-
Get involved.
We want your input!
Apply for Membership and join the conversations about everything related to broadcasting.
After we receive your registration, a moderator will review it. After your registration is approved, you will be permitted to post.
If you use a disposable or false email address, your registration will be rejected.
After your membership is approved, please take a minute to tell us a little bit about yourself.
https://www.radiodiscussions.com/forums/introduce-yourself.1088/
Thanks in advance and have fun!
RadioDiscussions Administrators
Multipath Simulation for SSB FM Stereo
- Thread starter k6sti
- Start date
I agree, much more testing will need to be done on SSB. I have no problem with simulations, but they are just simulations. There is something happening in the “Boston” video that has improved reception. I would not classify “Boston” as tough multi-path terrain, but the SSB is improving something in the recorded signal. My stab in the dark would be that SSB might reduce perceived distortion caused by AM synchronous noise in the transmission system. Terrain and manmade objects create multipath, transmitters to not. Poor transmission systems can create noise that will cause the receiver to blend, distort or make existing multipath seem worse. It also appears that each manufacturer is implementing their own scheme of SSB. The artifacts created by the SSB overshoot removal methods may differ with each manufacturer’s software implementation. Those artifacts could certainly change the perceived evaluations depending upon how they are filtered or masked.
Brian,
You present interesting information. As testing continues, we will factor your simulations in with our findings. It needs to be pointed out SSB in now on-the-air in numerous markets with various terrain and coverage circumstances.
To date, the SSB system:
1. Does not generate any modulation overshoots, without degrading audio performance. This claim is true of Omnia.11, Omnia.9, and the Optimod 8600.
2. There have been no reports of SSB degrading conventional coverage or reception. This refutes Bob Tarsio's paper.
3. Has shown a reduction of multipath for quite a few broadcasters, and improved signal coverage.
4. Has not caused any negative feedback regarding stereo separation.
I started this process by doing a lot of simulating in MatLab. From there, we implemented it, in one of our prototype processors, and made a few in-field tests. Now, broadcasters have applied for Experimental Authorizations, and put it on-the-air, with positive results. I'm still the first to say, the concept needs additional testing. Some of that has started, but there is still more to do.
For the sake of discussion, I wish to share a story about why simulations, equations, and white papers, are only half the answer...
Back in the 1940's, engineers at Bell Laboratories had tried to design a way to null out sidetone noise in long distance telephone lines. They never could make it work. They were so convinced this was impossible, that they wrote a very long paper on why this would *never* work. They included page after page of equations, theory, test results, and more. In their view, this was not possible...ever!!!
Well, in 1985, a broadcast engineer had an idea. He read this Bell Labs paper, did some research, and made use of this new (at the time) tech called DSP. He created an adaptive hybrid, that produced a trans-hybrid null never achieved before...EVER!! That broadcast engineer is my buddy Steve Church. They said it could never be done...ever. Well, he thought outside the box, and proved them wrong. He didn't rely on the closed loop thoughts of a white paper, and findings of those scientific engineers whom preceded him.
The same applies here. White papers, simulations, and prior efforts are valid, but only up to a point. If we don't challenge them, we discount the possibility of advancing the medium. This is why the other half of the process is to bring the idea to life, and get it out into the marketplace for real world testing. The results shared here, are beyond those of white papers, simulations, and bench tests.
We need to see this process through. Your input is valid, and welcomed. Broadcasters are realizing improved coverage with SSB, and we'll continue the ongoing research and testing. I hope you'll plan to be a part of this exciting experience!!
-Frank Foti
You present interesting information. As testing continues, we will factor your simulations in with our findings. It needs to be pointed out SSB in now on-the-air in numerous markets with various terrain and coverage circumstances.
To date, the SSB system:
1. Does not generate any modulation overshoots, without degrading audio performance. This claim is true of Omnia.11, Omnia.9, and the Optimod 8600.
2. There have been no reports of SSB degrading conventional coverage or reception. This refutes Bob Tarsio's paper.
3. Has shown a reduction of multipath for quite a few broadcasters, and improved signal coverage.
4. Has not caused any negative feedback regarding stereo separation.
I started this process by doing a lot of simulating in MatLab. From there, we implemented it, in one of our prototype processors, and made a few in-field tests. Now, broadcasters have applied for Experimental Authorizations, and put it on-the-air, with positive results. I'm still the first to say, the concept needs additional testing. Some of that has started, but there is still more to do.
For the sake of discussion, I wish to share a story about why simulations, equations, and white papers, are only half the answer...
Back in the 1940's, engineers at Bell Laboratories had tried to design a way to null out sidetone noise in long distance telephone lines. They never could make it work. They were so convinced this was impossible, that they wrote a very long paper on why this would *never* work. They included page after page of equations, theory, test results, and more. In their view, this was not possible...ever!!!
Well, in 1985, a broadcast engineer had an idea. He read this Bell Labs paper, did some research, and made use of this new (at the time) tech called DSP. He created an adaptive hybrid, that produced a trans-hybrid null never achieved before...EVER!! That broadcast engineer is my buddy Steve Church. They said it could never be done...ever. Well, he thought outside the box, and proved them wrong. He didn't rely on the closed loop thoughts of a white paper, and findings of those scientific engineers whom preceded him.
The same applies here. White papers, simulations, and prior efforts are valid, but only up to a point. If we don't challenge them, we discount the possibility of advancing the medium. This is why the other half of the process is to bring the idea to life, and get it out into the marketplace for real world testing. The results shared here, are beyond those of white papers, simulations, and bench tests.
We need to see this process through. Your input is valid, and welcomed. Broadcasters are realizing improved coverage with SSB, and we'll continue the ongoing research and testing. I hope you'll plan to be a part of this exciting experience!!
-Frank Foti
I have updated my multipath simulation with several new graphs. I have also added a table that gives the SSB S/N advantage for multipath delays between 1 µs and 100 µs in 3-µs steps and multipath levels from -3 dB to -33 dB in 3-dB steps. See if you can guess the average of the 374 numbers before looking at the table.
Brian
Brian
I've updated the SSB FM stereo simulation to eliminate a couple of distortion products near -90 dB. They were due to a tiny amount of time jitter in the oscillators. Changing the sine arguments to double precision got rid of it. I also moved the tone frequencies to the center of FFT bins to minimize spectral leakage. Neither of these changes had significant consequences. But since I'm plotting the spectrum to -100 dB, some plots became a bit cleaner near the floor.
http://ham-radio.com/k6sti/ssb.htm
I forgot to mention that I put my Sony XDR-F1HD on the bench last night. With my ST-1000A generating a normal L-only, R = 0 DSB stereo signal using a 1-kHz tone, I summed a second oscillator at 39 kHz and cancelled the USB tone. As expected, the tuner blended fully to mono on the resulting SSB FM stereo signal.
Brian
http://ham-radio.com/k6sti/ssb.htm
I forgot to mention that I put my Sony XDR-F1HD on the bench last night. With my ST-1000A generating a normal L-only, R = 0 DSB stereo signal using a 1-kHz tone, I summed a second oscillator at 39 kHz and cancelled the USB tone. As expected, the tuner blended fully to mono on the resulting SSB FM stereo signal.
Brian
FFoti1 said:
No. The normalized SSB subband levels were L+R = 1, L-R = .5. After matrixing this yields 2L = 1.5 and 2R = .5, for a separation of 20*LOG(3) = 9.5 dB. (This happens to be the value I use when converting hi-blend to flat blend to avoid spectral smearing of the stereo image.) I was listening to the output with headphones. The DSB separation of > 50 dB collapsed to mono, not to 9.5 dB.
Brian
I neglected to provide a reference for what I call flat blend. Here it is:
http://ham-radio.com/k6sti/blend.htm
Brian
http://ham-radio.com/k6sti/blend.htm
Brian
k6sti said:
Just played with my Sony XDR-F1HD tuner, and SSB. Placed the Omnia.11 in SSB, audio on L only. There is plenty of stereo separation, but the right channel does have artifacts similar to what you would get with composite clipping as well as some stranger sounding things that sound like the tuner's active noise masking system.
Did not run any lab tests, just listened. When I am in the office, I will bring it so Frank can post official numbers as he has the gear to do that already. I am at home this week, and my gear is also in the office.
In any case, mine was NOT mono with SSB. This post got my curiosity up, and I do not recall the Sony tuner being in mono when I tried it W/ SSB several months ago. Just did a quick listen again just to make sure.
-Cornelius
What sort of modulation were you using? I used a constant tone. It's possible the noise reduction algorithm behaves differently with varying modulation. It does have several time constants associated with it. I switched my cancelling signal on and off several times. The tuner switched rapidly between the tone in the left channel for DSB and in the center for SSB.
Incidentally, the XDR-F1HD mutes, or greatly attenuates, a steady L = -R test signal. I don't know whether this is intentional or just a byproduct of its noise reduction algorithm. It doesn't seem to affect broadcast signals, although I doubt they often exhibit pure channel antiphase. But the muting can be confusing if you switch a signal generator to L = -R while testing.
Brian
Incidentally, the XDR-F1HD mutes, or greatly attenuates, a steady L = -R test signal. I don't know whether this is intentional or just a byproduct of its noise reduction algorithm. It doesn't seem to affect broadcast signals, although I doubt they often exhibit pure channel antiphase. But the muting can be confusing if you switch a signal generator to L = -R while testing.
Brian
Hi there!
I was using typical programming content (music & talk), fed through an FM exciter, and picked up off-air via the tuner.
Interesting observations with the tone test, though. The Sony HD tuner is really an odd duck. Very interesting to observe how it reacts to things. I find it hard to listen to (critically) in general as the noise reducing schemes on the analog side have some annoying MP3 like qualities (annoyances) to *normal analog* stereo content.
First blush, though, it sounded great. After a while I started noticing all these anomalies with analog reception. In HD Radio mode, though, it sounds great.
-Cornelius
I was using typical programming content (music & talk), fed through an FM exciter, and picked up off-air via the tuner.
Interesting observations with the tone test, though. The Sony HD tuner is really an odd duck. Very interesting to observe how it reacts to things. I find it hard to listen to (critically) in general as the noise reducing schemes on the analog side have some annoying MP3 like qualities (annoyances) to *normal analog* stereo content.
First blush, though, it sounded great. After a while I started noticing all these anomalies with analog reception. In HD Radio mode, though, it sounds great.
-Cornelius
k6sti said:
Try a L-only, R = 0 tone and see what happens.
I put my XDR-F1HD back on the bench and verified the reversion to mono for 1 kHz = 1, 37 kHz = 0.5 relative levels. Those are the levels for an acoustic image partway between center and extreme left. As I change the phase of my 39-kHz cancelling tone, the tuner stays in mono over a fairly large range, suddenly popping back into stereo when the level of the 39-kHz tone gets high enough. There is a dead band where it's strictly mono.
I tried another approach where I put the ST-1000A into L+R mode with pilot. Then I injected 37 kHz via the SCA jack. This let me try equal levels for 1 kHz and 37 kHz. It behaved differently, with no sudden reversion to mono. I could get the output to go from full L through mono to full R by changing the 37-kHz phase. But what was strange was that best separation did not occur for maximum or minimum composite level, where I believe the 37-kHz tone should be in or out of phase with 1 kHz. It seemed to occur for some intermediate phase, perhaps 90 degrees. Because the generator's 38 kHz isn't exact, I can't judge relative phase with a scope (though I tried). I'm not sure what to make of this experiment.
I measured the muting level for L = -R as 16.5 dB.
Incidentally, I don't hear artifacts from my XDR-F1HD except under exceptional circumstances, as when it's suppressing noise from a signal so badly degraded that it's otherwise unlistenable. See my review for details:
http://ham-radio.com/k6sti/xdr-f1hd.htm
The review contains a block diagram of the noise reduction system. You can trace the quadrature signal and see how it is involved. For clean DSB stereo, the quadrature signal is zero. For SSB stereo, it is huge.
Under ordinary conditions with a clean strong signal, I can't distinguish the sound from that of any other tuner in an A/B comparison. The audio is clean enough that I think I'm hearing artifacts from Arbitron encoding on one local station.
Brian
I put my XDR-F1HD back on the bench and verified the reversion to mono for 1 kHz = 1, 37 kHz = 0.5 relative levels. Those are the levels for an acoustic image partway between center and extreme left. As I change the phase of my 39-kHz cancelling tone, the tuner stays in mono over a fairly large range, suddenly popping back into stereo when the level of the 39-kHz tone gets high enough. There is a dead band where it's strictly mono.
I tried another approach where I put the ST-1000A into L+R mode with pilot. Then I injected 37 kHz via the SCA jack. This let me try equal levels for 1 kHz and 37 kHz. It behaved differently, with no sudden reversion to mono. I could get the output to go from full L through mono to full R by changing the 37-kHz phase. But what was strange was that best separation did not occur for maximum or minimum composite level, where I believe the 37-kHz tone should be in or out of phase with 1 kHz. It seemed to occur for some intermediate phase, perhaps 90 degrees. Because the generator's 38 kHz isn't exact, I can't judge relative phase with a scope (though I tried). I'm not sure what to make of this experiment.
I measured the muting level for L = -R as 16.5 dB.
Incidentally, I don't hear artifacts from my XDR-F1HD except under exceptional circumstances, as when it's suppressing noise from a signal so badly degraded that it's otherwise unlistenable. See my review for details:
http://ham-radio.com/k6sti/xdr-f1hd.htm
The review contains a block diagram of the noise reduction system. You can trace the quadrature signal and see how it is involved. For clean DSB stereo, the quadrature signal is zero. For SSB stereo, it is huge.
Under ordinary conditions with a clean strong signal, I can't distinguish the sound from that of any other tuner in an A/B comparison. The audio is clean enough that I think I'm hearing artifacts from Arbitron encoding on one local station.
Brian
Played some more.
That Sony tuner is an odd duck for sure.
When SSB is used, the separation depends on what frequency you're talking about. I did the 1K tone test, yes, it's mono. Change the frequency, full separation. Do a tone sweep, and it has a comb filter effect. At regular intervals, you get full separation, mono, full separation, so on.
It definitely plays some games with the demodulation process. Using it side by side with a conventional tuner, the conventional tuner had full separation the whole time, SSB or DSB. The conventional tuner had a little less separation at high frequencies in SSB, but nothing that would be a show stopper, and nothing anyone would notice unless you deliberately set out to run tone and measurement tests.
Again, the Sony tuner is an odd duck here. This must begin explain the oddities I hear in the radio under typical listening conditions that annoy me. Yes, if you give it a perfect full quieting signal, it's great. Otherwise it sounds odd, MP3-ish. These tests are revealing a lot of what I was hearing...again, using it like a typical listener would. If I hook it to the roof antenna, I'm sure it would sound phenomenal.
-Cornelius
That Sony tuner is an odd duck for sure.
When SSB is used, the separation depends on what frequency you're talking about. I did the 1K tone test, yes, it's mono. Change the frequency, full separation. Do a tone sweep, and it has a comb filter effect. At regular intervals, you get full separation, mono, full separation, so on.
It definitely plays some games with the demodulation process. Using it side by side with a conventional tuner, the conventional tuner had full separation the whole time, SSB or DSB. The conventional tuner had a little less separation at high frequencies in SSB, but nothing that would be a show stopper, and nothing anyone would notice unless you deliberately set out to run tone and measurement tests.
Again, the Sony tuner is an odd duck here. This must begin explain the oddities I hear in the radio under typical listening conditions that annoy me. Yes, if you give it a perfect full quieting signal, it's great. Otherwise it sounds odd, MP3-ish. These tests are revealing a lot of what I was hearing...again, using it like a typical listener would. If I hook it to the roof antenna, I'm sure it would sound phenomenal.
-Cornelius
cgould said:
The XDR-F1HD assumes that the FM signal has both L-R sidebands per the FCC spec. It takes advantage of them by using the power in the quadrature channel to estimate the S/N in the in-phase channel in 17 different frequency bands. Since quadrature power normally is zero for a DSB signal, the tuner can use its presence to detect signal impairment. It shields the listener by lowering the L-R summing coefficient in the frequency bands where the quadrature power is high and the in-phase L-R channel therefore is noisy or multipathy. When fed an SSB signal, it sees quadrature power equal to in-phase power in all bands. For an FCC-spec DSB signal, this means that the signal has been severely impaired. You shouldn't be surprised that the tuner functions oddly on an SSB signal. It wasn't designed for it. The tuner take advantage of the redundancy present in a DSB signal. That redundancy has been removed from an SSB signal.
If you feed the tuner a very noisy signal, occasionally it will let some L-R noise through on louder program peaks for which it thinks S/N is high enough. On softer sounds it blends the channels to mono to suppress the noise. In this case the program material may gate or modulate the L-R noise. It can sound rather odd, kind of "grassy." But signals for which the tuner behaves this way are completely unlistenable in stereo on an ordinary tuner because they contain gobs of L-R noise. In essence the Sony is trying too hard to preserve some stereo content when it would be better off just blending everything fully to mono. Can you make a recording for me? I'd like to hear what you refer to as MP3 sounds. They may simply be due this noise modulation effect.
Brian
Doesn't have to be noisy.
One annoying artifact with this tuner I hear & observe under any condition is low level audio signals (signals that modulate the transmitter at low levels) are switched to mono. This thing doesn't blend, BTW...it switches to half separation, then to Mono. Maybe a couple more "in between" steps in there.
I'm not turning down the composite levels here. Talking about standard modulated composite with the pilot tone at 9% all the time.
If the audio modulation (stereo, DSB) drops below a certain level, it switches to mono. On tone, this level is pretty low, but on music content, it's more active.
I'd imagine that noisy signals (as you pointed out) would fair much worse. With a standard tuner, it's just a bit noisy, no funny games going on.
The more I play with this tuner closely, the less I like it. I thought it "sounded funny" before. Now these tests are putting a face on things I couldn't put my finger on before. My old standard reference tuner performs much better. :-\
-C
One annoying artifact with this tuner I hear & observe under any condition is low level audio signals (signals that modulate the transmitter at low levels) are switched to mono. This thing doesn't blend, BTW...it switches to half separation, then to Mono. Maybe a couple more "in between" steps in there.
I'm not turning down the composite levels here. Talking about standard modulated composite with the pilot tone at 9% all the time.
If the audio modulation (stereo, DSB) drops below a certain level, it switches to mono. On tone, this level is pretty low, but on music content, it's more active.
I'd imagine that noisy signals (as you pointed out) would fair much worse. With a standard tuner, it's just a bit noisy, no funny games going on.
The more I play with this tuner closely, the less I like it. I thought it "sounded funny" before. Now these tests are putting a face on things I couldn't put my finger on before. My old standard reference tuner performs much better. :-\
-C
k6sti said:
Interesting observations guys! As a Sony HD tuner owner and a DX guy too, I have noticed that it acts certainly different that an average tuner. For the most part, it's a positive, but, now that Corny mentions it, sometimes it does sound a bit different to my ears at lesser signal levels. What you guys have been discussing sort of proves out that subtlety of audio differences. Overall I love the tuner for what it is, but I agree something like a FanFare or one of the real high-end tuners actually sounds slightly better at times. I have a Fanfare, a Sony, and a BW Broadcast tuner. The Fanfare has the best fidelity, then the BW, then the Sony...
I just had a listen to the Sony tuner and SSB.
With program material - not single tones - it does not switch to mono when SSB is transmit. Using content with lots of separation like the Beatles, it's easy to subjectively hear the results. It does seem to play games with audio separation a bit, but it is not mono, as we've been lead to believe. Our tests were done using a broadcast setup, not a bench test, where the output from an audio processor was connected to a known FM exciter, and modulation set to 100% (75kHz deviation). FYI: Pilot injection set to 9%. Switching between DSB and SSB was done in the broadcast grade stereo generator.
Worth noting is an interesting anomaly with it in DSB mode. When the audio signal (music) is extremely low, it does switch to mono even though the content is stereo. Must be something in the noise sensing algorithm. The same test done with another receiver does not possess this issue. I've downloaded the Sony patents, and will read them with interest.
Performing tests using program content, as compared to test tones, revealed what a listener will experience in the real world. That is, unless some broadcaster decides to create an all tones, all the time, format. We use tones for testing and design purposes, but, as has happened many times, the end result might be different when real world content is employed. Tone tests are great for reference purposes, but they are not always the absolute answer. Especially when the issue has a dynamic aspect to it, as this does. We're not dealing with steady state scenarios like tones, but extremely random dynamic levels and frequencies such as music and voice. Here's a case where test results are significantly different when program material and test tones are used.
This further supports the notion that white papers, test tones, and bench tests are only half the equation. Yes they have value, and I'm not discounting them, but they are not the final say in a matter such as this. Here is a good example where performing a test, as would be done in a broadcast facility, reveals quite a bit more than one might think using test tones.
-Frank Foti
With program material - not single tones - it does not switch to mono when SSB is transmit. Using content with lots of separation like the Beatles, it's easy to subjectively hear the results. It does seem to play games with audio separation a bit, but it is not mono, as we've been lead to believe. Our tests were done using a broadcast setup, not a bench test, where the output from an audio processor was connected to a known FM exciter, and modulation set to 100% (75kHz deviation). FYI: Pilot injection set to 9%. Switching between DSB and SSB was done in the broadcast grade stereo generator.
Worth noting is an interesting anomaly with it in DSB mode. When the audio signal (music) is extremely low, it does switch to mono even though the content is stereo. Must be something in the noise sensing algorithm. The same test done with another receiver does not possess this issue. I've downloaded the Sony patents, and will read them with interest.
Performing tests using program content, as compared to test tones, revealed what a listener will experience in the real world. That is, unless some broadcaster decides to create an all tones, all the time, format.
We use tones for testing and design purposes, but, as has happened many times, the end result might be different when real world content is employed. Tone tests are great for reference purposes, but they are not always the absolute answer. Especially when the issue has a dynamic aspect to it, as this does. We're not dealing with steady state scenarios like tones, but extremely random dynamic levels and frequencies such as music and voice. Here's a case where test results are significantly different when program material and test tones are used. This further supports the notion that white papers, test tones, and bench tests are only half the equation. Yes they have value, and I'm not discounting them, but they are not the final say in a matter such as this. Here is a good example where performing a test, as would be done in a broadcast facility, reveals quite a bit more than one might think using test tones.
-Frank Foti
- Status