more.
2 of these suggestion is "look for another opamp".
We cannot do nothing to extend OLBandwith of NE5532?
Yes, but......
They had the same reaction to some power amps that used the same "current feedback" technique. IOW, Alexander amps, and variations of it. Some of which were discrete.
I seem to recall that EAD wrote a "white paper" on using "C-F" op amps as I/Vs, and their claim was that the settling time was too long to allow 16-bit resolution, maybe 10 or 12-bits. Maybe someone else will recall the details of it. It was a few years ago.
I dunno........I don't have "the reason".
Jocko
They had the same reaction to some power amps that used the same "current feedback" technique. IOW, Alexander amps, and variations of it. Some of which were discrete.
I seem to recall that EAD wrote a "white paper" on using "C-F" op amps as I/Vs, and their claim was that the settling time was too long to allow 16-bit resolution, maybe 10 or 12-bits. Maybe someone else will recall the details of it. It was a few years ago.
I dunno........I don't have "the reason".
Jocko
Here's a thought: PSRR.
The typical CF model does not work to supress PS noise, since it is singled-ended in the front end.
A personal observation: Audio stages using CF op amps seem to be more sensitive to the quality of the power supplies feeding them. An example would the ckt. cited in post 483 of this thread. This is my own observation, and has nothing to do with the AD846 and/or anyone's audible asessment.
wj
Folks, any dual IC's in an 8 pin package are difficult to change much, because there aren't any spare pins. However, there are other, better op amps that can be substituted. Some have a reasonably high open loop bandwidth, others have, at least, a more linear input stage than is possible with the 5532.
After saw the discussion of the "gurus" here, I can see (at least a little part) why Mr. JC insisted on putting JFET for differential input and why Mr. Pass likes to built amp with Mosfets all transistors. A single mosfet has low gain, compared with single bipolar. This will ends up with wide OLG. That reduces the distortions discussed here, wont they?
And why some designers likes simple design approach. It does the same effect too, for those nasty PIM family.
And why some designers likes simple design approach. It does the same effect too, for those nasty PIM family.
john curl said:
Looks like there was one paper left off this list. It also appears to be a thorough treatment of the subject, with lots of measurements. The author built a special distortion analyzer to measure PIM, and also constructed an amplifier that could be used both with and without feedback to test the effects of feedback on PIM. The description of the design of this distortion analyzer is quite interesting. I'd love to see a schematic of the thing. I believe Jan posted a reference to this article earlier, but I didn't have access to it at the time and just came upon it a little while ago while searching for some other info.
It's been mentioned before in this thread that there are other sources of PIM that are not created by feedback. One such source is nonlinear collector-base capacitance, which can cause a level-dependent phase shift. This form of PIM is reduced by feedback in the normal way. The author refers to this as "preexisting PIM". He does measurements that show the level of PIM both before and after applying feedback. Here's a quote from the article:
.
The article is:
"Phase Intermodulation Distortion Instrumentation and Measurements"
Robert R. Cordell
AES Journal Vol. 31 No. 3, March 1983.
Just shoot me an email if interested.
I realize there were personality conflicts at the time and I'm not bringing it up to rub that in. Since the discussion has centered on PIM, it seems reasonable to mention an article that does measurements of it using a piece of equipment especially designed for that purpose.
Folks, it is really a matter of attitude. I am sure that the Cordell article is worth reading, but its conclusions wlll be essentially negative, much like many on this thread that PIM is very important. What is interesting is that Barrie Gilbert revived this subject aproximately 15 years later.
We have had problems with Bob Cordell, previously, with TIM. We wrote a rebuttal to him in 1980, a synopsis of which, was published in 'Audio' in 1980.
We have had problems with Bob Cordell, previously, with TIM. We wrote a rebuttal to him in 1980, a synopsis of which, was published in 'Audio' in 1980.
The BG paper is interesting in more ways than have been mentioned here. If you read it, at the end he promises paradise (so to speak) by, among others, Active Feedback.
This was just around the time that the AD830 was introduced. I vividly remember having a chat with Barrie in I think around 1995 in Eindhoven, on the AD830, where the travelling ADI circus was doing another application seminar at that time. We all were enthralled by the '830, and I encourage you to study the data sheet.
Looking back, the breakthrough of AF hasn't happened. Why? Probably several reasons, but one IMHO is the fact that it works with a specific input voltage between V+ and V-, unlike a 'normal' feedback amp where in theory the input voltage approaches zero. That specific input level of course brings about many of the non-linearities discussed here.
In hindsight, that 1995(?) talk by Barrie was a strong plug for AF by elaborating the pittfalls of claasic opamps, and it is quite ironic to think that AF didn't take off as much for precise the reasons classical opamps are flawed.
Ahh, who said Mother Nature isn't even-handed?
Jan Didden
This was just around the time that the AD830 was introduced. I vividly remember having a chat with Barrie in I think around 1995 in Eindhoven, on the AD830, where the travelling ADI circus was doing another application seminar at that time. We all were enthralled by the '830, and I encourage you to study the data sheet.
Looking back, the breakthrough of AF hasn't happened. Why? Probably several reasons, but one IMHO is the fact that it works with a specific input voltage between V+ and V-, unlike a 'normal' feedback amp where in theory the input voltage approaches zero. That specific input level of course brings about many of the non-linearities discussed here.
In hindsight, that 1995(?) talk by Barrie was a strong plug for AF by elaborating the pittfalls of claasic opamps, and it is quite ironic to think that AF didn't take off as much for precise the reasons classical opamps are flawed.
Ahh, who said Mother Nature isn't even-handed?
Jan Didden
Rodolfo,
Not fully answering your question, but shedding light on the problem, read the interview with Rich Cabot, of Audio precision fame. Very readable.
http://www.stereophile.com/interviews/850/
To wet your appetite:
"Robert Harley: Do you believe that every audible phenomenon can be measured with existing instruments?
Cabot: No, I wouldn't say that. I think an awful lot of it can be, but just isn't measured. But I have no doubt that measurements will be found in the future that we don't know how to do today. I'm sure they will be found. If you can identify the problem, I can figure out a way to measure it. The problem is identifying the problem. If you tell me the physical effect that is causing what you hear, then I know I can find a way to measure that physical effect. But if you don't know what the physical effect is, only that you hear something, I can't necessarily find a way to measure that. "
Jan Didden
Not fully answering your question, but shedding light on the problem, read the interview with Rich Cabot, of Audio precision fame. Very readable.
http://www.stereophile.com/interviews/850/
To wet your appetite:
"Robert Harley: Do you believe that every audible phenomenon can be measured with existing instruments?
Cabot: No, I wouldn't say that. I think an awful lot of it can be, but just isn't measured. But I have no doubt that measurements will be found in the future that we don't know how to do today. I'm sure they will be found. If you can identify the problem, I can figure out a way to measure it. The problem is identifying the problem. If you tell me the physical effect that is causing what you hear, then I know I can find a way to measure that physical effect. But if you don't know what the physical effect is, only that you hear something, I can't necessarily find a way to measure that. "
Jan Didden
Cabot interview / Audiophile
Jan:
Thank you very much for the link.
It is sobering to learn how someone with the background of Cabot is openminded enough to confess there is the possibility of audible effects not measured yet.
And this is declared by a person devoted precisely to audio measurements in all applicable domains.
Of course he quickly points out once a relevant physical connection can be established, it will be duly characterized.
Some days back I downloaded all relevant technical papers from the AP site. Worth reading carefully, enlightening stuff.
Yet as you said, this does not answer the former question. I guess Walt's EDN article series in some way point in the direction of an audio OpAmp reference design, but stops short of the design itself, or at least to an actual device approaching it.
Rodolfo
Jan:
Thank you very much for the link.
It is sobering to learn how someone with the background of Cabot is openminded enough to confess there is the possibility of audible effects not measured yet.
And this is declared by a person devoted precisely to audio measurements in all applicable domains.
Of course he quickly points out once a relevant physical connection can be established, it will be duly characterized.
Some days back I downloaded all relevant technical papers from the AP site. Worth reading carefully, enlightening stuff.
Yet as you said, this does not answer the former question. I guess Walt's EDN article series in some way point in the direction of an audio OpAmp reference design, but stops short of the design itself, or at least to an actual device approaching it.
Rodolfo
Swedish Chef said:Rodolfo, since we can not agree on the use/abuse of global feedback in the first place I think there will be approximately as many opinions as there are people...
/Magnus [/B]
True. I suggested some posts back global feedback by itself should not be relied on to cure a grossly nonlinear gain path no matter how high the gain.
I guess global feedback could shine when applied to a fairly linear gain path to begin with, which may consist of cascaded stages with moderate amounts of local feedback. You can see this approach in some designs like the Leach amplifier for example.
I am currently toying with simulation of an ouput stage with 2 steps of voltage gain and local feedback. Not that I will end building it, I prefer class D for high power and chips for moderate power, but the idea comes back form many years ago.
I will post results if something interesting comes out. First I must come to good terms with LTSpice (I used MicroCap for years).
Rodolfo
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.