Audio Proccessing Pre or Post STL

[masterphreak]

I would like to hear any arguments from both camps as to why one way is better that the other.

 

[Kelly A]

It depends on your program chain: Composite STL? Best to have the audio processing-stereo generator at the studio. AES STL? Audio processing at the studio, stereo generator at the transmitter (unless your exciter takes AES in directly). Discrete L/R analog STL? Aphex Compellor limiter ahead of the STL to keep studio levels under control and audio processor/stereo generator at the transmitter site.

 

[Grounded Grid]

Thanks for starting that, Kelly. I was looking at those combinations and had a couple more, but with caveats.

I've seen a couple of stations use the AES in/out of a pair of Bric-Links. I'm pretty sure this equipment (as well as the Barix Exstreamer 1000) does a D/A - A/D conversion, then compresses the audio to whatever form you select for bandwidth purposes. I have a set that uses fiber for the link between sites, so running the Bric-Links in PCM format works fine. Over the public internet however, I have a question about using compressed formats and expecting there will be no overshoots in a processed stream at the receiving end.

There are two other stations at the site I do some work in. My STL is a 2-hop composite, and without a clipper at the transmitter, I have a bit of an overshoot issue (hope to move that to backup-status soon). The other two stations use digital STLs, but have chosen to go L/R in/out and have their processors and RDS generators at the transmitter site. Now that so many processors, remote controls and switching devices are network-accessible, perhaps there's less reason to put a lot of effort into locating the processor at the studio. Even in the case of AM, I have web access to air monitors, processing and modulation monitors, and administer all of it from my office. In my case, there's not much more I could do if the physical equipment was at the studio.

 

[Kelly A]

Thanks for starting that, Kelly. I was looking at those combinations and had a couple more, but with caveats.

I've seen a couple of stations use the AES in/out of a pair of Bric-Links. I'm pretty sure this equipment (as well as the Barix Exstreamer 1000) does a D/A - A/D conversion, then compresses the audio to whatever form you select for bandwidth purposes. I have a set that uses fiber for the link between sites, so running the Bric-Links in PCM format works fine. Over the public internet however, I have a question about using compressed formats and expecting there will be no overshoots in a processed stream at the receiving end.

I'm a big fan of the Comrex Bric Links, not so much with the Barix. We use over a dozen of the Bric Links for IFB and PL connections overseas rather than phone lines. For STL purposes, I suggest treating them similar to a discrete L/R and put some form of limiting ahead of them and the processing at the TX. You can't be sure the talent or guy running the board is going to pay attention to their levels.

There are two other stations at the site I do some work in. My STL is a 2-hop composite, and without a clipper at the transmitter, I have a bit of an overshoot issue (hope to move that to backup-status soon). The other two stations use digital STLs, but have chosen to go L/R in/out and have their processors and RDS generators at the transmitter site. Now that so many processors, remote controls and switching devices are network-accessible, perhaps there's less reason to put a lot of effort into locating the processor at the studio.

Two hop composite STL hops are a bitch. The s/n is never very good and filter ringing/overshoots is always a possibility. What I've done in the past with Moseley 606 units, is bypass the filters at the second hop site. That knocks the filter ring down considerably, but doesn't necessarily help with s/n.

Wherever possible, I always try to keep the audio processing at the studio. Modern processor-powered processing are in essence a PC, with remote transmitter sites being generally inhospitable environments for things like PC's. Remember hearing horror stories of transmitter manufacturers putting hard drives in their exciters? Yeah, that didn't go so well. The old days of 8100 Optimods which were designed for installation at TX sites are long past, that is of course unless you're still running an 8100.

 

[Grounded Grid]

I'm a big fan of the Comrex Bric Links, not so much with the Barix. We use over a dozen of the Bric Links for IFB and PL connections overseas rather than phone lines. For STL purposes, I suggest treating them similar to a discrete L/R and put some form of limiting ahead of them and the processing at the TX. You can't be sure the talent or guy running the board is going to pay attention to their levels.

Agreed. I'm a fan of the Compellor as well.

Two hop composite STL hops are a bitch. The s/n is never very good and filter ringing/overshoots is always a possibility. What I've done in the past with Moseley 606 units, is bypass the filters at the second hop site. That knocks the filter ring down considerably, but doesn't necessarily help with s/n.

My STLs are Nicoms, but I'll look into that.

Wherever possible, I always try to keep the audio processing at the studio. Modern processor-powered processing are in essence a PC, with remote transmitter sites being generally inhospitable environments for things like PC's. Remember hearing horror stories of transmitter manufacturers putting hard drives in their exciters? Yeah, that didn't go so well. The old days of 8100 Optimods which were designed for installation at TX sites are long past, that is of course unless you're still running an 8100.

Generally, I agree with this as well. In my case I'm fortunate that, at the sites I don't have direct control over, the owners pay some attention to environmental conditions... and at the sites where I do, they're mostly air conditioned and closed-looped. Two of my buildings were old AT&T relay sites, and they were pretty finicky about how they applied their filtering and HVAC. Where I do put in computers, I replace the hard drives with SSDs and strip out everything but the required software & operating systems. Haven't lost a computer at a transmitter site in well over 18 years of using them up there.

I used to be worried about CPU fans, as that's the only thing left in my computers that spins. Still, I see processors and STLs from legacy names that have fans in them, so I seem to have come out ahead on that one.

I do understand the concern about putting PCs at the transmitter. I've also heard engineers cringe about putting a PC in their transmitter audio chains, though, as you've said, that's what processors have become now. For various reasons, most related to the budgets of some of my clients, I've jumped into that pool and taken the risk... and haven't been burned yet.

 

[Kelly A]

My STLs are Nicoms, but I'll look into that.

I'm not familiar with Nicom STL radios, but if they are like some of the others built overseas I've run across (BEXT Italian and Spanish models as an example), some have the IF in and out available on the rear panel of the radios. If so, then I recommend using the IF in and out at the repeater point rather than demodulating composite audio only to re-modulate it again. The filters ringing and subsequent overshoots, along with improved s/n would be a significant improvement.

 

[Groove1670]

We don't use anything before our Marti STL 10's. At the site, we have a 8100a/xt2 combo. The result, no modulation or distortion issues for us, and a very clean transparent sound.

 

[Kelly A]

That's dangerous. You deviate those old Marti's and one or more things would occur: 1. Exceed your licensed bandwidth (overmodulation). 2. Distort the audio badly. 3. Cause the rock-bound oscillators and AFC to go out of lock. None of these affects would be very clean sounding.

Your management must have a 'watch-your-levels-or-die' edict with the people running the board or doing production.

 

[Groove1670]

Nope. No modulation issues. No distortion. I've yet to kill a DJ for overmodulation Now, I don't overdrive the Marti's. I let the 8100A/XT2 do the work. However a Compellor is in the budget for the future.