Transmit audio phase with AM mode

N3WWL

New Flex 6300 owner, well versed in SDR and Flex since the purchase of a new 1000 many years ago, then upgrading to a 3000 in 2009, and now the 6300.  The transmitted audio in AM mode is out of phase as viewed on my oscilloscope.  The audio is being fed at line level at the 15 pin d-connector on the rear of the 6300...pin one mic hot and using pin 5 for ground (unbalanced).  This is exactly how I fed the audio into the 3000 with no inverted phase problem.  I confirmed the improper phase by defeating the outboard audio using the hand mic supplied with the 6300, directly in to the front 8 pin foster connector, and the audio is still out of phase in the AM mode.  

Steven Hess

I'd raise a support ticket on the issue. 

David Altekruse

Pardon my ignorance, but I have two silly questions.  

1)  When you say out of phase do you mean 180 degrees out of phase?  

2)  What difference does it make?   That is, under what situation would this be a problem?

Please email me your answer so I won't waste any more of other people's time.  

David (dna@comcast.net), N1DNA

W1AEX

Hi Jay,

I don't think this is an unusual problem when running AM with a lot of present day ham gear but would be very surprised if it was a widespread issue with the Flex 6000 line as Flex has a good history of being very much on top of that kind of detail.

Since I use the same external hardware audio chain with multiple transmitters (with some exhibiting inverted phase) I finally gave up and now use a 1:1 (600 ohm) transformer at the output of the audio chain with a DPDT switch on the secondary so I can flip the phase so that the waveform is correct on the scope. The Edcor Transformer company " WSM600/600" 20-20k flat response transformer for just over 10 bucks works beautifully.

Many dynamic elements used in ham-type hand mics don't even have a plus or minus label on the element so it's totally random whether or not they produce in-phase audio so that an inward excursion of the mic engine produces a positive voltage. Not a problem with SSB but as you have noted with AM it makes it impossible to reach 100% positive peaks before reaching 100% negative peaks.

73,

Rob W1AEX

Jim KJ3P

Hi Jay,

Out of phase when compared to what?  Do you mean that the upper AM sideband is out of phase with the lower sideband?  If so, to what degree?  Any scope photos to show-and-tell?

  --jim KJ3P

Al K0VM

It could only make a difference if the amplitude of the positive peaks of audio where different than the negative peaks of the audio  AND if the difference where predictable.

AL, K0VM

W1AEX

Assuming perfect symmetry in the audio source (a sine wave for example) when an AM transmitter is modulated, positive peaks will reach 100% at the same time 100% negative is reached. Since most human voices are asymmetrical, if the transmitter is modulated in-phase you will see positive peaks that exceed 100% positive before 100% negative is reached. Unfortunately, the reverse is also true and you will reach 100% negative well before you get close to 100% positive if the transmitter is modulated out-of-phase. Unlike plate modulated transmitters this does not result in splatter with SDR transmitters but it does result in a less than optimal AM signal that sounds weak and distorted when compared to the in-phase condition.

In practice, this makes no difference at all with SSB but it does make a significant difference when running AM. When an AM transmitter is being modulated out-of-phase (100% negative reached before 100% positive) it's easily corrected if you are using a balanced audio source by flipping the balanced pair leads.

Rob W1AEX

pa0bie

There is also a certain amount of delay in the DSP TX chain.

rfoust

I'd rather see the response posted here. Great questions and a good chance for me to learn something. A lot of threads on this forum have been very educational to me.

N7AIG

I'm still trying to wrap my head around why any of this matters.

Running less than 100% modulation is a way of buying yourself some margin against fading. I just checked the WWV pages and also find that they run most of their modulation at levels between 50% and 75%, presumably to help stay ahead of selective fading on the carrier.

N7AIG

This isn't a stereo system where running one channel out of phase with the other produces incorrect spatialization. So what is the phase reference here? And furthermore, the description about having more than 100% modulation would always produce distortion no matter what you think of the phasing.

I understand fully about the asymmetry of voiced speech waveforms. Any negative excursions in the modulating waveform would produce modulated levels near the 0 line, and should never drive below that, or you will induce distortion. Positive excursions produce the high amplitude levels of the modulated signal.

Phase rotators have been used in the past to help even out the positive and negative excursions in the modulating waveform and keep higher average modulation levels in the transmitted signal.

Not sure what you mean by stating that SDR modulation is different from plate modulation?

N7AIG

oh, perhaps you mean that having greater positive excursions than 0 dBc are okay, which they might be if the amplifier can handle it, but that negative excursions greater than 0 dBc would drive the system into distortion, which they would. Sounds like a phase rotator *is* called for here...

W1AEX

I think most guys who are concerned about fading will use the synch detector on the receive end rather than under-modulating the transmitter.

One of the nice things about the SDR platform is that it excels with the AM mode so Jay's concern with the apparent inverted phase is understandable since the unbalanced source he used with his earlier Flex transceivers produced an in-phase AM signal when the RF sample was viewed with his scope.

I love your backyard picture on QRZ Dave. Tucson is one of my favorite places to visit because it seems like all the views are spectacular!

Rob W1AEX

N7AIG

Hey! Perhaps this is related to the problems discussed earlier in the distorted reception of AM signals? Maybe something in the Flex internals is inverting these modulation+ signals in the receive chain and forcing them into conditions of induced distortion, very much like when a peaky signal is fed inverted into the modulator?

W1AEX

Actually, the positive excursions beyond 100% simply result in louder sounding audio without sounding distorted (assuming you are not flat-topping in the PA). In fact, many AM broadcast transmitters use processing to modulate their signals to well beyond 100% to sound louder than their competition.

 The difference with the SDR platform is that the negative excursions beyond 100% don't pinch off at the baseline as they do with a plate modulated transmitter. Instead, they pass through the baseline and produce an inverted "wavelet" which sounds fine with a synch detector but can be problematic with some older receivers that use a diode detector. You can see this at 30 seconds into a video (at the YouTube link below) that I made of my Flex 5000A when I was feeding it triangle waves while observing it with a scope. A plate modulated transmitter would pinch off and begin to misbehave but the "software balanced modulator" of the SDR platform simply produces the little wavelets when 100% negative is exceeded.

73,

Rob W1AEX

https://www.youtube.com/watch?v=KemFreHZLJ0

N7AIG

I'm not an AM'er, but maybe this kind of modulation+ is a commonly used method among the crowd? If so, it really is a form of overmodulation. And if rigs try to use average voice levels to set the reference DC level, that is a big mistake unless they take care ahead of time to even out the excursions, as with a phase rotator.

We need the DC reference level to be halfway between positive and negative excursions, and the carrier needs to be that same half p-p amplitude in order to guarantee distortion free transmission and reception. Carriers at less than that amplitude will produce distortion under some conditions.

N7AIG

just as an aside... how bad can things get? Here is an example of a waveform that sounds exactly like a square wave (or at least the first 9 partials):

It was generated by feeding a sinewave into a sum of Tchebyshev polynomials, acting as the modulator of the sinewave. of orders 1, 3, 5, 7, and 9, and with amplitudes 1, 1/3, 1/5, 1/7, and 1/9. But if you play this waveform on a synth it sounds exactly like a square wave. The ear can't tell the difference, all it needs are partials and amplitudes.

N7AIG

yep! this is exactly what I was trying to describe when I stated that more than 0 dBc negative excursions would produce distortion. Interesting that Sync-AM doesn't have difficulties with this. I guess the synthesized carrier is always as strong as it needs to be.
And yes, I do remember seeing mention of some international shortwave stations using diminished carrier modulation. So they must be using something to limit the negative waveform excursions on the modulating signal. Do you happen to know any details about how they successfully implement reduced-carrier AM?

Tim - W4TME

I have entered this report into our bug tracker for further investigation.  Please report any additional information related to this specific issue in this community topic.

Doug Hall

I'm not an AMer either, but I listen on 3885 and 7290 from time to time. Many of the serious AMers there (at least the boatanchor crowd) seem to be running well in excess of 100% modulation positive while limiting the negative modulation peaks. I've always understood it was because speech output from a microphone was naturally asymmetrical.
73,
Doug K4DSP

k3Tim

Rob - W1AEX is correct about why phasing matters.  A detailed explanation :

http://www.qsl.net/wa5bxo/asyam/aam3.html


Enjoy!

k3Tim

N7AIG

Hmm... okay, so I looked at my own voice recording that I use for tuneup. Use Audacity or any other tool. My recording shows almost no asymmetry. Why? Where does the asymmetry ascribed to vocal speech come from?

Turns out it seems to come from the bass frequencies. Harmonics exhibit almost no asymmetry. So a quick fix until Flex comes up with something better is to use a high pass filter to saw off your fundamental voice frequency. You don't need that anyway.

There is a psychoacoustic effect where the brain, on hearing a progression of harmonically related partials will reconstruct a fundamental for itself. That's the secret behind "bass boosters", and while not labeled as such, was well known at least back to the time of J.S.Bach for his grand pipe **** music. Just play a C and a descending 4th at G and lo and behold you hear a C an octave lower.

For radio communications you simply don't need those fundamentals. This isn't high-fi sound, it is communications.

N7AIG

here is a picture of that same waveform, shown earlier, sans the fundamental:

Jim KJ3P

As I’ve mentioned in other threads, voices vary widely in their asymmetric characteristics.  For AM broadcasters, the classic Kahn Symmetra-Peak was a passive chain of all-pass filters, placed *before* all other audio processors.  Its output yielded audio that was symmetric around the baseline, so that there were no lop-sided peaks to artificially force the compressor/limiter into unnecessary gain reduction.

At one small AM station, long ago, one announcer always seemed to have lower voice levels than the others.  It turned out that his voice was strongly asymmetrical, forcing the limiter into more gain reduction than usual.  Since the station could not afford a Symmetra-Peak, a phase reversal switch was installed in the main DJ mic circuit for that particular announcer to use.  We went on to find the optimum phase settings for the other announcers.

Modern AM broadcast processors, of course, do all of this in DSP. Further, the processor “modifies” (distorts?) the waveform so the positive peaks consistently reach 125% modulation, while limiting the negative peaks to 100%.  The FCC allows these limits for broadcasters.

Bottom line:  Any good voice processor “scrambles” the voice phase into symmetry before any compression/limiting takes place.  This should be easy with DSP (several all-pass filters in series), and I’d love to see it added to the Flex mic processor for both SSB and AM.

Ken - NM9P

I have played with a broadcast audio software program called "Stereo Tool" which has a feature that allows just such a 125% positive peak modulation for AM. It also has a multiband compressor, bass boost, and an EQ that you can simply draw the response curve you want on the screen. Best of all is that it is FREE. But it also added about a quarter second latency delay on my TX audio.

W1AEX

Actually, passing through the baseline does not produce distortion with the balanced modulator approach used in the SDR platform. Look again at the wavelets in the YouTube video link above. They are perfect replicas of the injected waveform. The signal will sound perfectly fine when received with a synch detector, however, some receivers that employ a diode detector may have difficulty reproducing the signal without distortion.

So... think about why it's important to assure that the proper phase is used with AM: You can reach 100% positive before you reach 100% negative. If your voice has enough asymmetry you can also exceed 100% positive peaks and still have a good margin well before you approach the baseline in the negative direction to attain a loudness factor to your audio that sounds clean and smooth in any receiver. However, if you are running your AM transmitter out-of-phase these inherent advantages will be lost.

You can implement reduced-carrier AM with your Flex simply by using the AM Carrier Level slider to change the ratio of carrier to DSB energy. However, if you go too far your AM signal will begin to sound like poorly detected sideband unless it is being received with a synch detector.

Rob W1AEX

W1AEX

Ken, I've reached the point where I keep trying to simplify things at my station but every time I see (and hear) what some of my friends are doing with digital audio I can feel the pull to the dark side of making it all complicated again! The number of free VST plug-ins is astonishing and in addition to the functions you have mentioned you can grab a phase rotation plug-in as well. Having heard how well a multi-band compressor works might be enough to drag me even faster in that direction. The SDR platform has opened the door to a lot of interesting approaches to our hobby.

Rob W1AEX

Ken - NM9P

Rob, I found Stereo Tool to be too much complexity for me.  Not that I couldn't "get it" but that there was just so much more than I needed for my usual SSB Rag chew stuff.  And the Processor in SSDR was very ample for my DXing and Contesting.  So I stopped using it.  But now that they have the LAN Codec I would like to find something simple but effective that would allow some of these things - i.e. multiband compression, EQ, and phase rotation, but that didn't add a lot of latency with my i3 computer!  I don't want much, do I?  

I want to sound loud, clean, balanced, articulate, and natural all at the same time, without causing latency delay on the TX.  So far, I have not found a lot that is simple that makes me sound much better than I can achieve with careful adjustment of the built in EQ & processing in SSDR.  But that is MY voice.

Other's may vary.

BTW, I have appreciated (and recommended often to others) your many posts on PowerSDR setup, SSB & AM Audio quality, and others.  It started me down the road to pursuing the max I can achieve simply on both my 1500 and 6500.  Thanks, and keep up the good work.

Ken - NM9P

W1AEX

Dave, I think all those guys who are running a PR-40, RE-27, RE-20, your pick of the Neumann line or one of the other high quality microphones hanging from thousands of ham radio microphone booms might disagree with your last statement. Many operators bought their Flex because it has the ability to reproduce the human voice as perfectly as you could want.

Rob W1AEX

N7AIG

Well, I know this subject is emotionally laden for some. But honestly, the fundamental tones of a human voice conveys zero information. I have been researching human hearing for more than 15 years. If you want to communicate, it doesn't help to shout -- you need to bend your high frequencies louder and the bass frequencies softer. The information of speech is in those higher frequency, low power, broadband consonant sounds. But I also hear a lot of guys on the air that have taken this a bit too far.

Ken - NM9P

I really appreciate your analysis, and plain language explanations.

For the most I agree with you.  But it depends upon what the goal is.  

If the goal is simply to exchange information - i.e. in a contest, DX operation, Awards collection, or weak signal contact, then the high frequency boost, low frequency cut, moderate compression routine is preferred.

If, however, the object is a long, pleasant rag chew, with low listener fatigue, and where the voices sound as natural, warm and rich as possible, then a wider, richer, more balanced audio response is preferred.  If I tried to rag chew with my DX profile for long, my buddies would throw me off the frequency after a while!  (It is "communications effective" but irritating for a long period of listening.)

While the fundamental tones convey little or no actual information, they do provide a certain amount of presence and balance to the "sound" of the voice, which impacts listener fatigue.  

(BTW, the overly bass-boosted signals of some hams I have heard on the bands are as equally tiring as the overly treble-boosted DXer profiles I  have heard.  The constant droning of 20dB-plus boosted bass tones is like a sledge hammer in my headphones!  WHen I hear one I just roll it off with my brick-wall filters!)  

Granted, this is a matter of personal preference.  But that is the blessing of SDR, isn't it?  We have the option of multiple TX profiles customized for a variety of purposes.  

It would be a great option of FRS was able to implement some of these options into their phone profiles for SSB & AM.  A separate pull-down menu with a few of these options would be an amazing addition.

Ken - NM9P