The AM Forum

It's not a new idea, but craftily-chosen two-tone combinations can produce asymmetrical audio waveforms that can really exercise the positive peak capability of an AM transmitter without going over 100% on negatives. A simple, no or low cost implementation:
http://www.w1tag.com/2_Tone.htm

John, W1TAG

Very nice. I've used the two tone approach to simulate asymmetrical voice in circuit and signal simulators like Matlab and LTSpice, but never on a transmitter.

Thanks for sharing.

John

This is a "neat" idea.

Note: there is a subtle issue/problem with classic vacuum tube push pull modulators:

Sometimes, but not always, the grids of the modulator tubes are AC coupled to the audio phase splitter/driver.

If the positive-going modulator tube is driven hard enough to draw grid current on positive modulation peaks, then it's grid bias will drift downward (more negative) with this audio input waveform. This is because the repeating positive peaks will charge up the grid coupling capacitor (i.e. the self-biasing effect).

A similar downward drift of the grid bias will occur, with this audio input waveform, in transmitters that employ a relatively high source resistance bias supply, with a high value bypass capacitor across its output.

With real speech, positive peaks occur a much smaller percentage of the time... and the grid leak/biasing resistor (or the bias supply) will be able to remove the charge produced by the grid current that flows on positive peaks... with negligible effect on the grid bias.

I made a test waveform using one of the freeware programs that consists of: a positive pulse of amplitude 60mV and 2ms duration, followed by a negative pulse of amplitude 30mV and 4ms duration, followed by a 200ms pause of amplitude 0.

This waveform has zero average value (i.e. every positive pulse has a area: 60mV x 2ms = 120uV-sec, and every negative pulse has area: -30mV x 4ms = -120uV-sec)... and repeats five times per second.

I believe it is better for simulating voice peaks because of the relatively long pause after each positive/negative pair of pulses.

I recorded it as an MPEG file, and I can play it into my audio chain from my computer's sound card (as you do with the 2-tone audio test waveform)

Since it is periodic, the test tone and the resulting modulated transmitter output envelope are readily observed with an oscilloscope.

The MPEG file is attached below.

Stu

Is there a formula to calculate the increase in dB of audio by modulating above +100%? I am aware of the additional distortion in some receivers, the effects on signal fading, and the higher voltages on transmitter components, so I am trying to weigh those against the increase in perceived audio.

This should get you started.

http://amwindow.org/tech/htm/modulation.htm


IMO, it's more important to have a big carrier/signal, since this largely defines the SNR on the receiver end. Then is important to have a high average level of modulation. Anything over 100 percent is just icing on the cake and won't usually make much difference, except when your signal is weak (that extra 3 dB might be important).

That's a great idea too. Each polarity then contains the same energy, very nice!

Heard some of those big carriers with low modulation. Bad business. RX AGC is working normally for RF and IF, but there is usually none for audio because it is assumed that all modulation is targeted for 100%.

You turn the receiver up to hear the big carrier/small modulation station, and the next person who has good modulation blasts you out.

Picking up on the idea of re-use of circuits as done in transceivers, I wonder if a good and clean compressor in the audio chain could be switched in to level the audio output of a receiver, to accommodate all levels of carrier and modulation.

Great link and info.. Thanks for posting

Thanks for a novel idea! Going at the link the function equation doesn't look right.
Shouldn't it be Y=SIN(X+PI/4) + SIN(2X)
Joe
 

Actually, what is shown in the graph is: f(x) = cos(x) + cos(2x) = sin(x+ pi/2) + sin(2x + pi/2), where x is the horizontal axis, in radians. As plotted, the waveform repeats when x = 2pi = 6.28...

You can define a new function of x, which is f(x) shifted to the right by pi/4

g(x)  = f(x - pi/4) =

sin[(x - pi/4)+ pi/2)] +  sin [2(x - pi/4) + pi/2] = sin(x + pi/4) + sin(2x)

Therefore the graph of g(x) = sin(x + pi/4) + sin(2x) would have the same shape as the graph that is shown, but it would be shifted to the right by pi/4

In any event, the graph that is shown is f(x) = cos(x) + cos(2x); and it does not have the same shape as sin(x) + sin(2x), even if you allow for shifting f(x) to the right by some amount.

Stu

Prolly a better test for linear operation in the AM mode.

Fred

Thanks for the link: http://amwindow.org/tech/htm/modulation.htm

Does this account for the fact that the negative peaks have to stop increasing at -100%? For modulation levels from 0 to 100, both halves of the waveform increase together, but above 100%, only the positive half of the waveform can increase which results in asymmetry.

Wouldn't the formula have to change above +100%?

Stu,
Thanks for rationalizing the graph equation.
Joe

The peak-to-peak voltage of the envelope is determined entirely by the positive peaks. Matters not that the negs are limited to 100%.

What I was getting at has more to do with the perceived increase in volume to the listener as a modulated signal goes beyond +/- 100%. Increasing a symmetric modulated wave from 50 to 100% increases both the positive and negative halves of the waveform whereas an asymmetric increase beyond +100 only results in an increase in half the waveform.

If the negative half of the modulated waveform contributes to "loudness" then that contribution would diminish beyond a fully modulating symmetric waveform.

I've seen pinched carriers and the resultant distortion being confused for loudness on more than one occasion....

--Shane
KD6VXI

You're stuck on positive and negative. It not about either of those. It's about peak-to-peak (p/p). If you increase the p/p of the envelope, the audio will be louder.

Yes, but, the benefits erode over +100%. From Frank Foti (Omnia) regarding this discussion:

"There's actually a couple of aspects at play here. Increasing the size or volume of the positive cycle will increase loudness a bit. By example, in AM, asymetrical modulation is allowed to be +125%/-99%. A [theoretical] increase in volume of symmetrical modulation to +/-125% would offer about a 2dB in loudness factor. Since we're gaining only half of that waveform increase, the loudness shift would be about 1/2 that, or 1dB.

At +125%, the added odd order harmonics is minimal. At +180%, it would get quite audible. Also, the method of creating asymmetry will be an issue. We employ psychoacoustic masking when generating asymmetry so the added harmonics are minimized."

The result is that, given the 1500 watt PEP limit, the cleanest and loudest signal will be achieved with careful clipping of a 375 watt symmetric waveform rather than by using a lower carrier power and driving a negative peak diode circuit to achieve higher positive peaks up to 1500 total.

Apologies for the confusion over the plotting. The plot shown was indeed generated by the cosine function. When I did the write-up, my brain was on auto-pilot! I have corrected the article.

John, W1TAG

I installed that TCube twice and  I cannot find it on my computer. I looked at my add/remove area and on the C drive and its not to be found. Windows 8.1? Any ideas?

Terry,
Same thing happened to me but did eventually find the three files in the OS Win 7. When I ran the exe file the virtual generator did appeared but didn't run properly. The program wouldn't close had to do a system restart.
Joe  

It's not mentioned on his web site what OS's it even works in. Maybe it doesn't work in Win 8.1.

But to find the files in Win 8.1; scroll down to the Task Bar (the bottom bar with the thumbnails on the left side); mouse over them to find "This PC"; click it on; upper right, enter in the field, "Search This PC", the name of file (I would start with tcube).

 I am happy to report that it works fine in Windows 8.1 ;D ;D

Thanks Pete & Joe, I installed it again in a jump drive and it worked. The Icon is extremely small but it gave me what I needed to look for. I went back to my download file and found that the program installs itself into the download file and not in the C drive programs area, so that is why I could not locate it. I sent an icon to the desktop and now its very easy to find.

Terry

Can you record about 15 seconds of the combined cos(x) and cos(2x) waveforms as an MPEG file... and post it in this thread.

That way, folks who want to inject the waveform into their AM transmitter could just down load it, and play it on their PC or their iPod or their smart phone.

Stu

Didn't work here on windows 7. Oh well that is why there are real TTGs in the operating rack.

Here is the two tone signal as per specs but I think the user will have to adjust the level at the user site. I hope its ok as I am not used to doing anything like this ;D ;D

Attached is a 100 second MPEG3 file (mono)

It consists of sin(2pi 500.01t) + sin(2pi 1000t)

My audio generator application allows me to generate two sine wave tones, but it doesn't allow me to adjust their relative phase.

So... I set one of the frequencies at 500.01 Hz instead of 500Hz. Over the 100 second duration of this recording, the two tones slowly shift their relative phase over 2pi radians.

If you look at the whole 100 second waveform on a scope, with the sweep set to 0.2ms or 0.5ms per division, and watching for 100 seconds, you will see the effect of the phase shift on the sum of the two sine waves.

For a good portion of this 100 second waveform (at least 5 seconds), the phase relationship is such that what you have is basically sin(2pi 500t + pi/4) + sin(2pi 1000t) ... which has the same shape as cos(2pi 500t) + cos(2pi 1000t)... which is the desired test waveform shape.


Stu

Maybe there is something wrong with my tones? i can hear both tones but not the sum of the two?

Terry

Your file has one tone on the left channel and the other tone on the right channel.

When I combine the two outputs of my sound card with a Y-adaptor... the result is exactly what is needed. The phase relationship of the two sine waves is correct for the purpose of producing the desired test tone.

Stu

Stu, with this setup, I cannot combine them. Can you post the combined tones from my mp3?

tnx Terry

Here's a mono mixdown version.

Thanks guys,

Kevin, WB4AIO.

Here's a look at the waveform. The center horizontal graticule line is zero volts.
Joe

     
Some time back I was doing two tone tests on my Gonset G-76. The frequencies were 700hz and 1900hz.

I include a scope image of the test. Getting a good scope trigger was impossible until I changed the variable time base.

Jim
Wd5JKO