Perhaps you would rather try an IM test, comparing 60/7000 Hz
performance with 1/7000 Hz, for example. The apparatus is easily
fabricated, has high sensitivity, and removes noise from the
results.
😎
performance with 1/7000 Hz, for example. The apparatus is easily
fabricated, has high sensitivity, and removes noise from the
results.
😎
Nelson Pass said:Perhaps you would rather try an IM test, comparing 60/7000 Hz
performance with 1/7000 Hz, for example. The apparatus is easily
fabricated, has high sensitivity, and removes noise from the
results.
😎
I agree, Nelson. I actually built a coherent IM analyzer about 20+ years ago that could measure amplitude IM or phase IM. It used PLL and synchronous detection techniques. It had four or five different LF frequencies between 5 Hz and 200 Hz. I must admit that I never did much testing with those off-normal frequencies, though. I guess I should give it a try. 5 Hz might be low enough to show up some of these effects if they are there.
Bob
I had something a little less sophisticated in mind - An op amp
mixer to the DUT and the output high pass filtered, rectified,
and then low pass filtered.
😎
mixer to the DUT and the output high pass filtered, rectified,
and then low pass filtered.
😎
Further thoughts
So I was thinking on this a little more and then I remembered reading a couple of papers from Otala where he described the possible effects of back EMF from the loudspeaker in circuits with a high loop negative feedback. He called this distortion interface intermodulation distortion or IIM distortion.
In the case of a highly reactive speaker, like an electrostatic speaker for example, much of the signal put into the speaker goes back to the output stage of the amplifier. Now, according to Otala some of this goes through the feedback loop back to the input where it can be reamplified as distortion. The rest is dissipated as heat.
Now this is what seemed interesting to me. The transistors will be heated further by the returning signal, which in some cases is nearly as large as the signal sent out. It seems unlikely that this will heat the transistors evenly.
The question then is can the back EMF affect the bias of the output stage to further drift out of symmetry thus resulting in an increase in zero crossing distortion?
So I was thinking on this a little more and then I remembered reading a couple of papers from Otala where he described the possible effects of back EMF from the loudspeaker in circuits with a high loop negative feedback. He called this distortion interface intermodulation distortion or IIM distortion.
In the case of a highly reactive speaker, like an electrostatic speaker for example, much of the signal put into the speaker goes back to the output stage of the amplifier. Now, according to Otala some of this goes through the feedback loop back to the input where it can be reamplified as distortion. The rest is dissipated as heat.
Now this is what seemed interesting to me. The transistors will be heated further by the returning signal, which in some cases is nearly as large as the signal sent out. It seems unlikely that this will heat the transistors evenly.
The question then is can the back EMF affect the bias of the output stage to further drift out of symmetry thus resulting in an increase in zero crossing distortion?
With all the discussion of banana plugs, I have to ask, what's wrong with the usual spades and binding posts? If the surfaces are clean and you tighten very hard, you can get something close to a cold weld.
Nixie said:......what's wrong with the usual spades and binding posts? ....
Absolutely nothing!
Nelson Pass said:I had something a little less sophisticated in mind - An op amp
mixer to the DUT and the output high pass filtered, rectified,
and then low pass filtered.
😎
Great hint!
Bob Cordell said:
Glen, You're right about one thing: this discussion is going around in circles. Tell us the title and author of the book. Just because something is written in a book does not make it right. If what you were saying had merit, the concept and point would be made in plenty of books that are still in print. The concept that open loop gain that is so high at low frequencies that the error signal must be smaller than the input-referred noise of the input stage seems nicely intuitive, but it is simply wrong.
The book is out of print. It is called OP AMPS Explained, author Bryan Maher (Chapter 3, Noise and distortion) It has been used as a basic University and TAFE text for electronic engineering here in Australia for probably a couple of decades now. In fact, I was taught from it.
Tell us more about this mysterious amplifier of yours. What is the gain crossover frequency? What is the compensation rolloff rate? What kind of compensation are you using if it is not Miller. I told you about mine, why don't you elaborate on yours. Cheers, Bob
Oh, yes, my “mysterious” amplifier. I don’t particularly care if someone disagrees with me on a technical point, but I do care if what I say is misrepresented (eg 67K resistors) either due to laziness or pomposity on behalf of the reader, and the continual condescending tone of your posts in reply is getting rather tiring.
Now this discussing has been going in circles. I’ve offered to scan the relevant pages of the book cited and send them to you in the hope of diverting it into a straight line. If you choose to make generalisations and assumptions about the validity of it content without having seen it instead, then fine. If so, then this will be the last I have so say on the topic.
Cheers,
Glen
Nelson Pass said:I had something a little less sophisticated in mind - An op amp
mixer to the DUT and the output high pass filtered, rectified,
and then low pass filtered.
😎
Sounds like a good job for a modified Heathkit IM5258. Let me check eBay....
Jan Didden
You should have done it, as it took 30 seconds and you wouldn't have posted this line if you had. Worldwide search in both titles and descriptions for IM5258 returns zero results.janneman said:Let me check eBay....
Nixie said:
You should have done it, as it took 30 seconds and you wouldn't have posted this line if you had. Worldwide search in both titles and descriptions for IM5258 returns zero results.
Thanks for the help. Too bad, no hits. Mind you, I had one of those, but sold it...

Jan Didden
Big deal.
Schematic
Here, kitty kitty : http://cgi.ebay.com/Heathkit-Harmon...78QQihZ015QQcategoryZ3284QQrdZ1QQcmdZViewItem
Schematic
Here, kitty kitty : http://cgi.ebay.com/Heathkit-Harmon...78QQihZ015QQcategoryZ3284QQrdZ1QQcmdZViewItem
jacco vermeulen said:Big deal.
Schematic
Here, kitty kitty : http://cgi.ebay.com/Heathkit-Harmon...78QQihZ015QQcategoryZ3284QQrdZ1QQcmdZViewItem
Jacco,
I think I had the type number wrong, this is the THDF analyzer. The IM analyzer should be the IM5248 IIRC. But isn't this becoming too off-topic???
Jan Didden
Nixie said:With all the discussion of banana plugs, I have to ask, what's wrong with the usual spades and binding posts? If the surfaces are clean and you tighten very hard, you can get something close to a cold weld.
Good point. There's nothing wrong with these and other alternatives to the banana plug. The banana plugs are just very convenient in some situations.
Bob
G.Kleinschmidt said:
The book is out of print. It is called OP AMPS Explained, author Bryan Maher (Chapter 3, Noise and distortion) It has been used as a basic University and TAFE text for electronic engineering here in Australia for probably a couple of decades now. In fact, I was taught from it.
Oh, yes, my “mysterious” amplifier. I don’t particularly care if someone disagrees with me on a technical point, but I do care if what I say is misrepresented (eg 67K resistors) either due to laziness or pomposity on behalf of the reader, and the continual condescending tone of your posts in reply is getting rather tiring.
Now this discussing has been going in circles. I’ve offered to scan the relevant pages of the book cited and send them to you in the hope of diverting it into a straight line. If you choose to make generalisations and assumptions about the validity of it content without having seen it instead, then fine. If so, then this will be the last I have so say on the topic.
Cheers,
Glen
Thanks for the book title, Glen. Yes, please do scan the relevant pages and post them or email them to me at bob@cordellaudio.com.
Guess you got out of the wrong side of bed this morning - your personanl attack seemed to come out of no-where. Let's try and keep this revived discussion on a technical level and see where it leads, hopefully not in a circle again. I'll be happy to take a look at the relevant pages in the book and let you know what I think.
As for your amplifier, I'm apologize for referring to it as mysterious, and I am interested in seeing the details that I referred to.
Bob
Nelson Pass said:I had something a little less sophisticated in mind - An op amp
mixer to the DUT and the output high pass filtered, rectified,
and then low pass filtered.
😎
Yes, this is the way the old SMPTE IM analyzers used to do it. It works fine, but its sensitivity isn't as good as with a coherent detection approach. I'm just not sure how subtle the alleged effect might be here, so I don't know if the envelope detection approach would be sensitive enough.
Cheers,
Bob
Re: Further thoughts
These are good points. The idea of injecting a current back into the amplifier for test purposes can be quite useful for this and similar issues. You're right, the injection of a low frequency current back into the amplifier could help bring out the thermal-based asymmetry you mention. Another advantage of back-injection is that the low output impedance of the power amplifier helps tend to suppress the LF test signal, making it easier to see the effect on the desired signal (for example, you might propogate a 6 kHz signal in the forward direction as the "victim" signal, and pump in a 1 Hz "agressor" signal in the reverese direction).
I have a paper on my website at www.cordellaudio.com that shows analysis and measurements of Interface Intermodulation Distortion that you might be interested in reading. The paper shows that, all else remaining equal, negative feedback does not bring on or increase this distortion. Nevertheless, the phenomenon of sometimes unexpectedly high speaker currents, in some cases a result of back EMF, is real, and is one of the reasons that it is desirable for power amplifiers to be able to supply high currents at their outputs.
Cheers,
Bob
morricab said:So I was thinking on this a little more and then I remembered reading a couple of papers from Otala where he described the possible effects of back EMF from the loudspeaker in circuits with a high loop negative feedback. He called this distortion interface intermodulation distortion or IIM distortion.
In the case of a highly reactive speaker, like an electrostatic speaker for example, much of the signal put into the speaker goes back to the output stage of the amplifier. Now, according to Otala some of this goes through the feedback loop back to the input where it can be reamplified as distortion. The rest is dissipated as heat.
Now this is what seemed interesting to me. The transistors will be heated further by the returning signal, which in some cases is nearly as large as the signal sent out. It seems unlikely that this will heat the transistors evenly.
The question then is can the back EMF affect the bias of the output stage to further drift out of symmetry thus resulting in an increase in zero crossing distortion?
These are good points. The idea of injecting a current back into the amplifier for test purposes can be quite useful for this and similar issues. You're right, the injection of a low frequency current back into the amplifier could help bring out the thermal-based asymmetry you mention. Another advantage of back-injection is that the low output impedance of the power amplifier helps tend to suppress the LF test signal, making it easier to see the effect on the desired signal (for example, you might propogate a 6 kHz signal in the forward direction as the "victim" signal, and pump in a 1 Hz "agressor" signal in the reverese direction).
I have a paper on my website at www.cordellaudio.com that shows analysis and measurements of Interface Intermodulation Distortion that you might be interested in reading. The paper shows that, all else remaining equal, negative feedback does not bring on or increase this distortion. Nevertheless, the phenomenon of sometimes unexpectedly high speaker currents, in some cases a result of back EMF, is real, and is one of the reasons that it is desirable for power amplifiers to be able to supply high currents at their outputs.
Cheers,
Bob
Regarding IIM I have a question.
Assume the amplifier has HIGH output impedance in OPEN loop and dominant pole compensation at a very low frequency.
For given mid-audio frequency it can have around 70degrees of phase shift because of freq. compensation.
Now, load impedance affects open loop tranfer function by inverse of open-loop-damping-factor, thus introduces considerable phase shifts near mechanical resonances of speaker.
For some frequency band total phase shift in open loop can therefore become very considerable, like 120deg or more, so that amplifier has very very little feedback for this very band (=bad luck!!!).
Is my resoning OK?
Assume the amplifier has HIGH output impedance in OPEN loop and dominant pole compensation at a very low frequency.
For given mid-audio frequency it can have around 70degrees of phase shift because of freq. compensation.
Now, load impedance affects open loop tranfer function by inverse of open-loop-damping-factor, thus introduces considerable phase shifts near mechanical resonances of speaker.
For some frequency band total phase shift in open loop can therefore become very considerable, like 120deg or more, so that amplifier has very very little feedback for this very band (=bad luck!!!).
Is my resoning OK?
- Home
- Amplifiers
- Solid State
- Bob Cordell Interview: Negative Feedback