Breakaway AM

[konbaasiang]

It worked!!

By clipping bass symmetrically and everything else asymmetrically, it turns out it's actually possible to accurately output an asymmetrical waveform through regular sound cards. 100% negative, 100 to 150% positive.

Check it out:

MPG movie (3 MBytes) showing Tektronix 465 analog scope displaying asymmetric signal from EMU 0404 USB, running Breakaway with -100/+150 clipping: here

Screen shot of settings: here

Audio example (Plutonium Preset, 60hz high pass, 10khz low pass, -100/+150%, Mono):
http://bredband.leif.cx/browse/bafm/audio_clips/bafm_am_mono_44.mp3

Breakaway FM now has the following AM-processing options:

• Strict output bandwidth control - Selectable 5.0, 7.5, 9.0, 10 KHz
• Brick wall high pass filter, selectable 30, 45, 60, 80, 100, 120hz. Brick wall filtering means for example full bass down to 60hz, but virtually nothing below, to prevent transmitter nonlinearities causing IM distortion.
• Tight peak (and valley) modulation control, Asymmetrical clipping, adjustable 100 - 150% positive peaks.
• Line-up tones, tilt and parametric eq control, to equalize transmitter frequency response
• Solid, effortless leveling and multiband compression
• ..and of course every other feature that makes Breakaway FM what it is, such as virtually inaudible clipping (even on difficult program material) within reason.

Download beta:
http://bredband.leif.cx/browse/bafm/ (Yes, Breakaway FM)

Looking forward to comments as always!

Best regards,
///Leif

 

[Tom Wells]

Loaded and running this morning, and I'm very happy.

I'm just waiting for the buy button to pop up now.

The bass hits as hard as a mean older brother, especially when no low cutoff is applied.

This sounds great at any listening level, and made a noticable difference in intelligibility of audio
at fringe coverage.

 

[gunterm]

Way cool! I thought AM was 5khz though?

 

[Tom Wells]

AM can be as wide as it needs. In analog TV video, it is 6 Mhz wide.
5 khz sounds pretty dull, and was discarded (by engineering practive) as an acceptable upper frequency limit in the 1930's.
It was not until the 1980's that anybody thought muffled audio on Broadcast AM was a good idea.
Part 15 AM stations need not limit upper end audio at all.
Most stations have more hi-end info than the average new AM radio will pass.

The BUY link is not ready yet.. guess I'll go to the website.

 

[gunterm]

Yes I MUST buy this! Best AM processing I have found yet, I assume my part 15 AM xmitter will pas 10khz audio, I'm clear on the nearby frequencies here in the 1600s so the interference that full 10khz would generate won't bother anyone...

Time to check my CC Radio specs to see if it passes 10khz, there's no switch so probably will need to borrow a GE Superradio.

Edit: wow this must be a typo on the website: AM: 100 Hz - 2.5 kHz (knob centered)

 

[radiorob2.0]

Yes I MUST buy this! Best AM processing I have found yet, I assume my part 15 AM xmitter will pas 10khz audio, I'm clear on the nearby frequencies here in the 1600s so the interference that full 10khz would generate won't bother anyone...

Time to check my CC Radio specs to see if it passes 10khz, there's no switch so probably will need to borrow a GE Superradio.

Edit: wow this must be a typo on the website: AM: 100 Hz - 2.5 kHz (knob centered)

Not a typo, that is all you get from a CCrane. They promoted the radio as being tuned for talk radio and voice is around 2-2.5 k.

 

[Sgeirk]

Leif,

I'm very impressed with what you're doing. Gotta say, I was listening to some of the FM samples on your site...comparing bafm to an off-air 8500 recording.

I'm in the U.S.A. So, it's irrelevant, really.... but I can't help but notice that every european popular music station I've ever heard sounds like hammered dog crap. In the US, thank god THAT much pumping and breathing is unacceptable...otherwise I'd have a splitting headache every day after work.

Keep moving ahead, in the 30 days I've been messing with bafm...I've seen some very cool features thrown in, it's very interesting watching this develop. Wish I had more time to play.

 

[konbaasiang]

Hi guys!

Thank you for all the great comments.

Gunterm, most AM stations today are 5.0 KHz thanks to IBOC, but I believe you're still free to run 10 KHz if you're not doing IBOC, which nobody is anymore, are they?

Sgeirk, thank you! I agree about that station (3FM), my god it sounds bad. It's actually gotten even worse since that recording was made -- they've cranked up the mids substantially. American stations tend to compress and pump less, but all too often they make up for it with more clip drive, which certain processors starting with an O can't handle very well without things falling apart.


New Version - Beta 42

I've added new lineup tones for AM radio! Please note that my background is in FM -- I have never even played with an actual AM transmitter (we just don't have them in Sweden, where I grew up) -- but I'm doing my best, learning as I go along, with open ears.

These test tones are for viewing on an oscilloscope connected directly to the transmitters RF sample output.

The new test tones are:

• 60hz Sawtooth
• 60hz Sawtooth with detent at 0%
• 400hz Sawtooth

Those three are for verifying transmitter linearity, since they cause nice straight lines on the scope. They follow the asymmetry setting, so if you set Asymmetry to 25%, the sawtooth will go from -100 to +125%, with an optional detent at 0%. When you use asymmetry, a flat bottom gets added to the sawtooth, to prevent non-dc-straight sound cards from decentering it, by keeping positive/negative energy in balance.

If you see the lines bending, your transmitter is not perfectly linear, either due to clipping or something else. If this happens, please take some photos for me, so I can start thinking about what adjustable pre-distortion software algorithm could be appropriate to linearize your transmitter .

Also:

• 1000hz Square Wave

Because Breakaway FM is a digital processor, and must adhere to physical bandwidth limitations, the 1000hz squarewave is not really all that square. Nevertheless, it is useful for adjusting high frequency response of the transmitter.


I have taken photos of an oscilloscope showing AM modulated versions of these signals, to provide an ideal baseline. I did the AM modulation in software (70 KHz carrier) and output the signal through a 192 KHz sound card, so the oscilloscope display is theoretically perfect -- if you can make your transmitter match it, you're good to go! The photos are ideal results except where noted otherwise.

Please see the photos here:

http://leif.cx/temp/amcal

Again, the 1k square is not really all that square. Because BaFM is a digital processor, high frequency response is limited by sampling rate. True squarewaves require infinite frequency response – so I instead created most useful approximation I could. The important part is the sharpness of the corners (as the arrows indicate). Ideal response is the best compromise between sharp corner and overshoot. A negative overshoot means dropping carrier, and we certainly don't want go off the air, it would defeat the purpose . Use the built-in Parametric EQ to adjust.


If you know any other lineup tones useful for AM calibration, please let me know.

Oh, and version 0.90.42 is available here as usual: http://bredband.leif.cx/browse/bafm

Best regards,
///Leif

 

[Martijn]

Leif,

I'm very impressed with what you're doing. Gotta say, I was listening to some of the FM samples on your site...comparing bafm to an off-air 8500 recording.

I'm in the U.S.A. So, it's irrelevant, really.... but I can't help but notice that every european popular music station I've ever heard sounds like hammered dog crap. In the US, thank god THAT much pumping and breathing is unacceptable...otherwise I'd have a splitting headache every day after work.

Keep moving ahead, in the 30 days I've been messing with bafm...I've seen some very cool features thrown in, it's very interesting watching this develop. Wish I had more time to play.

No offence, but I heard some airchecks from US based radiostations also pumping and breathing like hell all because of the loudness war in ... the US.
I agree that the sound of most European radiostations is giving me headaches like French stations, UK based radiostations and, of course, a lot of Dutch radiostations, but than again, there are also radiostations knowing what they are doing with their soundprocessor and what the goal is they want to reach.
Try Radio M (http://mediaplayer.rtvutrecht.nl/stream/radiomutrecht_hi) the still have an 8200 on air and are amongst the best sounding in the netherlands.

 

[jesseg]

Let's not make this about countries or locations. Bad radio happens everywhere.

And need I remind you guys, that this is a thread about the AM part of this new audio processor. Not FM.
8)

 

[Tom Wells]

I'm enjoying the results now for the last few days on my station at home.

Everything sounds as good or better, nothing sounds worse, there's no breathing or pumping as I have it set.
The thud and pow of a giant 50kw from the 1960's is there along with flawless control of high frequency info.

I still maintain that this software is performing actual restorative work on some files which I swear now have
"formerly lost" details once again audible. It's worth buying for audio production work even if you don't have anything to do with radio.

I give my highest possible recommendation for this product.

Kudos to all involved.

 

[konbaasiang]

Thank you kindly, Tom!

Finally, I managed to please you! I am overjoyed .


NEW NAME:

Breakaway Broadcast Processor

AM friends, rejoice! No longer will I treat you like second class citizens by asking you to use an FM processor for your AM stations .

Breakaway Broadcast Processor is for ALL of us.

New convenience feature too: Slider settings are now saved separately for each preset, making A/B comparison MUCH easier.

http://bredband.leif.cx/browse/bafm <-- still the same URL (I apologize in advance ), but the official site is only days away!

///Leif

 

[Ted Russell]

Leif,

Is there any possibility that you can design an option for Matrix Processing for AM Stereo? So instead of processing the discrete Left and Right, it would process L+R and L-R. You could also add a control on the control panel for more or less L-R to enhance or decrease the stereo image in Matrix mode.

TR

 

[konbaasiang]

Ted,

Interesting idea!

After at first dismissing the idea entirely for the reason that it won't work the way the algorithm is written, I realized after thinking about it some more that it actually might work.

One important fact: My AGCs and compressors are hard coupled in code. There is no control for the coupling between channel 1 or 2, because there is really only one gain value shared between the two.

Thus, it will not be a stereo enhancer even if set to mid/side mode, but maybe that's a good thing! I believe uncoupling a mid/side compressor is not a great way to do it.

So, I could add a mid-side matrix to the beginning of the chain, and a dematrix at the end, after the clippers. This might actually work. Thus, L+R and L-R peaks would be accurately controlled, *without* the stereo image being affected. What do you think? Would it be useful?

///Leif

 

[Ted Russell]

Interesting idea!...
One important fact: My AGCs and compressors are hard coupled in code. There is no control for the coupling between channel 1 or 2, because there is really only one gain value shared between the two.

Thus, it will not be a stereo enhancer even if set to mid/side mode, but maybe that's a good thing!...

So, I could add a mid-side matrix to the beginning of the chain, and a dematrix at the end, after the clippers. This might actually work. Thus, L+R and L-R peaks would be accurately controlled, *without* the stereo image being affected. What do you think? Would it be useful?

///Leif

I have given considerable thought to the same problem over the years, and I do believe that there does need to be coupling between the L+R and the L-R pathways, otherwise the stereo image can get way out of control. As a result I would always end up lowering the L-R side considerable to prevent the whole thing from going out of whack.

I do think what you are saying could work, and I think its worth a try. I also believe that this could be an asset for webcasters who want to offer both a high quality stereo stream and a lower quality mono stream at less bandwidth. This way you can simply just run two encoders off the same audio source. It would maintain the punchy-ness in the mono stream and add extra vibrancy to the stereo stream.

Thank you for considering this idea.

Here's another one:

I see you made provisions for stereo multiplexing out of sound cards that can handle it. It looks like there is also some kind of RDS pass through. How hard would it be to incorporate RDS encoding to Breakaway Broadcast? That'd be a real hoot.

Thanks again. Take care!

-TR

 

[konbaasiang]

Hi Ted!

That's encouraging. I'll definitely try it and see how it comes out.

Interesting idea for webcasting. Only thing is, webcasting still has to be peak controlled in the L/R domain -- I don't know how the peak control in the L/R domain will look if I'm clipping in the Mid/Side domain. Only one way to find out.

RDS isn't exactly hard, but it's *tedious*. Apart from the modulation part, there's the whole bitstream.. Endless twiddling of all these little bits of data, tightly packed together.. It's a major project, and my hat is off to the Airomate author for doing it. Airomate is also the reason why I won't do it -- Airomate already does a *perfect* job of RDS encoding.

In audio, there was enormous room for improvement over the previous state of the art - and quite honestly there still is, even beyond the current Breakaway! It CAN still get even louder and even cleaner.

For RDS, there simply is no room for improvement. It's perfect already. Thus, there is no challenge -- and frankly (or bobly), that's what drives me. Plus, Airomate is only $29, so there's no profit in it either .

In short, I recommend using Airomate. It's what the RDS input is for, and due to the way it's structured, the RDS encoder is out of the critical path. Even if something were to go wrong with it, you're still on the air, you're still in stereo -- in fact there is not even a click, as Breakaway crossfades between the incoming pilot and internally generated pilot.

(Correct usage mode for the RDS input is to feed both pilot and phase-locked rds into it. Breakaway will then filter and level both signals and use it for the stereo encoder. In fact, you can even just feed a straight composite signal with RDS, audio and everything, into the RDS input -- Breakaway will take the pilot, the RDS, and throw away everything else.)

Best regards,
///Leif

 

[konbaasiang]

I don't know how the peak control in the L/R domain will look if I'm clipping in the Mid/Side domain. Only one way to find out.

And now I have, and the answer is: Lousy!

Doing mid-side matrixing before the front-end, and dematrixing after the clipper, produced some remarkably lousy peak control. In hindsight -- of course it would. If we have a perfectly tight peak-controlled mono signal, and then add a stereo difference signal to that, of course we're going to have lots of very thin peaks! Up to 100% overshoot, in fact.

So, that approach won't work so well.. Of course, if we go back to Mid/Side domain afterwards (which I guess we'll do for C-QUAM), perfect peak control will be restored again -- of the Mid signal, that is. The Side signal don't get much clipping at all unless there's extreme stereo imaging going on.

So, maybe this is what you were asking for after all?


Mid-side matrixing before the front-end and the dematrixing BEFORE the clipper, worked a lot better. (Dynamics processing Mid/Side, Clippers Left/Right). In fact, if I omitted the 6dB attenuation that is customarily done when dematrixing, it was as loud as L/R processing, and as clean, with the dance track I was playing.

Then, I switched to Good Lovin' by The Rascals, with its glorious hard-panned stereo, and was greeted by *extreme* distortion from clipper overload. Since there wasn't much mid-content to push the compressors down, it ended up driving the clippers way too hard.

Also, in running the clippers on the left/right, we would have lousy peak control when we go to mid/side mode.

It seems, there is no way to have the cake and eat it too (who would have thunk it!). If I attenuate 3dB to compensate for that distortion, mono content will be 3dB quiet.

I may be missing something -- in particular, how exactly C-Quam works -- maybe Mid/Side dynamics and clipping is exactly what is needed? If that's the case, I'll be happy to make the option available.

Best regards,
///Leif

 

[Sgeirk]

Back to RDS for a moment, if you don't mind...we're off thread here...sorry.

Airomate, the way I've been told it works, will only read from a text file. We get our now playing data via a bitstream. So Airomate won't work in that regard. I still believe there's room to improve, but I wouldn't take away from working on breakway to do RDS.

...work on it as your next project.

 

[konbaasiang]

Sgierk, I see your point.

How about, say, a utility that receives your bitstream and writes a text file as it updates?

///Leif

 

[Tom Wells]

It seems, there is no way to have the cake and eat it too (who would have thunk it!). If I attenuate 3dB to compensate for that distortion, mono content will be 3dB quiet.

I may be missing something -- in particular, how exactly C-Quam works

I need a re-read on C-Quam theory but I do seem to remember there being a difference in mono loudness vs stereo loudness, and it may be the same 3 db you are finding.