John Curl's Blowtorch preamplifier part II

[Bob Cordell]

Originally Posted by Steve Eddy

Yeah, my 5.0 wasn't bad for just eyeballing off a jpeg image. And my actual measurement was 4.9 and some change I just rounded it up to 5 even. I wasn't looking at this in terms of intermod frequencies, but simply making relative measurements in inches based on fig. 3.

By the way, here are the approximations of all the unlabeled peaks in fig. 3.



se

Hi Steve,

Thanks for your effort here. This is very close to my earlier assertion that the frequencies were 2f2-9f1 = 1380 and 11f1-2f2 = 4980.

Throughout this exercise, we all must be mindful that this is not a high fidelity measurement by any stretch of the imagination. The Figure 3 example is pounding the cr@p out of a known cr@ppy op amp.

It would be far more interesting to see the same results for a modest-fidelity modern op amp under the same conditions. Maybe something like an OPA-604.

It is not the least bit surprizing that this kind of amplitude spectra can be created by fm modulation, but that only underscores the fact that with amplitude spectra alone it is virtually impossible to discriminate between TIM and PIM. It in no way supports the assertion that these sidebands show that there is PIM.

Cheers,
Bob

[Wavebourn]

Originally Posted by anatech

Hi Anatoliy,
With FM modulation, you get a continuous spectrum. With square wave modulation that is working perfectly, you will get what is called FSK, or frequency shift modulation. The only time you will have a discontinuous spectrum with FM that I know of is if you are sampling in some way. However, the center frequency (or carrier) is changing continuously between the peak high extreme and the peak low extreme of the deviation, or change in frequency. That continuous sweep back and forth "looks" like a continuous spectrum. I'm looking at an FM IF sweep right now, here it is.

Notice that it is continuous in nature. The input is a swept 10.7 MHz center, 800 KHz range. In other words, the high frequency is 10.7 MHz and the low frequency (modulation) is 5 Hz (inverse of sweep time). My earlier post showed 10.7 MHz modulated by 1 KHz, so take your pick from these. I can, and later will, try with the lower frequencies that this thread is concerned with. I don't have time right now and the generator I'm using (HP 8656B) only goes down to 100 KHz. I'll need to use an HP 3324A and 3336B in order to recreate the exact test setup used by John. That is why I have been asking for his test setup - so I can see what he is seeing.

Modulation in a linear analog power amplifier will be AM for the most part, FM modulation is what I think they are looking for, and may occur at very low modulation depths. Can you imagine listening to an amplifier that is being modulated using FM? That might be somewhat like listening to a turntable suffering from wow and or flutter. A frequency effect hard to miss.

Chris; you are right about sampling rate. Imagine very fast sampling when you amplitude modulate the signal. Do you see sidebands? No. Each measurement you take such a way show you the signal of the same frequency, but of a different amplitude. Similarly, when you observe FM such a way you see as if it is the single sinewave signal with slightly different frequency. But it is not what you see on a spectrum analyzer. Here is the picture of sidebands of FM (however, -180 degrees are not shown):



I've found this picture for you googling for FM sidebands:

http://www.radio-electronics.com/inf...m-tutorial.php

Edit: you may see that for FSK you've mentioned bandwidth will be needed is double deviation plus double baud rate. I.e. if you limit bandwidth of transmitter and receiver by double deviation you won't be able to transmit FSK.

Also, you may search for phase modulation methods to obtain SSB. They were (and are) still used in radio. The plus is, they don't require expensive mechanical filters, and such SSB modulators can be micro-miniaturized.

[john curl]

Wavebourn, .5 looks pretty good to me, for this. What do you think?

[Wavebourn]

Originally Posted by john curl

Wavebourn, .5 looks pretty good to me, for this. What do you think?

About what?

If you mean how nasty sounds mixed phase and amplitude intermodulation, I think you may get an impression listening to SSB transmission on AM receiver.

By the way, here is a SSB modulator designed by Russian ham radio operator A. Polyakov (RA3AAE), you see how it is easy to get a nasty SSB sound just summing clean and phase modulated signals (that happens in opamp-based amps easily):

[Steve Eddy]

Originally Posted by janneman

Steve, I had an appointment with Jan Lohstroh today and took the opportunity to go through his files from way back. Came up with this:

http://www.linearaudio.nl/Miscellane...poo%201976.pdf

I haven't gone through it in detail, just scanned and posted it, but it does offer some other perspectives on the DIM method and results.

jd

Sorry, Jan! I missed this the first time around and only just caught on what your bump post was about.

Thanks for providing the article. Though I really don't want to get in to the whole TIM/DIM/PIM debate.

It's just that y'all were arguing over what was shown in fig. 3. You'd previously labeled some of the peaks in it, and John questioned whether or not they were accurate.

So I thought I'd use some of my CAD/graphics skills to get a better fix on what frequencies the unlabeled harmonic components were.

se

[anatech]

Hi Anatoliy,
Thank you, I am very familiar with modulation types.

Now, just above the pictures you posted for us, there is some text that I feel is relevant and important to consider.

The total spectrum is an infinite series of discrete spectral components expressed by a complex formula using Bessel functions of the first kind.

Now, since you are offering that we are using a low modulation index, which I don't think is accurate here, you can look at two situations due to the use of a square wave. Remember that I am considering the input signal now.

First, an infinite series of frequencies (Fourier equivalent of a square wave) all modulated together by a since wave. Man, that's some kind of spectral mess you'll have there!

Second, a sine wave modulated by that same series from the square wave. Even limited by the 100 KHz filter, you will have your hands full figuring out that mess.

Look at the input signal. From the posted test circuit, you can see it's a linear adder. The resultant signal will be an amplitude modulated signal. If you pass this through a non-linear amplifier (A 741 qualifies for higher frequencies), you will produce a fair amount of harmonics added into the output signal. Okay, no problem. So, how are we producing a large amount of FM modulation here? Wouldn't the phasing effects be clearly audible if this distortion was as high as John says it is?

What I would expect to see is a small amount of FM produced. These should be at a low amplitude. Maybe even buried in noise compared to the AM products. The audible distortion produced by a 741 op amp does not have a phase component that is large. I've heard 741s in audio circuits often enough, and they produce more of a nasty harmonic type sound. Not that I was ever concerned with looking at them closely. Perhaps I will soon enough.

Imagine very fast sampling when you amplitude modulate the signal. Do you see sidebands? No.

Actually, I will disagree with you on this on a hunch. The energy from the carrier appears in the two sidebands. The higher the modulation index is, the greater the amplitude will be in the sidebands. I think I'll give that a try soon to make sure. If you can see the amplitude of the carrier reduced, the energy must be somewhere. It can't simply disappear. Maybe I misunderstand what you are saying here?

Hi John,

Wavebourn, .5 looks pretty good to me, for this. What do you think?

Are you assuming that because you want to see FM modulation, and this one single case appears to be, that this must be what you are seeing? As Anatoliy pointed out earlier, the two side bands would be 180° out of phase for the FM case, and the diagram does not furnish this information. There isn't enough information available to prove either case, so you're grasping at straws unless you know something we don't.

I don't have a problem with some small amount of FM modulation being produced by an op amp or circuit in distress, but I balk at the possibility of seeing such large amounts, comparable to the amount of AM distortion products. I guess that's my main beef with the interpretation you're reaching for. Not only that, but we've seen you're a little hazy on the details as well. No surprise, because the experiment happened so long ago and I believe you have lost all your original data and notes in the fire you had.

[john curl]

Lazy, ME! Here I am, 67 1/2 years old, half blind, 2 fingers asleep that make it difficult to type, and here I am, trying to help others understand a little of what I have found over the last 40 years. Think what I could do, if I found something important to do.

[john curl]

OH, you said 'hazy'. NEVER MIND !

[anatech]

Hi John,
Yes, hazy. Things that I did 30 years ago are pretty hazy, if not entirely gone from my memory.

Many of us have various problems. I come down with hands that go asleep with no warning, plus a host of other issues. Take your time and post as you can. There are days that a few posts represents all I can do for that day, and it has nothing to do with being lazy.

Anyway, John. There was a test setup posted a short while ago. Does this look like the setup you fellas used. If not, could you, or would you let me know so that I may try some of this out? What order of magnitude would you expect the FM (or phase) products be?

Thanks John, Chris

[Wavebourn]

Originally Posted by anatech

Hi Anatoliy,
Thank you, I am very familiar with modulation types.

So you should know the taste of a small spoon of SSB added to the whole bucket of honey of a clean sound.