EAM - - Enhanced Amplitude Modulation article by W2WLR

[W2INR]

I received the following PM from Don K4KYV . It is an article written by Watertowns little radio W2WLR  (The PM is edited )

Gary,

I just scanned a copy of  "EAM, Enhanced Amplitude Modulation" written by George, W2WLR many years ago.  Hoisy, W4CJL (SK), founder of SPAM sent me this copy sometime back in the 1970's, pre-dating George's articles on audio that appeared in The AM Press/Exchange.

George's theory of audio and the human voice in quite interesting, and this article influenced how I set up the audio curve on my transmitter, although I cut off at a somewhat higher frequency than George recommended.

I think a lot of readers might find the article interesting.

73,

Don k4kyv

The article is in PDF format.

http://amfone.net/Tech/EAM.pdf

Thanks Don for sharing this

G

[WA3VJB]

I always thought George's audio sounded thin and kind of muffled.
When he asked for an audio report and I told him as much, he hit the roof.

All I said was that there was lots of upper midrange, sort of like a typical two-meter rig, but that I could tell it was lacking bass because the register of his voice clearly suggested there was a lot more low end than what his setup was able to pass.

This article explains why.

Not something I would aspire toward, that's for sure, although Don you've improved upon what George may have been trying to accomplish, both in the high and low ends, from having heard you in person.

[k4kyv]

I think his theory about the human voice is interesting, where he says that it scans repetitively across the band of frequencies in the voice range, and that it is the scan rate that determines whether a person's voice is high or low pitched, not the distribution of frequencies over that person's voice range.

He once told me he picked up that theory from something he read that was published by Bell Labs, although I have never seen or heard of anything on the topic other than from George.

In my opinion, he cuts off the upper end of his response curve at too low a frequency, but remember, this was written in 1970, during the era of the spectrum conservation crusade.  And, as I recall, he used a D-104 microphone, which may explain his lack of lows since, as shown in Fig. 7, his response curve is flat down to 80 Hz, with the presence rise beginning somewhere in the lower midrange.  In my mic preamp, the response begins to rise at about 800~ or 900~ and rises steadily up to 9 dB somewhere beyond 2 kHz, where it flattens out like a plateau to stay uniform up to the cutoff frequency of my low-pass filter, which is switchable between 3400~ (sharp cutoff) and 5000~ (more gradual cutoff, with complete attenuation near 7500~).

Does anyone know what ever happened to George?  As far as I know he hasn't been on the air for years, but I haven't heard anything about him going SK.

[k4kyv]

Hoisy proved that it is indeed possible to have very restrictive 'quiet hours' imposed by the practice of 'letting it rip' whenever you desire. Hoisy disappeared from the air for a long time before becoming a SK. Obviously 200+% pos peaks will not always serve one well.

To my knowledge, Hoisy's "quiet hours" were not imposed by the FCC, but by his XYL.  She was very demanding that he spend his prime evening hours with her usually playing cards or watching TV, plus some of the telephones he interfered with belonged to her friends.  I think his rig was clean of harmonic TVI.  It was audio equipment that he wiped out, and the subdivision where he lived was densely packed.

[flintstone mop]

I have the ultra modulation articles Hoisey sent to me when I was a fledgling AMer in the early 90's. His pics were published in ER when he became SK. He had some very big rigs in his shack. I mean they looked pretty serious! 4-1000's and HUGE tank coils + the 200% modulation. What I have been reading on this subject is that the typical Rx can't handle more than 130% pos peaks anyway.
Oh yes, tell me about the audio processing. I have heard fellow ops on the air and when I met them in person, could not connect the big broadcast voice I heard in my radio to the fella I met.
Fred

[k4kyv]

Hoisy may have got involved with a court case before he went SK.  But he endured the XYL-imposed quiet hours (no operation before 10:30 PM) from the time I first started talking to him over the air in 1970.  I think his absence from the air in later years was due to his frail health.  It is my understanding that his wife developed Alzheimer's and Hoisy was taking care of her even though he was well into his 80's.

I believe she was eventually moved to some facility in Florida near where their son lives, and at the son's insistence, Hoisy reluctantly moved to Florida as well.  He sold or gave away all his big radio stuff and took a transceiver with him, but I don't believe he ever got on the air with it before his death.

His son is a ham and supposedly interested in AM, but I don't think he is very active.

[WA1GFZ]

http://www.tech-systems-labs.com/books.htm

go to this site and download the 1959 handbook with good information. Also notice the SSB section and you will see  a QSD mixer done with tubes that looks similar to the H mode mixer.

[WA3VJB]

I recall the call/name of W2WLR/George but cannot recall what he sounded like?

Don't think I have any recordings of him.
His voice reminded me of Paul Harvey, suggestion of heavy jowls and rotundus resonance.

I remember he was very much into corn.  Regardless of where the QSO happened to be, it ended up about corn.  Odd.

[KB2WIG]

The last time I heard Geo. was on the 2m CB, moblie on I90 near smAlbany, ny... this was sometime in the late 90s....klc

[N3DRB The Derb]

wasnt his phonetics Watertowns Little Radio? His super modulation always sounded weird, my diode detectors simply couldnt handle it.

[k4kyv]

There are so many varieties of "super modulation" to consider.  Ultramodulation is nothing but a little bit of controlled carrier combined with substantial even harmonic distortion.

I tried the classic circuit sometime back in the early 70's.  The signal reports I got were that the audio wasn't any louder, but definitely more distorted.  Besides, audio power is too difficult and expensive to generate, to waste heating up a big power resistor.

What was effective was symmetrical high level speech clipping, by deliberately driving the modulator stage into flat-topping on both positive and negative peaks, inserting a low-pass audio filter between the modulation transformer and class-C load to suppress the splatter.  But it was rough on modulation transformers - I blew about a half dozen kilowatt size transformers in a period of 6 months.  I finally gave up on the idea, removed the splatter filter, cut back on my audio, and haven't blown a mod xfmr in over 30 years.

And besides, even though it was effective under "battlefield" conditions, the sound bordered on "space shuttle audio" when the modulator was driven to heavy clipping.

[WA3VJB]

Oh yes, do check the characteristics of the diode.

Asymmetrical modulation gets gnarly on the diode of the R390A after about 130 percent positive.

This type of envelope can also be viewed as reduced-carrier DSB although people rarely call it that.

Bill DUQ, and I think Chuck WA1EKV (or whatever his new call is) have both dialed up and back for me to actually hear it getting gnasty past 130%, with lots of ugly distortion by 140%

Clicking the BFO "on" at zero beat restores the proportion of carrier and cleans it right up.

[Pete, WA2CWA]

I think his theory about the human voice is interesting, where he says that it scans repetitively across the band of frequencies in the voice range, and that it is the scan rate that determines whether a person's voice is high or low pitched, not the distribution of frequencies over that person's voice range.

He once told me he picked up that theory from something he read that was published by Bell Labs, although I have never seen or heard of anything on the topic other than from George.

Back when I was on the initial Merlin project with the Labs, we pulled a number of articles on the theory of human voice from the early Bell Labs Technical Journals to help the developers design a new acoustical handset and microphone for the project. When I moved back into data, I dumpstered all that stuff. From what I remember, it was pretty interesting stuff.

[Bacon, WA3WDR]

WLR-George's thing with corn was that it was very bad for people to eat.  He blamed corn (in the form of maize) for the failure of the Aztec and Mayan civilizations.

Yes, he has a sort of Paul Harvey voice, but maybe half an octave or an octave higher in pitch.

I remember that George had some acronym for his AM system idea back around 1970, but I never could remember what it was, and either could he.  EADS might be it.  He generally believed that the voice spectrum should be flat, meaning flat spectrum, which requires huge higher-frequency boosting.  I think he did some sort of speech EQ on a two-meter repeater, too, but I never heard it.  I am finally finishing scanning AM Press/Exchange issue 78 (12/1989) and there is an article in it from George about speech equalization.

Once I got the RF out of my mike audio and added the series screen resistors to my modulator tubes, I got a "very natural" report from KNX Deano with my 180-200% positive modulated Viking-II.  The receiver IF has to have the headroom, and the receive AGC has to be the DC-average type, like old tube receivers normally have.  Receivers with peak-oriented AGC will make AM sound badly compressed, and more modulation only makes more bad compression on the receive side.  This is the typical sound of a modern solid-state receiver, and it can be greatly improved by turning off the AGC and using manual RF gain.  Changing it to average AGC would fix the problem as long as there was enough IF headroom.

Also, thermionic diodes detecting over ten volts of IF output are more linear than silicon or germanium diodes detecting an IF signal of around one volt peak to peak.  An active-linearized solid-state detector would be better for envelope detection.

If the supermodulation involves negative peak clipping, well of course it's going to sound distorted.  If it involves the 'upside-down' tube, or a DC-offset balanced modulator, it will sound even more distorted on an envelope detector, although it will be perfect with a synchronous product detector.  I have come to favor voice-polarity supermodulation, because it does not involve transmission nonlinearity, harmonic distortion, etc.  Just polarize the speech the right way, and don't overmodulate in the negative direction.  Asymmetrical peak limiting with a dual or triple time constant is best for transmit-side speech compression - it is remarkable how bad the old-style single-constant symmetrical peak limiting really sounds, in comparison.  Probably the best way to go would be an unbalanced-balanced modulator and voice polarity supermodulation.  Only short little spikes would be allowed to go beyond -100% modulation, and the receive selectivity would probably cut that back to 100% or less on the receive side, so it would sound fine.  Some bass-phase shifting would enhance the natural speech waveform asymmetry.

The old idea of a pulsating signal that makes meters swing up when you talk is all wrong.  It is better to have the meters swing down when you talk, as long as it does not result in distortion.  Think about it - if you had a 1.5KW carrier between syllables and words, it would push the AGC and make the receiver quieter between syllables and words at the receive end.  And that's when you notice noise the most, between syllables and words.  So what if a light bulb would get dimmer when you talk!

  Bacon, WA3WDR

[WA3VJB]

Thanks, Bacon, I had forgotten which side of the corn controversy he was on.

Quite the health kick it seems. I googled him and found this:
http://www.marketlaunchers.com/bonadio.html

He has not been ill enough to need to stay indoors in the last 25 years, ever since he started experimenting with Indium. He works a full week, every week, at least 50 weeks per year.

Born 1917.

No mention of corn, however.


The Foundation For Health
George Bonadio, Executive Director
337-373 East Avenue
Watertown, NY 13601-3829

Or call us at: (800) 724-7460 or (315) 782-6664 and ask for George Bonadio.

[Steve - WB3HUZ]

Probably the best way to go would be an unbalanced-balanced modulator and voice polarity supermodulation.

That's what I'm running. I set the audio polarity such that the postive peaks exceed the negative peaks the majority of the time. Then I run the negatives up to 100% and let the postive go where they may, without exceed the peak power capability of the transmitter. With my voice, this works out to be around 120-135% positive peak modulation.

Mike, N9MC, visited George about a year ago. At that time he was still alive and well. Never heard if he was still active with amateur radio.

[WA1GFZ]

I remember George bragging about his 80 meter X beam performance that he could never demonstrate it on TX.

[Todd, KA1KAQ]

There was a huge thread a year or two back when I brought up George's name and call to the group. I wondered if he was still around too, apparently he was still alive and kickin' back then with no issues.

Yes, corn was bad, not good. I remember him telling me that it keeps you awake at night, but can't remember the reasoning or description he gave me. This was eons ago, when I was still A.M. wet-behind-the-ears. He had an old 32V transmitter in his basement he gave to me as a parts unit, all I had to do was go get it. Should've done it, meeting him would've been a hoot.

He did have somewhat peaky audio as I recall, but crystal clear and strong to VT. I remember his signal as one of the AM beacons of 75 meters in the northeast back then. Interesting guy, for sure. I always enjoyed talking with him.

"INDIUM is the "Missing Link" unifying two major Mineral groups, for the first time in history INDIUM is made bioavailable by George Bonadio's patented process #6,007,847." 

a.k.a. 'IndiumEase' which is what George was always talking about and credited his excellent health and lack of aging to.

But here's the really good stuff. Just goes to show that you never know who you're talking with on the other end!

http://www.military.com/Content/MoreContent1/?file=dday_0054p1

Cool!

[flintstone mop]

I'll have to agree somewhat about the BC FM stations and the audio processing creating that extra low end they seem to have. The high end audio is beautiful. It makes cheap radios sound really good and the so-called upgraded systems in some autos to really sound nice. It's like the Fletcher-Munson curve on steroids.
The RDS information is way out the passband of audio and there is no way to be able to hear any artifacts from it. It's like an SCA.
Musicast was an SCA of WWDC FM in Washington DC. When DC101 was doing 'sweeps' we could hear little snibbits of the "Grease Man" in the Musicast. The engineers were pushing the limits of the main and Stereo subcarrier right into the SCA.
Today's FM BC sounds so clean, why are we messing with another format of "HD RADIO"
Just .02 worth
Fred

[Bacon, WA3WDR]

I did hear from one fellow that George was indeed heard well by some long DX stations.  His signal strength stateside was not very impressive, though.  Not bad, but nothing special at all.

His strong high-frequency speech pre-emphasis made his audio waveform spiky, and his overall loudness was not great. His clarity, however, was quite good.

I think that some good audio compression, with a little bit of peak clipping, would have made his audio a lot more powerful.  I agree that strong high-frequency boost is appropriate for speech processing on amateur radio, but my ear tells me that George was overdoing it a little bit.  Then again, I could have applied some de-emphasis equalization to the receive audio, and enjoyed better throughput quality on his signal.

   Bacon, WA3WDR

[W1DAN]

Hi all:

I was not here when George was on the air. Sorry I missed him. I do remember W4CJL...he is a main person why I got into AM as well as K4KYV (which I'll never tell him;0). I enjoyed listening to them on my Hallicrafters S120 in grade school.

Thanks for posting the article. It is kinda confusing. His figure 8 seems to be pretty flat to 2.8khz. No pre-emphasis. Figure 2 is where the pre-emphasis comes in. Is there a part 2 with amplifier EQ circuits?

To me the usual pre-emphasis that a D104 or broadcast chain would do seems to be similar to his idea, except for the "scanning" of the listener, which I still do not get.

My idea on AM audio (if anyone cares!) is the transmitter should do pre-emphasis to correct for the average receiver rolloff. Then do overmodulation protection. That is all....does not need to be more complicated.

If one does the traditional "supermodulation" that clips the negative peaks, does this not create 3rd order harmonic distortion?

Good thread!

73,
dan
W1DAN

[Bacon, WA3WDR]

Asymmetric clipping produces mostly even-harmonic distortion, but even-order distortion creates harmonics much like odd-order distortion does.  Soft-clipping of negative peaks is better than overmodulation, but it still produces harmonic distortion, and some splatter will result if post-clipper filtering is not applied.  The transient characteristics of the splatter filter are also important; if ringing/overshoot peaks stick out and overmodulate, you will still get splatter.  Also, the frequency-dependant load presented by a splatter filter makes things worse on the mod transformer - KYV Don has told the sad tale of his many blown mod transformers that lost their lives to this phenomenon.

I think WLR George's voice 'scanning' idea comes from the vibrating vocal cords producing pulsating overtones.  He probably saw overtone pulses from some frequencies coming through before others, and called that scanning, because it repeats like a scanner with each cycle of the fundamental voice frequency.

[Steve - WB3HUZ]

The scanning frequency George writes about is actually the pitch of the voice or the rate at which the vocal cords of the speaker vibrate. The vocal cords also create harmonics of the fundamental pitch called formants. Building a bandpass filter around the first two formants will allow recovery of speech, although it won't sound natural. So, 300-3000 Hz isn't really needed for speech - more like a 500-800 Hz of total BW, maybe less. This allows for intelligible speech, even in very poor SNR situations. The trick is to properly estimate or measure the pitch. Once this is done, DSP can build those bandpass filter at the correct frequencies.

[W1DAN]

Guys:

Thanks for the responces...I now see what he was picturing.

I agree that clipping of any kind produces distortion that needs to be correctly filtered. I have studied clipping and filtering in great detail, primarily with what the Optimod does. There are some good low level clippers and filters out there.

73
Dan
W1DAN

[k4kyv]

If one does the traditional "supermodulation" that clips the negative peaks, does this not create 3rd order harmonic distortion?

 

Once upon a time I could write out the mathematical formula as a series function, just as once upon a time I could derive Maxwell's Equations from basic electrostatic equations, and explain how they relate to actual physical phenomena.  I haven't manipulated those equations in so long now that I'd have to get an elementary electricity /magnetism text, brush up on advanced algebra and calculus and work my way through all the theory and equations.  Kind of like forgetting a language that you haven't spoken for decades.

But I do recall off the top of my head that since symmetrical clipping approaches a square wave as the clipping level is increased, symmetrically clipped waveforms result in predominantly odd harmonic distortion.

Ultramodulation does not clip the negative peaks, but reduces the amplitude of the negative half of the audio sinewave, resulting in predominantly even harmonic distortion.  This is basically what happens with a single ended audio amplifier, but in the case of ultramodulation it is carried to extreme.  The "grid leak" triode detector works on the same principle.

In any case, we can establish that ultramodulation, as well as negative peak clipping, generate a severe case of harmonic distortion.  In addition, some of the audio is rectified due to the nonlinearity of the negative peak attenuation circuit, and the resultant DC is applied to the final, added to the DC  from the power supply.  The result is that the effective DC plate voltage is increased with modulation, the sum of power supply DC + rectified audio.  So this produces exactly the same product as a controlled-carrier transmitter.

These "super-modulation" schemes result in nothing more than an AM signal with a little controlled carrier plus a lot of even harmonic distortion.

The plate current meter will kick upwards due to the rectified audio, and the rf ammeter will jump due to the increased carrier power plus all the added audio harmonic energy in the sidebands, but the actual increase in real sideband power (product of the modulating signal and carrier) is minimal.  The s-meter might even kick upwards slightly, and the signal might sound "louder" due to all the hash being generated along with the audio, but the intelligibility of the signal is increased little, if any.

A far better approach is to take advantage of the natural asymmetry of the human voice, by using a low-distortion audio chain and modulator, with at least one octave of flat frequency response above and below the limits of the intended frequency response,  plus a modulator with extra, undistorted, peak power output capability.

The power advantage is not in the positive peaks that exceed 100%.  The power advantage lies in the increased depth of modulation, down to near 100% in the negative direction, that is possible without the severe limitations on peak audio power caused by a modulator lacking enough headroom to accommodate the positive peaks that naturally result when the carrier is modulated to near 100% in the  negative direction.  Otherwise, the positive peak flat-topping causes exactly the same type of distortion and splatter as overmodulation in the negative direction.

Whether this scheme or some form of "ultra" modulation is used, one of the requirements is a modulator with enough undistorted peak audio power output to modulate in excess of 100%.  And this power requirement increases as the square of the modulation percentage.  This means that a modulator capable of modulating the carrier 200% positive, must have undistorted peak power output capability four times that of a modulator just barely capable of modulating 100%.  It takes double the peak power output capability just to modulate up to 140% positive.

And it ain't gonna happen when one adds some rectifiers and resistors between the mod xfmr and class-C final of a DX-100, and turns the mic gain wide open without first souping up the modulator, or adding a big dropping resistor to cut back on the DC plate voltage to the final for "reduced carrier" output. 

The modulators in many commercial, military and amateur transmitters are intentionally designed so that the transmitter cannot be modulated very much in excess of 100%.  Drive the audio any harder, and you get flat-topping and distortion.