Ah, ok. They were just talking about the regular intermod series. I thought you meant they'd actually identified some of those peaks as specifically being what you'd labeled them.
Thanks for the clarification.
se
As John Curl once said, they've found an asymmetry of sidebands. I.e. heights of strips are different while in case of pure IM they should be equal, that means additionally to IM there is some PM.
Thanks for reminding them, Wavebourn, but the sidebands are symmetrical, which implies FM. However, back then, while I was a participant in the original testing, 34 years ago, I did not notice these 'extra' pips as being anything special, either, until the possibility of FM modulation was pointed out to me, recently.
I think that Matti first mistook the added pips to be the 2F2-NF1 series, because the 2F2-9F1 seemed to fit fairly closely. It is just that some of them DON'T fit closely enough. It was just the wrong path to take, but all pioneers go up 'blind alleys' some of the time. Everyone, please understand that.
In those days, before personal computers, and the smallest practical computing device was the PDP-11, we didn't use computers at all, and this test was performed by an HP3581, and some sort of mechanical plotter.
It is a good machine, but all analog, so far as I know. I have one sitting in my closet, and I loaned SY one, years ago. He still has it, I guess, and unless he decided to use it for a 'boat anchor' or door stop, it still does a respectable job of measurement.
I think that Matti first mistook the added pips to be the 2F2-NF1 series, because the 2F2-9F1 seemed to fit fairly closely. It is just that some of them DON'T fit closely enough. It was just the wrong path to take, but all pioneers go up 'blind alleys' some of the time. Everyone, please understand that.
In those days, before personal computers, and the smallest practical computing device was the PDP-11, we didn't use computers at all, and this test was performed by an HP3581, and some sort of mechanical plotter.
It is a good machine, but all analog, so far as I know. I have one sitting in my closet, and I loaned SY one, years ago. He still has it, I guess, and unless he decided to use it for a 'boat anchor' or door stop, it still does a respectable job of measurement.
Last edited:
Sorry John, I misunderstood you. I thought you were referring to asymmetrical pips that may signify presence of PIM, but I don't understand how some extra pips may signify PIM. For more than 2 pips to be visible in case of PM there should be horrible deviation that can't be produced by the device in question. And as soon as both IM and PIM present pips should have different height, due to compensation.
Last edited:
It is really easy to see this. Just print out Fig 3, especially at 2 times size, then FOLD IT BACK, on itself around F1 or 3.18KHz. You WILL find symmetry in the contested distortion byproducts. This hint is due to Jocko, a former contributer here, and still one of the best engineers I know.
Why I should not predict John if I know what happens when PIM and IM are mixed together? Do you remember, I told you that I've experimented with different kinds of distortions in order to create musical signals? Do you remember I told you that when I made PIM device for a bass guitar nobody liked the sound because it was horrible? I learned radio and electronics then, trying to apply knowledge to synthesis of sounds and sound effects. We did not select which disciplines to take; it was impossible since an education system was carefully planned and everything prescribed was compulsory. I wish we had a choice, in such case I would take more of related to radio and electronics disciplines instead of humanitarian ones, like CPSU History, Philosophy, Economics of Socialism, and so on. I barely remember Economics of Socialism. But I remember well how to create SSB using a phase modulator. Just mix them in proper proportion until one pimp and a carrier disappears. When proportions are in between pimps have different heights.
Last edited:
Pimps are only coincidently of the same height and usually are not FM or even AM. We have quite a few here in Toronto downtown around Jarvis St.
No matter how do you cal that strips on the paper, but it is a forum. Be it a democratic institution we could vote for PM or against PM, be it a tribe we could ask our Elders to decide, but here on the forum some physical evidences have to be presented. I believe that PIM were present in that experiment because I know how to create it, but I don't see a proof of it's presence on the graph.
The best cellphone I ever had was long like an energy bar gray Siemens, rectangular one, like 1x2 inch, about 4.5 inch long, in 1999, in South Africa. It was GSM then over there, and all telephony was digital. Sound quality was very good. Car kits were required. All of them were wired, no blue teeth, just jack plugs. No echo, no intermittent loss of signals, and there was an option to see on display which cell station it is connected to currently.
I guess I'm going to have to read more on this. From what I know, both AM and FM modulation produces sidebands of equal amplitude. Given normal instruments, there is no easy way to see if the lower sideband is 180° out of phase with the upper sideband.
We do know that pure AM modulation will give you a carrier and two sidebands, and that's it. FM modulation will generate increasing sidebands as you increase the level of modulation. This may be the best way to determine whether you are dealing with an AM or FM issue. Only very low deviation FM modulation will generate one carrier and two sidebands.
-Chris
We do know that pure AM modulation will give you a carrier and two sidebands, and that's it. FM modulation will generate increasing sidebands as you increase the level of modulation. This may be the best way to determine whether you are dealing with an AM or FM issue. Only very low deviation FM modulation will generate one carrier and two sidebands.
-Chris
Can we expect wide deviation from an opamp?
I don't think so. That means, we can't see more than one pair of strips generated such a way.
But since both AM and FM present, their lower sidebands will be out of phase, upper sidebands in phase, and as the result we'll have higher upper sideband than lower one. This would be the chance to prove presence of PIM.
My method is to push the topology into the direction to highlight the phenomenon in question. How? To conduct series of experiments with different phase parameters, and added non-linearities, then see the whole picture in dynamics.
As the result of experiments I know what to do in order to minimize the phenomenon in question.
Here is one way to avoid PIM generated by a MOSFET with feedback: to use feedback in phase with it's own non-linear capacitances, so errors and feedback would be aligned in phase, and no PIM generated by feedback:
Please note folks, the basics are still not agreed on. John, as Gerhard showed ALL the tones are EXACTLY on NF1+-MF2.
Hi Scott,
That's exactly what my earlier line of questioning to John was all about. With no targets in sight, we're simply spinning our wheels. At least with a verified test circuit, we can see what comes out, therefore confirming or disproving any previous statements in that direction. From what I've read and understand so far, this distortion effect seems to be in a special case. Only when the amplifier in question has been driven to the point where it's gain starts to compress will any trouble come about. That's assuming that the amplifier will suffer gain droop at high output levels before gross distortion sets in.
Hi Anatoliy,
We both agree that the actual amount of this particular distortion created is very likely to be a very small amount. So small in fact that sidebands may not appear above the noise floor for starters. The entire argument is really premature until we have a sense of magnitude. I suspect that all you may see is AM modulation in nature, the FM contribution being so small.
So how is your testing method different from anybody else? It seems to be the natural way to test.
You know, the first time I saw the acronym "TIM" applied to distortion, I found it to be pretty amusing. Why? Well, because "TIM" stands for Technician Induced Malfunction. Now, that's pretty funny if this turns out to be an error in test method!
-Chris
That's exactly what my earlier line of questioning to John was all about. With no targets in sight, we're simply spinning our wheels. At least with a verified test circuit, we can see what comes out, therefore confirming or disproving any previous statements in that direction. From what I've read and understand so far, this distortion effect seems to be in a special case. Only when the amplifier in question has been driven to the point where it's gain starts to compress will any trouble come about. That's assuming that the amplifier will suffer gain droop at high output levels before gross distortion sets in.
Hi Anatoliy,
We both agree that the actual amount of this particular distortion created is very likely to be a very small amount. So small in fact that sidebands may not appear above the noise floor for starters. The entire argument is really premature until we have a sense of magnitude. I suspect that all you may see is AM modulation in nature, the FM contribution being so small.
I don't understand where you are going with this to begin with. I do not expect a large amount of PIM or TIM from any amplifier, chip or otherwise. As I mentioned, I intend to read more information on this before making a factual statement of any kind. I think I want to run the experiment myself to see what the result actually is.
You really lost me here Anatoliy. Each sideband will be displaced from the center frequency by the same amount - the frequency of the modulating signal. The amplitudes will be the same as an absolute value, we can't see the phase when using a spectrum analyzer.
What topology are you talking about?
So how is your testing method different from anybody else? It seems to be the natural way to test.
Well, why else do the experiment? Mind you, the mechanism may not be so obvious. You may well find out that this particular thing is not of any concern as well.
I see a current generated feedback voltage in your diagram. Besides, I don't think anyone has said that PIM is created in the feedback loop. I understand that the issue exists in the forward amplification path.
Didn't this experiment take place before DSP was in significant use? In other words, I don't think John and company used any DSP techniques. Therefore, we are not concerned with DSP in any way unless you want to use these methods to uncover the distortion products. In that case, it's doubtful that anyone on that team saw any PIM or TIM.
You know, the first time I saw the acronym "TIM" applied to distortion, I found it to be pretty amusing. Why? Well, because "TIM" stands for Technician Induced Malfunction. Now, that's pretty funny if this turns out to be an error in test method!
-Chris
Chris; if you are talking of PIM you've lost AIM. If you are talking of AIM you've lost PIM. If PIM is generated by non-linearity that means AIM must already present, so if you sum their products you will get asymmetrical specter. It is well known in radio fact, and let me repeat again, it is the method of generating SSB signals well known in radio, one of basic things.
Any topology. Did you read the paper Jan Didden posted?
It exists in both; but M. Otala's and J Curl's paper deals with PIM generated by an amp with no PIM, but with a phase lag, when a negative feedback is applied. "Fixing" non-linear distortions it creates PIM, as stated in that paper.
I was kidding about DSP.
Hi Chris,
Anatoliy is correct about the sidebands being of unequal amplitude if both am and fm at the same modulating frequency are present. I believe I also mentioned this several posts ago. Because the upper and lower sideband phase relationships are different for am and small-deviation fm, the sidebands will augment each other on one side and fight each other on the other side.
On another note, I do expect a lot of TIM and PIM from some amplifiers under some conditions. The 741 on which Fig. 3 was based is a very good example. That op amp is being crushed, and all kinds of bad things are happening.
Matti was probably right to choose that amplifier to make clear what he was talking about, but that is not a good example upon which to base a high-end audio discussion.
Cheers,
Bob
Yes, but in order to understand where is the right direction sometimes it is good to see an opposite side, since errors are visible to everyone, but clean reproduction may be questioned by some who expect specific sound to be reproduced. For example, my wife did not object sounds from mass production systems until she realized at once how close reproduction may be to real sounds. The same happens to other people who don't hear the direction to clearness, but hears well distortions, and sees them on spectrum analyzer screens. Show them the wrong direction, and they would not object the opposite way. However, I have to admit that there are more of wrong ways, but the single right one.
- Status
- Not open for further replies.