Interesting books....

Status
Not open for further replies.
Jump to Latest
#223
john curl said:
......PIM is real, and probably the MAJOR problem with global negative feedback......
Click to expand...

Mr Curl, you apparently have several years experiance producing audio electronics.

I therefore find your inability to rigorously defend your positions wholly inexplicable, and i dare say, rather disappointing.

Merely repeating something many times without number does not validate it, and conversely.....

PIM, TIM etc, are red herrings...mere fig leaves used by subjectivism to prop up its shaky edifice..... 🙁
 
#225
Mikek, I don't teach 'high school'. That is one of my personal rules. SOMETIMES, some of you have to study the subject, on your own, to gain enough understanding in order to follow the opinions put forth. While it takes heavy math to DERIVE the equations for PIM, it is relatively easy to predict it. This does not take heavy math.
When Till asks me to give him the 'in and outs' of TIM and PIM in easy terms, well, I don't have the time or patience for it.
I can barely get you people to comprehend how harmonic distortion changes with level.
Still, Bob Cordell, like you Mikek, is a minimalist, and worked hard to compromise Dr. Otala at every turn. The AES paper is most probably from him. However, years later, Barrie Gilbert, of Analog Devices, (remember him) rediscovered PIM and gave it a good deal of significance, and a serious problem in op amp type designs.
 
#226
john curl said:
When Till asks me to give him the 'in and outs' of TIM and PIM in easy terms, well, I don't have the time or patience for it.
Click to expand...

Then why bother 'contributing' or participating on the forum??

In my experiance, people who really know their subject material have no difficulty explaining any facet of it in lay terms, without invoking the 'heavy maths required' fig leaf..... :smash:
 
#229
john curl said:
I can barely get you people to comprehend how harmonic distortion changes with level.......
Click to expand...

How patronising!!

Which 'you people' are you refering to here?? 🙄

I avoided that particular piece of elementary obscurantism....

This does not mean i did not 'comprehend' that this had totaly NOTHING to do with whether or not major loop feedback is desirable.... 🙄
 
#230
John,

I really don't want to give the impression that I want to attack you at every turn, because that's not my intention at all. But really! That AES paper (indeed from Bob) shows the guy went to a lot of trouble to building specialised test equipment just for the purpose of analysing PIM. He gave Matti credit where due, and went about his measurements painstakingly. He put his money where his mouth is, and I am not aware of any challenge to his paper. His conclusion was as I quoted.

On your side, we have nothing, nada, zilch so far, except a bunch of offhand statements.

I don't think you can honestly blame us for not being convinced by your case. Sorry.

One thing I learned on this forum early. This is not a nice snug club of your buddies admiringly looking up to your eyes for the next miracle. You deliver, or not. Simple as that.


Jan Didden
 
#235
Well folks, this is the situation:
It is good to go to the university and get a solid technical education.
With a good technical education, it is possible to understand the derivation of many distortion sources, many of which are important in audio design.
There are many technical books that are useful to understanding circuit design, provided that you are interested in the subject, and don't attack every opinion offered in the technical book, UNLESS you are prepared, with a good technical education and the mathematics, to back up your difference of opinion.
If you have a good technical background, then it is EASY to learn about specific distortion types, such as TIM or PIM, by doing a 'Google' search and reading what comes from it. Some of it, at least, will be at a level that you can grasp, even if you don't have heavy math skills.
As an engineer, I am normally interested in 'engineering'. This means that I use previously derived formulas in most cases to estimate my design and its performance. I also measure what I make and see if it matches my estimate.
Deriving first principles, doing heavy math, or re-inventing what is already in a book, is not what I am interested in doing.
SO, I tend to believe the formulas and equations that I can get from a book, unless I have a good reason NOT to do so.
Audio design is based on what the ear hears, ultimately, not what you want it to do, or what someone tells you is unimportant. It is the proof of performance that counts, and why I am well known in audio design.
 
#238
janneman said:
And what does it mean that it is only for 'weak non-linearity'. Does that exclude things like overloading or clipping only, or is there more to that??
Click to expand...

To see this, you'll need to go back to the Taylor series expansion of the nonlinear function that relates the output voltage to input voltage, from http://www.diyaudio.com/forums/showthread.php?postid=480771#post480771

We saw that the cubed term resulted not only in a third harmonic, but also a spurious signal at the fundamental frequency. In order to arrive at the result that the ratio of the third harmonic amplitude to the fundamental amplitude was proportional to the input amplitude squared, we had to neglect the effect of this spurious signal at the fundamental frequency on the amplitude of the fundamental. That's part of, but not all of, the "weakly nonlinear behavior" assumption. As the input signal gets larger and larger, the "third harmonic amplitude proportional to the input voltage cubed" assumption no longer holds because of higher-order terms in the expansion. When we truncated the Taylor series expansion to the squared and cubed terms, we were really making an assumption about the characteristic of the nonlinear transfer function. That assumption is that it's close to linear, and its deviation from linear is a smooth, gradual curve. If that deviation contains abrupt "squiggles", we'd need lots of terms in the expansion (a high-order polynomial) to represent those squiggles accurately. When you apply the formula:

cos(a)*cos(b) = 1/2(cos(a+b) + cos(a-b))

over and over again to the higher powers of the expansion to relate nth power terms to nth harmonic terms, you'll find that the fourth power term results in a fourth harmonic, a second harmonic, and a DC distortion term. Likewise, the fifth power term results in a fifth harmonic, a third harmonic, and a fundamental distortion term. So as the input signal gets larger, the third harmonic doesn't just come from the cubed term, but a combination of the cubed, fifth power, seventh power, ninth power, and so on.

So what kind of circuit would have a deviation from linearity that has "squiggles" that require a high-order approximation to the nonlinear transfer function? A class B output stage. You can see this with SPICE in fact. You can set up a sim of a simple class B output stage, and do a DC transfer curve on it with very fine voltage steps. If your simulator can export a file of output voltage vs. input voltage, you can bring it into Excel. There, you can perform a linear regression on the data, finding the slope and intercept of the best-fit of this curve to a straight line. By subtracting the data of the transfer curve from the best-fit line, you can see with very high resolution the deviation of the curve from linearity. You'll see the squiggles. Unfortunately, they're at nearly zero volts in. So for small input amplitudes, these squiggles are significant compared to the amplitude of the signal. We therefore need many high-order terms in the Taylor series expansion to represent them accurately, even when the input amplitude is small.

In other words, the "weakly nonlinear" assumption just doesn't hold for a class B output stage. The "weakly nonlinear" approximation assumes the third harmonic is due only to the cubed term, when in reality it comes from the fifth, seventh, ninth, eleventh, etc. Likewise, all these odd-order terms have a spurious term at the fundamental which causes compression (or expansion in some odd cases) which is neglected in the weakly nonlinear approximation.
 
#239
till said:
What exactly causes TIM and PIM? (please in laymans terms, so i´m able to follow)
Click to expand...

Hi till,

A good place to start is here http://users.ece.gatech.edu/~mleach/lowtim/bckgrnd.html

See also
http://users.ece.gatech.edu/~mleach/lowtim/instage.html

Here's a particularly interesting quote:

"Many of the critics of TIM disputed the design criteria that the open-loop bandwidth must be greater than the signal bandwidth to prevent TIM. After my article was published, I came to realize that this is true, provided the open-loop gain and bandwidth are varied in such a way that the product of the two remains constant. If this is done correctly, static distortions such as THD and IM can be reduced while not affecting the stability of the amplifier or its susceptibility to TIM."

In that statement, he's basically admitting that Otala was wrong about the wide open-loop bandwidth "requirement" for eliminating TIM. "The critics of TIM" are really "The critics of Otala". This admission is nothing new. He said the same thing in a class I had with him sometime around 1978 or 1979.

I am totally amazed that, more than 25 years later, this myth about wide open-loop bandwidth persists. It could be that some are worried about reviewers like this http://www.audioinsiders.com/htdocs/dcforum/DCForumID13/4.html#10 finding out that they really are using significant amounts of feedback.
 
#240
Andy C, the requirement for high open loop bandwidth is NOT a myth.
It is true that you can eliminate TIM, but you can't eliminate PIM without high open loop bandwidth. Matti Otala 'knew' that something was amiss, so he did not back down from the requirement of high open loop bandwidth, even when the slew rate requirement was met. It turns out that he was correct. PIM is as important as TIM. This is why many IC op amps with fairly high slew rates, can still sound relatively lousy.
If you don't believe me, well 'Live in ignorance!' ;-)
 
Status
Not open for further replies.
Top Bottom