Digital Audio Chain

I'm very good with balanced analog audio, and know it very well, however, now, with a Voltair, PPM encoder, and sometimes a profanity delay, in the totally analog audio chain, en route to the transmitter audio processor, I'm wondering if there are a number of A-D and D-A conversions taking place in the Voltair, PPM encoder, and profanity delay, which now (cumulatively) degrade the final product. Our audio (that audio chain that I just mentioned), sometimes sounded different coming out of various exciters (using the exact same audio chain). There was a Continental 802B (dense sounding), a Harris MS-15 (amazingly clean), and now a Nautel 3.5 (producing the worst sounding audio of them all). I'm feeding a composite signal from the Omnia 11 into the Nautel. Before you all condemn me for doing that, also know that that is the way I fed the 802B and MS-15. We still have analog boards, so I don't, as of yet, have a digital signal to work with. I am wondering if I convert the console analog output, to digital (AES), and keep it digital throughout the audio chain I just mentioned above, if that will clean up my audio coming out of the Nautel. If I do that, do I have to watch out for clipping on the console A-D conversion, as we have some air talent that still peg the console meters? Any recommendations for a A-D converter to use, in this application?

Brian Bowers said:

I'm very good with balanced analog audio, and know it very well, however, now, with a Voltair, PPM encoder, and sometimes a profanity delay, in the totally analog audio chain, en route to the transmitter audio processor, I'm wondering if there are a number of A-D and D-A conversions taking place in the Voltair, PPM encoder, and profanity delay, which now (cumulatively) degrade the final product. Our audio (that audio chain that I just mentioned), sometimes sounded different coming out of various exciters (using the exact same audio chain). There was a Continental 802B (dense sounding), a Harris MS-15 (amazingly clean), and now a Nautel 3.5 (producing the worst sounding audio of them all). I'm feeding a composite signal from the Omnia 11 into the Nautel. Before you all condemn me for doing that, also know that that is the way I fed the 802B and MS-15. We still have analog boards, so I don't, as of yet, have a digital signal to work with. I am wondering if I convert the console analog output, to digital (AES), and keep it digital throughout the audio chain I just mentioned above, if that will clean up my audio coming out of the Nautel. If I do that, do I have to watch out for clipping on the console A-D conversion, as we have some air talent that still peg the console meters? Any recommendations for a A-D converter to use, in this application?

Click to expand...

The Omnia 11 has a headphone jack. Does the audio sound OK there?
I find it interesting that your best audio comes from a MS-15! A modern Nautel exciter should have pristine audio performance. A 802B, if in good operating condition, should not color the audio in that way.
There is something not right with your audio chain. You also posted a while back that your Orban 8500 would not catch peak energy which is the first time I've ever heard someone have that issue. Do you have another engineering friend who could maybe look over your shoulder? Sometimes even the best tech/engineer misses the obvious...

BTW...although some have a different opinion, I run analog composite into my Nautel. You may be able to measure the difference with lab grade equipment but I seriously doubt ANY listener would notice the difference. If your Nautel exciter has a digital multiplex input you might try that. Again...the difference can be measured but can listeners hear it?

>>>The Omnia 11 has a headphone jack. Does the audio sound OK there?<<<

Good question. I will check. This is mostly a part-time gig for me now, so I am only mostly there on the weekends.

>>>I find it interesting that your best audio comes from a MS-15!<<<

I know it, but it's true (IMO). Out of all the exciters that I have heard over my lifetime, I would have to say that that one MS-15 had the highest fidelity of them all (crazy, I know). When I was driving one day, while listening to the backup (which was using the MS-15), I couldn't believe what I was hearing. The highs were immaculate. We, however, are no longer using that exciter, as we now have the Nautel 3.5 (for the main), and a Harris PT4, (? I think that's the one) for a backup. Even though the Harris PT4 exciter looks like a Harris Digit, or Harris CD Digit, I think that I saw that it was marked SuperCiter. That's the first time I have ever seen or heard of that model. If you're wondering if the SuperCiter sounds different than the Nautel, I'll have to get back to you on that one, as I don't remember, but I am curious to find out.

>>>A 802B, if in good operating condition, should not color the audio in that way.<<<

One would think, but in my opinion, almost every exciter has it's own audio personality/signature.

>>>You also posted a while back that your Orban 8500 would not catch peak energy<<<

That is correct. That was me, however, that was/is for a totally different FM. That situation is on the back burner for now.

>>>Do you have another engineering friend who could maybe look over your shoulder? Sometimes even the best tech/engineer misses the obvious...<<<

I have several, but my situations and questions usually stump them, even though they are all seasoned veterans. Personally, I have a good 20-30 years of this stuff myself. I definitely don't mind picking people's brains though.

I have also set up a selector monitoring system, where I can select a tuner (which can cleanly monitor the local competition), mod monitor output, or the transmitter audio processor input. As of yet, I have not connected the processor input audio to the selector switch. I plan to do that this weekend. That system is in place for the Nautel 3.5/Harris PT4 facility. The output of this system feeds some Dorrough Loudness meters, a decent external headphone amp, and a speaker amp.

I'm noticing that the Harris SuperCiter is an analog only input exciter, but I also notice that the composite input is selectable between balanced and unbalanced. From the SuperCiter spec sheet:

Wideband Composite Performance
Input: Unbalanced, jumper selectable 50Ω or 10,000Ω (resistive), BNC
jack or Balanced (jumper selected), 10,000Ω (resistive), XLR jack.

I believe the Omnia composite outputs are also selectable (balanced/unbalanced). Maybe this is where my problem is. If one is set for balanced, and the other unbalanced, could this cause a difference in sound? I know that balanced analog audio is much better than unbalanced, especially in the broadcast world (with lots of RF nearby), is it the same for running balanced/unbalanced composite?

I'm thinking that the unbalanced/balanced BNC composite has more to do with the length of the BNC cable being run, than the sometimes muddy audio that I am hearing. Balanced (composite) is better for long lengths, while unbalanced is better suited to shorter lengths. Anybody know if I'm right on that? When it comes to some of our audio sounding like bad MP3's, I don't think that it's related to the composite outputs that we're using. RDS and stereo pilot all seem to be fine. Modulation is perfect. I still think that this is a cumulative result of all of the A-D and D-A conversions taking place, en route to the audio processor.

For those that are curious about the BNC composite connection issue, here are responses from Telos/Omnia, and Orban.

Telos/Omnia:
The composite output of all the Omnias (and most stereo generators) is unbalanced. And most exciter and STL inputs are unbalanced "bridging" inputs. I think some exciters have that option for hum elimination (ground loop issues) on long runs etc. Normally, just set it to unbalanced mode.

Orban:
The answer is in the proof-of-performance. Use whichever mode is the most quiet. If using a composite line over 100’ use RG6 and an isolation transformer if needed.

Brian Bowers said:

I'm very good with balanced analog audio, and know it very well, however, now, with a Voltair, PPM encoder, and sometimes a profanity delay, in the totally analog audio chain, en route to the transmitter audio processor, I'm wondering if there are a number of A-D and D-A conversions taking place in the Voltair, PPM encoder, and profanity delay, which now (cumulatively) degrade the final product. Our audio (that audio chain that I just mentioned), sometimes sounded different coming out of various exciters (using the exact same audio chain). There was a Continental 802B (dense sounding),

Click to expand...

The 802B sounds dense due to excessive amounts if Intermodulation Distortion and noise they produce. Those exciters weren't very good to begin with.

Brian Bowers said:

a Harris MS-15 (amazingly clean),

Click to expand...

By today's standards, those exciters are antique junk. In fact, they were junk back in the 80's when new. If you did a proof on either of those exciters, you would be amazed how bad they really are.

Brian Bowers said:

and now a Nautel 3.5 (producing the worst sounding audio of them all). I'm feeding a composite signal from the Omnia 11 into the Nautel. Before you all condemn me for doing that, also know that that is the way I fed the 802B and MS-15. We still have analog boards, so I don't, as of yet, have a digital signal to work with. I am wondering if I convert the console analog output, to digital (AES), and keep it digital throughout the audio chain I just mentioned above, if that will clean up my audio coming out of the Nautel. If I do that, do I have to watch out for clipping on the console A-D conversion, as we have some air talent that still peg the console meters? Any recommendations for a A-D converter to use, in this application?

Click to expand...

Okay so you're probably not going to want to hear this Brian, but I'll try to be as kind as textual commentary can be.. First, I never want to discourage someone from getting into, or growing in the broadcast technical business, but sometimes we can get sucked into the Program Director or GM mentality of what "sounds good" or doesn't. Seasoned engineers learn over time that when the technical specifications and proper integration of equipment is correct, THAT is when you also get the best "sound". Modern gear will always function better than old gear, and swapping out exciters with old junk is not a sustainable methodology for reliable and consistent operation. In other words, taking the opportunity to make the signal chain as clean and simple as possible. That includes: paying attention to impedance of cables, noise sources that can bleed into gear or cabling, and above all; keep the program chain as SIMPLE as possible. Simple doesn't include daisy-chaining old gear with opamps and dried-up capacitors that contribute to cumulative distortion and noise to your signal.

Kelly A, I couldn't agree with you more. I could hear the distortion on the 802B, I'm sure that's why it sounded dense and fat. I agree with you on the MS-15 too. I'm sure if I looked at, or listened to, the S/N, or listened closely for distortion, I'm sure it was there, but I'm not kidding, those highs were smooth as silk. That box was way past its time, so I'm glad we no longer have to rely on it.

The investigation continues as to why the Nautel doesn't sound as clean as I think it should. I've audited and analyzed the entire chain, and connection wise, it is as clean as it can get. I simply have to keep in the EAS box, the Airtools 6100 profanity delay (we do bypass it when not needed), the Voltair, and the PPM encoder. There's no getting around those. The ultimate goal would be to have two routes from the on-air board, digital, and analog, and for quality comparison, simply switch back and forth. Then, one day when we get digital output boards, some of my work (and research) will already be done.

I've found the Nautel exciters to be pretty good. Have you checked to see if the preemphasis on the exciter is turned on? Your audio processor is already outputting a preemphasized signal, so it would certainly make a difference if the exciter was to to as well.

I'm not sure I'm understanding you correctly. I believe only one source should be the pre-emphasis provider - either the audio processor, or the exciter, not both. That's the way I've always understood it. I'm assuming that's what you're saying here too, yes?

My hunch here is that the PD is sometimes uploading less than .wav files for the new music automation (NexGen) source. These then get A-D'd, and D-A'd a few times en route to the Omnia 11. The final picture (so to speak) is then not as clear as the original snapshot. I'm sure that there is currently at least one pre-emphasis source on. I doubt two, otherwise the air signal would be overly bright.

If there really are a number of A-D, and D-A conversions taking place, maybe I need a digital word clock. Reading up on word clocks, I just stumbled upon this:

>>>However, clock timing is absolutely critical whenever data is being converted between the analogue and digital domains, so the clocking of A‑D and D‑A converters is of fundamental importance to the quality of the audio.<<<

I realize that most of you are already familiar with the digital AES/EBU world, but I'm pretty much just entering it now.

Brian Bowers said:

I'm not sure I'm understanding you correctly. I believe only one source should be the pre-emphasis provider - either the audio processor, or the exciter, not both. That's the way I've always understood it. I'm assuming that's what you're saying here too, yes?

Click to expand...

Not necessarily. Remember that Nautel exciters are made for worldwide distribution. Depending on the configuration, some stereo generators from overseas don't provide preemphasis because the European preemphasis curve is different than the US. You select the preemphasis in the exciter (75 or 150 microseconds). Running 75us from an audio processor into 150us (as an example) into the exciter will cause the sound to sound overly compressed or clipped. It's something to check anyway..

Brian Bowers said:

My hunch here is that the PD is sometimes uploading less than .wav files for the new music automation (NexGen) source. These then get A-D'd, and D-A'd a few times en route to the Omnia 11. The final picture (so to speak) is then not as clear as the original snapshot.

Click to expand...

Ah HA! You may be onto something here Brian.. Your comment about MP3's is key: First of all, MP3 files played over a radio station with digital processing and a digital program chain is BAD. Back in the early days of AES/digital/audio chains and broadcasting, there was a lot of concern about "dual-ling codecs". Rightfully so, the concern was the artifacts and poor quality that can occur when playing material with very different sample rates and codecs through a program chain. Let's say for example your audio processor is doing and A-D from your analog console and is set for 44.1kbps, or the same rate as CD's. Down the program chain your FM exciter sample rate 48kbps. Even though you're going in composite, there is an A-D to convert the analog composite into some form of digital sampling. So in this example; now you have 44.1kbps feeding 48kbps. No big deal, but play an MP3 file (with lossy compression), sampled at 32kpbs, and you end up with noticeably ugly sound on those songs. Now of course your PD will say that the song sounds fine on his car stereo, but it isn't being processed at varying sample rates through the program chain of a radio station.

I just received confirmation from Symetrix that my Airtools 6100 profanity delay does in fact go through an A-D and D-A process when using the analog inputs and outputs. I pretty much knew the answer to that one, as that degradation is easy to hear, with it engaged. Still checking on the Voltair and Nielsen PPM encoder. If both the Voltair and PPM encoder do the same, I think I've found my problem.

>>>Not necessarily. Remember that Nautel exciters are made for worldwide distribution. Depending on the configuration, some stereo generators from overseas don't provide preemphasis because the European preemphasis curve is different than the US. You select the preemphasis in the exciter (75 or 150 microseconds). Running 75us from an audio processor into 150us (as an example) into the exciter will cause the sound to sound overly compressed or clipped. It's something to check anyway..<<<

OK will do, but, let's just talk US here. I believe I'm correct in stating that only one provider of pre-emphasis should be turned on, yes? In other words, if I have 75us of pre-emphasis turned on, on the exciter, the audio processor's pre-emphasis should be turned off, and vice versa. That's correct, right?

>>>First of all, MP3 files played over a radio station with digital processing and a digital program chain is BAD.<<<

Definitely, and I'm trying to educate everyone on that. Unfortunately, the PD is the worst offender.

So when it comes to sampling rates in the exciter and audio processor (or any piece of digital audio equipment), I'm assuming higher is always better, yes? Is it better for all pieces of equipment ('s sampling rates) to match? I'll have to look over my manuals for some of these pieces of equipment, and see if this is explained in length. I don't think that it is though.

If you're hearing the artifacts of multiple code stripping or plain ol' mp3 audio you should hear those artifacts on every exciter. You should hear that on the Omnia headphone jack.
I'm still having trouble digesting the MS-15 audio comment.
Kelly: I don't have the intermod and noise specs to look at but I have heard some very good sounding audio come out of those 802 boxes. A properly setup 802B should not have audible grunge - mine never did. I do have a trained ear but I can't hear the difference between copper and gold plated wire so my ears may be suspect <grin>

Kelly A said:

Rightfully so, the concern was the artifacts and poor quality that can occur when playing material with very different sample rates and codecs through a program chain. Let's say for example your audio processor is doing and A-D from your analog console and is set for 44.1kbps, or the same rate as CD's. Down the program chain your FM exciter sample rate 48kbps. Even though you're going in composite, there is an A-D to convert the analog composite into some form of digital sampling. So in this example; now you have 44.1kbps feeding 48kbps. No big deal, but play an MP3 file (with lossy compression), sampled at 32kpbs, and you end up with noticeably ugly sound on those songs. Now of course your PD will say that the song sounds fine on his car stereo, but it isn't being processed at varying sample rates through the program chain of a radio station.

Click to expand...

You're confusing kHz with kbps here. CD quality audio in WAV is 44.1kHz sampling, 16 bits stereo for approximately 1411 kbps.

You can then encode that to MP3 at (for example) 44.1kHz, 16 bits stereo, and 256 kbps.
A 32kbps MP3 is like listening to a 1000 watt AM signal at 100 miles: barely intelligible.

Your point is clear: if you start with a minimally acceptable MP3 and start treating it badly, you will end up with something that is not acceptable. If you start with a pristine WAV, and treat it badly, you will have something that still sounds decent.


=====

Now, Brian, with respect to your various exciters: It is totally possible that the MS-15 sounds best paired with your airchain and transmitter, but that could simply mean that the MS15 is simply incapable of reproducing the signal provided by the processor (or the PPM generator).

Yes, you should have only one pre-emphasis filter applied to your signal.

Brian Bowers said:

OK will do, but, let's just talk US here. I believe I'm correct in stating that only one provider of pre-emphasis should be turned on, yes? In other words, if I have 75us of pre-emphasis turned on, on the exciter, the audio processor's pre-emphasis should be turned off, and vice versa. That's correct, right?

Click to expand...

Correct. You only want one device with a preemphasis curve. Usually that's the audio processor, since it takes into account the extra high frequencies into it's peak limiting scheme. The exciter should be set to flat.

Brian Bowers said:

So when it comes to sampling rates in the exciter and audio processor (or any piece of digital audio equipment), I'm assuming higher is always better, yes? Is it better for all pieces of equipment ('s sampling rates) to match? I'll have to look over my manuals for some of these pieces of equipment, and see if this is explained in length. I don't think that it is though.

Click to expand...

It's always better to have everything set for the same sample rates where applicable.
Sometimes 44.1 to 48 isn't a big deal, but 48 to 96 or 32 to 48 won't work well.

wavo said:

Kelly: I don't have the intermod and noise specs to look at but I have heard some very good sounding audio come out of those 802 boxes. A properly setup 802B should not have audible grunge - mine never did. I do have a trained ear but I can't hear the difference between copper and gold plated wire so my ears may be suspect <grin>

Click to expand...

That's why I'm not a big fan of judging the quality of things like exciters on how they sound. I've worked on several 802B's over the years, mainly because several of the stations I was involved with had Continental transmitters that came with the 802B exciters. Ultimately I replaced most all the 802's with BE FX-50 exciters. The complaints I had with the Continental 802B were:

Microphonics.. Even the stupid muffin fans would create noise in the modulator stage which in turn created noisy audio. I had a transmitter which I just couldn't get the AM noise down to reasonable levels. And I'm obsessive about low AM noise. Come to find out, it was the noisy 802B creating what was measured as synchronous AM noise.

IM Distortion: If you just ran tones through the exciter, distortion was reasonable until you reached about 10.0khz, then the LP filters started ringing. When you ran processed audio with a lot of high frequencies, the IM went through the roof. Some PD's loved that grungy, over processed sound, but it cost you usable modulation and colored high frequencies, but not in a good way.

PTBoardOp94 said:

You're confusing kHz with kbps here. CD quality audio in WAV is 44.1kHz sampling, 16 bits stereo for approximately 1411 kbps.

You can then encode that to MP3 at (for example) 44.1kHz, 16 bits stereo, and 256 kbps.
A 32kbps MP3 is like listening to a 1000 watt AM signal at 100 miles: barely intelligible.

Your point is clear: if you start with a minimally acceptable MP3 and start treating it badly, you will end up with something that is not acceptable. If you start with a pristine WAV, and treat it badly, you will have something that still sounds decent.

Click to expand...

You're right, I totally mis-typed. I was in the process of working on a report about some VoIP stuff at work, and transposed kbps instead of kHz. I do so much video and audio over IP anymore, my brain is getting slower at switching back.

I always try to impress upon anyone who will listen, is to stay away from audio recorded with lossy codecs like MP3, even though the sample rate is 44.1kHz. Back in the day I used to struggle with spots recorded as MP3 files literally being attached to an E-mail for cost savings and speedy delivery. Problem is, many times the client will be upset anyway when his spot sounds swishy or weird on your station. We used to insist on minimum 44.1 .wav files delivered to our dropbox.

Kelly A said:

Back in the day I used to struggle with spots recorded as MP3 files literally being attached to an E-mail for cost savings and speedy delivery. Problem is, many times the client will be upset anyway when his spot sounds swishy or weird on your station. We used to insist on minimum 44.1 .wav files delivered to our dropbox.

Click to expand...

Hah. Off topic, but out here in podunk we struggle with one particular agency sending us audio that must have been recorded on an iPhone at halftime of a basketball game. Immense background noise, echo, sometimes clipping. It's embarrassing to air. But their money is green...