What is causing this audio highs attenuation/ distortion problem?

[K1JJ]

Finally got to the point of running THD tests with Fabio II. (4-1000A plate modulated rig)  The rig is near to completion, though will get some more fine-tuning attention soon.

Using a one volt, low distortion sine audio waveform into the rig and using my HPSDR receiver to view RF spectrum, here's what I found:


1)  Loading the class C final heavily definitely helped to make THD better by as much as -10 dB. The groups of higher order products up the band became "better behaved" by making a nice orderly decline.  When loaded lightly there is more chaos in the groupings and at a higher level.  Just tuning C2 loading cap could actually find the sweet spot.

2) I found that running a triangle wave is a very interesting frequency response test as viewed on the SDR screen.

3)  Using a pure, low distortion sine wave, I found that at ~100% modulation, the THD generally ran about -40dB 3rd and -50 dB 5th order levels. The 7th order was always lower than the 5th, but was dependent on loading and tuning.  I am very happy with these numbers.

4) I ran some THD tests up at 8 - 10Khz and saw that the THD numbers were not as pristine, but still close to -35 dB 3rd order.   At 6-7 KHz I was seeing about the same THD as the lower frequencies, which is good.  At 12 KHz I did not see any unusual crud up the band and tells me that it is stable.

5)  I tried the THD tests with and without 8 dB of NFB.  With it on, I saw 3rd and 5th IMD generally better - and the farther out crud was better behaved and grouped well with an orderly decline.  Without NFB the higher order crud was higher and somewhat random looking. I would say the NFB helped to improve things maybe -4 to -5 dB at most.  This is using a sine wave, so TIM (transient Inter-mod)  would not show up anyway, only with voice or music programming.

6)  The deep lows swept very cleanly at about 20Hz minimum, which is FB for a rig with a mod transformer.

7) I ran some temperature tests with the laser temp gun and see that the 4X1 plate caps are running about 90 degrees F.  I have max air flowing for tests and this temp will increase once I back the Variacs down for on-air use.

I tried some voice modulation with plenty of S's  and CH's and the spectrum was smooth and contained.  No more spitting and highs distortion or splatter could not be seen.  I plan to add the CRL PMC-300A modulation limiter next week which will limit negative peaks to -95% and we'll have to see how far the positive peaks run in AB1.  My goal is to run the modulators as cleanly as possible, so no grid current is permitted. This will be the bottleneck in performance.

When the CRL limiter arrives, I also will try some aggressive voice audio tests to see if the NFB comes unglued. I want to run as much NFB as I can but still be stable.  (Probably limited to -8 dB)

All in all, I couldn't be more pleased with the test results and it looks like this may be the cleanest plate modulated rig I've built to date....  

Thanks again to all who have helped!

T

[w3jn]

Tom, not sure about your test.  Are you measuring THD by using a triangle wave input?  If so, all you're measuring is the audio response of the transmitter.

Remember that a triangle wave consists of many odd harmonics of a pure sine wave.  If you're simply measuring the amplitude of those odd harmonics with your SDR you're just measuring how the transmitter is affecting its response to a particlular component of the triangle wave.

A triangle wave is useful to see, in time domain, transmitter distortion as you can easily see bending of the wave's slope, rounding of the edges, etc.

In the frequency domain, however, you're best served by feeding the transmitter a pure, low-distortion sine wave and seeing what harmonics of that sine wave you get.

As you increase the modulation index of the transmitter, a more useful measurement might be IMD.  IMD is much more irritating to listen to, and while you can relatively easily see IMD in the frequency domain given an undistorted input signal, a THD test (generally) will not tell you a thing about IMD.

[W2VW]

  IMD is much more irritating to listen to,

Is he another one of those guys who has a few too many drinks and gets on 75 looking for a fight?

[Steve - K4HX]

 

[K1JJ]

In the frequency domain, however, you're best served by feeding the transmitter a pure, low-distortion sine wave and seeing what harmonics of that sine wave you get.

Hi John,

Yes, I've been using a low distortion sine wave for all of my AM tests now and in the past.  But just for the first time I looked at a triangle wave on the SDR and thought it might have some additional THD value, but I guess not.

One thing that cornfuses me...  On AM, the single sine tone gives information about THD.  Do we need a standard two-tone on AM to show the IMD just like on ssb?   When I ran a two-tone, it looked about the same as a single tone but with more peaks, as expected.   If the rig is very clean, can I expect a big difference in general peak levels between the two tests?

The AM carrier seems to supply a tone to beat against on AM, whereas a ssb signal does not. That's what cornfuses me.  It seems like the carrier acts like a supplied note for the two tests. Understand what I mean?

* I edited my post above to remove my comment about a triangle and THD -  to keep it accurate.


A friend commented:

"The reason for using two tones is in a nonlinear condition they mix to generate third order responses. Using 1 tone you would just generate harmonics of the tone. I bet a single tone is cleaner to 6 db above a two tone. The two tones not phase related gives you the most information because the vector addition generates modulation from 0 to 6dB above 1 of the tones.  I suppose if you have a clean system the results will be similar."


Think I will try some 2-tone tests (on the AM rig ) and see if the IMD tells me more.  When I tried it a year ago, it looked exactly like the single tone, but with more peaks filled into the same slope.  So I didn’t bother anymore with it for AM, just for ssb tests..


T

[K1JJ]

Just ran a 2-tone IMD test on Fabio II using an Electcraft  2-T two-tone board.  The tones were 700 Hz and 1900 Hz.  This board has good specs and a clean waveform.

At medium power, the 3rd order IMD was down about -35 dB and the 5th order was down about -55 dB at 100% modulation.  Above + 5KHZ up the band all peaks dropped off below -55 dB... quite well behaved.  I would need to generate audio frequencies much higher than 700/1900 Hz to test the extreme highs IMD, but this is a good start.  

I did notice the same THD pattern where heavier loading helped improve the IMD.   The improvement was slightly less than the effect it had on THD, but still showed the same trend.

For a class C, plate modulated final, I am pleased, since these IMD results are slightly better than an average ham ricebox  or ham linear amplifier.

Thanks for pointing out the THD/ IMD  information, Johnny.

T

 

[AMLOVER]

Tom,

what is your modified heissing configuration at this time?
Do you return the modulator's cold end to the HV+ or you ground it through the cap?

Stefano

[K1JJ]

Tom,

what is your modified heissing configuration at this time?
Do you return the modulator's cold end to the HV+ or you ground it through the cap?

Stefano

Hi Stefano -

I have the Heising cap connected to the cold end of the mod xfmr and the other end of the cap connected to the cold end of the Heising reactor.  The "cold" end of the Heising reactor goes to HV+.

BTW, your idea about using a HV relay and power resistor across the mod iron and screen choke to quench sparks is still working FB. No arcing problems since. That was you who recommended this last year, right?

T

[flintstone mop]

I also always thought it was a waste of time to run 200% modulation.
Phase the audio right, run it up to 90% negative and something like 120% positive.
Anything else is just distortion, no?

Some people on the air do the big carrier and do not push the modulation.
I remember when FM broadcast started. We had WYSP in Philadelphia, who played great music, and the DJ's spoke softly and at reduced volume. It was wonderful.

I guess its good to want to be very clean and hi fi, but I think its all in the EQ and the voice.
You can have a great sounding signal with some distortion (plenty really) if the EQ is good or the voice is a radio voice.

One of my favorite signals was a guy over in PA, forget the call, but he was running an old Sure S40 mic into a stock 32V3.
Sounded like god.
 

That's about it for me too. It's always nice to have a lot of headroom, and that makes the TX audio a lot nicer. Tom wants a lot of operating space for the modulator/transmitter so there are no bottlenecks for clean RF and clean sound on the air.
Fred

[N2DTS]

I know he has fun fine tuning his rigs for clean and stable operation.

Too many others do no tuning or checking at all!

[W3RSW]

KF1IMD again? w2VW already whacked his chops. 

[K1JJ]

I also always thought it was a waste of time to run 200% modulation.
Phase the audio right, run it up to 90% negative and something like 120% positive.
Anything else is just distortion, no?

Agreed.  There's no doubt in my mind that for the majority of receivers on the air (mostly diode detectors) the cleanest modulation level  is -95 negative and whatever your voice naturally does for positive peaks. This may be 105%, 110% or 120% positive, etc.   This level will satisfy the majority of receivers without excessive distortion.

Chuck said it well in an email... to paraphrase:  We usually have a certain limited amount of CLEAN audio power to work with.  (or pep RF power)  Why not run the carrier up so that the maximum audio power makes 110% positive?  This creates a better quieting effect and S/N ratio when using a bigger carrier.   In contrast, dropping the carrier way back and running the same amount of audio power (at 200% mod, etc) doesn't make sense, since the background S/N and S meter become lower.  Selective fading also becomes a big issue with heavy modulation for diode detectors - unless we are using sync detectors.

I plan to run a fat carrier and use limiting to keep at -95% negative  and protect the mod iron from rouge positive peaks above 130%.


Fred, yes, correct.  I need headroom so that I can run the modulators in AB1 - I cannot draw grid current, thus power output is lower. This is the bottleneck. So I need big tubes that can do their job cleanly.  I COULD draw grid current (AB2) but the THD would suffer based on my tests.

T

[N2DTS]

For me, its all about the carrier.
You can get ssb to sound good frequency response and distortion wise, but its got no carrier to quiet the frequency.

I remember my early days into AM, when I would hear someone with a strong signal key up and you could hear the clock in the background, or the guy next door with the lawn mower.
I do not seem to hear that much anymore, its mostly blowers or nothing at all.

[W2VW]

I remember my early days into AM, when I would hear someone with a strong signal key up and you could hear the clock in the background,

W2KTU 160 meters.

[Chuck...K1KW]

 "you could hear the clock in the background"


Like the original W1ZE on 75M in the 50's and 60's?

[KB2WIG]

And then there are the crickets and dishes being washed.


klc

[w1vtp]

 "you could hear the clock in the background"


Like the original W1ZE on 75M in the 50's and 60's?

Yup, Chuck, remember that well.  Fond memories.  Met Irving while he was visiting my aunt, W1UET's camp one summer.  I was a snotty nosed kid at the time and was impressed with his gentleness and good humor.

Al

Attached his QSL coming up on the 60th anniv of our QSO.  Note the address: how simple life was back in those days

[K1JJ]

Fabio II is coming along very well.  Just added  the PMC-300A and getting it to work. Should have it on the air and will post some last results soon.

* I do some Kayak fishing in the CT River, but not like this guy.  It's amazing the size fish some of these guys land in their little Kayaks from the ocean.

[I couldn't resist]

T

[N2DTS]

My pmc 300 arrived today, and was a bust.
Loud hum, very low output, when the gain is adjusted, it trys to blow my mod transformer with really loud low frequency garbage.

The thing is loaded with cold solder joints and bad connectors...

[K1JJ]

My pmc 300 arrived today, and was a bust.
Loud hum, very low output, when the gain is adjusted, it trys to blow my mod transformer with really loud low frequency garbage.
The thing is loaded with cold solder joints and bad connectors...

Aw, that's too bad. Sorry to hear.  Hopefully you can get a refund.

It might be worth giving it another try.   Jeff / W2NBC gave me quick start instructions to get mine running and I'm very pleased with it!  It's limiting the negative peaks to ~95% and I put a limit on the positives to about 130%.  I can scream into the mic and it stays within parameters.  The rig sounds wonderful in the monitor.  This was the crowning touch and Fabio II is now the best plate modulated rig I've ever had.

It raised my background blower noise very slightly simply cuz the audio level is higher since I can lean on the NPL now.  It seems quite transparent and just what I needed.  I played with the 4X1 modulator negative feedback and brought it up to -12 dB without problems. Tmw I will do some THD tests to see what setting is best.  Maybe -8 dB or so.  

So in summary, the PMC-300A is a worthwhile box if you can find a clean working one.

And again, my thanks to Jeff for his email assistance!  


T

[Opcom]

Is this a decent summary  about the frequency response, distortion, and feedback ?

It was the broken ss driver, for the actual repair that brought this into the open when the transmitter was tested afterwards.

The negative feedback around the modulator (taken from the modulation transformer secondary), switched in and out, accounted for 10dB gain.

Changing the .002 coupling and .002 plate bypass caps to 500pf each to remedy need for a lot of extra modulator drive and a slightly higher THD above 6500Hz, and reduced the modulator current from 700-800mA to 400mA @7KHz.

Changing the 1000pF screen bypass caps on the RF stage to 500pF to reduce phase shifting on the self-modulated screen.

Large caps like 0.001 or 0.003 are not to be used across the modulation transformer.

Transient instability in the modulator (coming from too much NFB and happening when the modulator is overloaded/clipped) can blow the modulation transformer. the maximum feedback should be more than 6dB less than the amount at which transient instability is seen. 3-6 dB of feedback may be enough around the modulator stage, 8dB max.

NFB signals should be taken from the modulation transformer primary because of phase shifts and leakage inductance in the transformer.

NFB can be taken from just one modulator plate. Or both. The effect should be the same.

[K1JJ]

Yes, I would say that's a good summary, Pat!

The only modification  listed that I don't think made a big difference is changing the screen caps from 1000 pF to 500 pF.  The screen was modulating well either way. 

The biggest improvements were repairing the SS driver followed by reducing the .002s to 500 pF in the plate coupling and choke bypass.


I have yet to see the effects of the NFB going unstable due to a transient, but with the new PMC-300A limiter and keeping the NFB under -8 dB, maybe I won't have to experience it.

T

[AB2EZ]

I'm trying to understand this issue:

The gain of the amplifier, with feedback, is: A/(1 - AxB), where A is the open loop gain and B is the fraction of the output that is fed back to the input.

If B were to go to zero, then the gain would jump up. But B is typically determined by passive components.

If the open loop gain saturates (I.e. the slope of the output v. the input gets smaller)... but does not introduce phase shift as it saturates... the closed loop gain will also saturate. I.e. the change in closed loop output corresponding to a change in input will get smaller.

For example:

If A=-100 and B=.05, then the closed loop gain is: -100/6 = -16.7

If the slope of the open loop gain decreases from -100 to -25, then the slope of the closed loop gain will decrease from -16.7 to -25/2.25 = -11.1

If the slope of the open loop gain decreases to -10, the the slope of the closed loop gain decreases to -10/1.5 = - 6.67

More generally:

The closed loop gain = A/(1 - AxB) = -1/[(-1/A) + B)]; which decreases as |A| decreases.

Therefore, it appears (to me) that a problem would occur only if:

A) The saturation of the open loop gain is accompanied by a significant increase in phase shift.

or

B) There is an amplifier in the feedback path whose gain drops significantly for some reason.

Stu

I have yet to see the effects of the NFB going unstable due to a transient, but with the new PMC-300A limiter and keeping the NFB under -8 dB, maybe I won't have to experience it.

T

[Steve - K4HX]

I've never heard of an amp become less stable with the application of negative feedback.

[K1JJ]

I appreciate the the math, Stu.  It appears from what you are saying there should not be a problem if there was a lot of NFB (-15dB for example) and the amplifier went into an overload (flat top) situation.  I don't know enough about this nor have seen it occur so far.

Talking with an amplifer engineer, he says that the Dynaco series 70 amplifier has -23 dB of NFB.... and there are a few BC transmitters (Gates BC-1T through the BC-1H)  that use =14dB around the 833As.  

I could make a mod to let me dial in some more NFB and may play around with it once I am sure the rig is very stable as is.   There is always the risk of having the audio system take off due to too much phase shift. That has happened before when I tapped off the mod xfmr sec. But tapped at the plate cap, I've not seen the limit yet.

Today I had the rig on 75M for an hour running about -12dB of NFB. Got some excellent audio reports. I didn't see any signs of NFB instability so far.  The extreme highs were very clean.  Looking at the SDR spec analyzer, when I hit an "S", the spectrum was contained to my pre-set limits. When I set the high cut to 6Khz bandwidth, it WAS 6 KHz and rolled off sharply.  In the past, I could hear tearing and would see the noise floor rise up way above 6Khz. That is the advantage of having clean extreme highs.


T