Ariane with 8100& XT2 or Broadcast Warehouse DSP-Xtra

[stace]

I have to disagree with Goran about using the built-in AGC's. Goran, have you ever heard what the 8200's AGC does on say Dreams by Fleetwood Mac? That really loud cymbal crash near the end of the song plays havoc with the 8200's AGC.

The AGC in the 8200 is it's biggest problem. It makes a mash of everything if you speed the release time up, but can't recover quick enough if you slow it down. We had an 8200 at the xtr site with ariane at the studio. Sounded great

 

[gtomas]

I have to disagree with Goran about using the built-in AGC's. Goran, have you ever heard what the 8200's AGC does on say Dreams by Fleetwood Mac? That really loud cymbal crash near the end of the song plays havoc with the 8200's AGC. Disabling the 8200's AGC and inserting a Compellor in front of the 8200 resolved this.

Never used 8200, so I really can't comment.

All I'm saying is if you are adding pre-processing, make sure it actually works better with the particular make and model of the processor, than what the designer carefully created and tuned in the box. I would stay away from generalizing and making "rules"...


Regards,
Goran Tomas

 

[gtomas]

But this whole thing is a judgement call.if one is not pleased with the internal agc,then go with an external.Just don;t sell yourself short by not having the internal agc set for it's best performance.

Exactly.


Regards,
Goran Tomas

 

[radioengr]

Be assured that DSPXtreme has a very capable and consistent wide-ranging window-gated multi-band AGC. Or better yet, try it! Plus we have a few more ideas to bring to the levelling table.

Regards,
Goran Tomas

Maybe there's more Arianne in the Extreme than their letting on. I think I saw a picture from FMister and the control and meter LEDs look just like the Arianne.

Maybe Mr. Reaves can tell us if they are using his code or Goran can tell us how their levelling table is different from the Arianne.

 

[PaulyBoy]

Never used 8200, so I really can't comment.

Hmmm...Never used this product, so how can you make any claim otherwise? Typical young self-proclaimed kid who should listen and learn before opening his mouth. But you're arrogant enough to make a claim that hardly supports your view. Maybe you should keep your mouth shut until your efforts match or exceed those of others.

 

[fm-engineer]

Never used 8200, so I really can't comment.

Hmmm...Never used this product, so how can you make any claim otherwise? Typical young self-proclaimed kid who should listen and learn before opening his mouth. But you're arrogant enough to make a claim that hardly supports your view. Maybe you should keep your mouth shut until your efforts match or exceed those of others.

Easy now.....

Most of us have been around block a few times with processing. I think you misunderstood Goran. He is hardly self-proclaimed, but more like a respected processing guru. We can debate and discuss without personal attacks. I find it hard to believe we still discuss the 8200 in terms of today's markets. Not pushed, it sounds fine for the first DSP box in 1991. As a comparison, the two-band Optimod 2300 is cleaner, louder, and brighter. If someone is stuck with an 8200, maybe they are searching for a competitive boost with Arianes's and such. You could sell the 8200 and have enough money to buy a DSPX or an Omnia 1. The DSP power is light years ahead of the 8200. If you like Orban flavor, the 5300 is the best value for the money. It will blow the doors off an 8200.

 

[busyradioguy]

Most of us have been around block a few times with processing. I think you misunderstood Goran. He is hardly self-proclaimed, but more like a respected processing guru. We can debate and discuss without personal attacks. I find it hard to believe we still discuss the 8200 in terms of today's markets. Not pushed, it sounds fine for the first DSP box in 1991. As a comparison, the two-band Optimod 2300 is cleaner, louder, and brighter. If someone is stuck with an 8200, maybe they are searching for a competitive boost with Arianes's and such. You could sell the 8200 and have enough money to buy a DSPX or an Omnia 1. The DSP power is light years ahead of the 8200. If you like Orban flavor, the 5300 is the best value for the money. It will blow the doors off an 8200.

Yeah we're stuck w/ the 8200 right now, primarily due to low budget issues. I think the unit is basically working in limiter mode only. Overall the sound is pretty good, although personally the EQ could use some minor tweaking. I'm sure one of these days, there will be enough funding to upgrade.

 

[dspxscott]

Maybe there's more Ariane in the Extreme than their letting on. I think I saw a picture from FMister and the control and meter LEDs look just like the Ariane.

Maybe Mr. Reaves can tell us if they are using his code or Goran can tell us how their levelling table is different from the Ariane. 

To pick up on this reply and other points made in this thread.

Customer feedback of the DSPXtra told us two things.
1. Some customers did not want to pay the premium for the Translantech AGC and had no problems with the AGCs already in BW products (the same way that many O and O customers feel about their respective boxes)
2. The customers wanted the option to (pre)process ahead of an STL. This would make the Ariane in the FM processor redundant and not the best economical decision.

A decision was made with the DSPXtreme to have the Ariane available as a plug-in upgrade. This is the most logical option for BW, TLT and Ariane fans.
The plug-in should be available later on in the year once BW and TLT have finalised details.
Unlike the DSPXtra which has the Ariane hard coded into the box at factory, the DSPXtreme Ariane plug-in can be installed at a later date and purchased from Translantech and possibly its distributors. Full details are to be confirmed in due course.

As to there being more Ariane to the DSPXtreme than we are letting on...
The AGC in the DSPXtreme is a four band design like the Ariane. Note: All of the multi-band AGCs in BW products are 4 band, including the original DSPX.
The DSPXtreme AGC and Ariane currently share a very similar X-over topology. This is scheduled to change as we let some of Goran's work filter up through the range.
The AGC in the DSPXtreme has a different control range to that of the Ariane.
The AGC in the DSPXtreme is coupled to a certain extent to the following multi-band limiters.
The AGC in the DSPXtreme does not support Matrix operation ( An Ariane hallmark feature)
The AGC in the DSPXtreme does not support adjustable window gating (Another Ariane hallmark feature). Note: The original DSPX supported adjustable window gating long before the introduction of the Ariane sequel.


As to the code and tables.....
All of the BW manufactured processing products (including the Ariane sequel) employ different processing topologies
All of the BW processing products share code and routines (I wrote them all so why wouldn't they)

Both BW broadcast and Translantech will endeavour to answer any technical enquiries or product questions. Certain questions relating to DSP code, routines and algorithms is BW and TLT company confidential and for obvious reasons will not be discussed on a public forum in any depth.

I am sure Goran will pick up on some of the other issues later in the day.

Best regards
Scott

 

[busyradioguy]

OK I am a bit curious here... Why is there ever a need for more than one band of AGC? It seems to me that multiband compression and limiting is the only place where band splitting is appropriate. If you split the audio in the AGC stage ahead of the multiband C/L processing, does that not further deteriorate the audio quality?

 

[gtomas]

The AGC in the DSPXtreme does not support adjustable window gating

Actually the DSPXtreme has an adjustable gating window (Scott must still be recovering from his jet-lag )


Regards,
Goran Tomas

 

[wgliradio]

Goran and I are usually on the same page, but I do have to say that almost all of the FM's in New York City employ an AGC over using the stock AGC. And six of those stations use Ariane.

 

[gtomas]

OK I am a bit curious here... Why is there ever a need for more than one band of AGC? It seems to me that multiband compression and limiting is the only place where band splitting is appropriate. If you split the audio in the AGC stage ahead of the multiband C/L processing, does that not further deteriorate the audio quality?

It may be advantageous to split audio in more bands even for AGC, to avoid spectral gain modulation effects such as pumping. This is particularly important with bass, as low frequency energy can dominate the wideband AGC's control signal and modulate the rest of the spectrum in an unnatural way. Very broadly speaking, multiband AGCs can be made to operate faster than the wideband AGCs while not suffering from the mentioned negative effects. Plus they will provide some dynamic re-equalization.

The actual advantages of multiband AGC over wideband AGC in the particular processor will very much depend on the design itself, the control circuit, speed of reaction times, etc. and how is the AGC integrated with the rest of the processing - how much "ground" (gain reduction) and at what speed is AGC expected to cover to keep the following dynamics sections in the "sweet spot".

As far as the distortion of splitting audio in more bands, if done properly it is not an issue. The FIR crossover will have linear phase, and all FIR filters in BW processors are designed to achieve system's overall +/-0.1 dB frequency response flatness. To put it in perspective, the acceptable frequency deviation in frequency response for a loudspeaker to be considered high-fidelity is +/- 3dB. Regarding quantization and round-off noise, our processors typically have over 90dB of SNR.

In some processors we use IIR crossovers as well. This crossover will not have perfectly linear phase, but if care and attention is taken to it's design, it can very much approach it. It should perhaps be useful to note that human ear is much less sensitive to phase deviation than it is to amplitude deviation. In fact, in audio processing phase deviation is often used deliberately (for example in phase rotator) because of the benefits gained are much more significant than the subtle audible effects of it (which are often inaudible). A typical passive loudspeaker will again introduce significantly more phase distortion. IIR crossovers are much less computationally costly and so allow the hardware to be more optimally used for other more critical processing stages.

Crossovers have gotten a bad reputation, because many of them weren't designed properly (even in respectable companies and products you might actually use). Additionally, a lot of crossover done today are still not good. They will be flat at unity gain as they are mostly tested that way, but when different bands have different gains (as it happens dynamically) will exhibit peaks and dips at crossover frequencies. They will also not pass transients unaffected. Obviously these will not sound as good as a properly designed crossover.

We at BW pay care, attention, research and testing into crossover design and they are state-of-the-art. As such, having an additional AGC crossover will introduce negligible sonic impairments.


Regards,
Goran Tomas

 

[FFoti1]

As far as the distortion of splitting audio in more bands, if done properly it is not an issue. The FIR crossover will have linear phase, and all FIR filters in BW processors are designed to achieve system's overall +/-0.1 dB frequency response flatness. To put it in perspective, the acceptable frequency deviation in frequency response for a loudspeaker to be considered high-fidelity is +/- 3dB. Regarding quantization and round-off noise, our processors typically have over 90dB of SNR.

In some processors we use IIR crossovers as well. This crossover will not have perfectly linear phase, but if care and attention is taken to it's design, it can very much approach it. It should perhaps be useful to note that human ear is much less sensitive to phase deviation than it is to amplitude deviation. In fact, in audio processing phase deviation is often used deliberately (for example in phase rotator) because of the benefits gained are much more significant than the subtle audible effects of it (which are often inaudible). A typical passive loudspeaker will again introduce significantly more phase distortion. IIR crossovers are much less computationally costly and so allow the hardware to be more optimally used for other more critical processing stages.

Crossovers have gotten a bad reputation, because many of them weren't designed properly (even in respectable companies and products you might actually use). Additionally, a lot of crossover done today are still not good. They will be flat at unity gain as they are mostly tested that way, but when different bands have different gains (as it happens dynamically) will exhibit peaks and dips at crossover frequencies. They will also not pass transients unaffected. Obviously these will not sound as good as a properly designed crossover.

We at BW pay care, attention, research and testing into crossover design and they are state-of-the-art. As such, having an additional AGC crossover will introduce negligible sonic impairments.

Nothing new here.

Your research about 'other' crossovers and performance can be accessed where? Anyone can make bold non-supported statements. Good engineering practice requires data and proof to support a claim.

-Frank Foti

 

[menotti1]

Nothing new with using external AGC's like ariane,compellor,texar,gentner,level magic when employing a STL.Lots of reasons to do that,but it does not necessarily mean the AGC in the processor is lacking in any way.

 

[dspxscott]

Goran and I are usually on the same page, but I do have to say that almost all of the FM's in New York City employ an AGC over using the stock AGC.  And six of those stations use Ariane.

I love the Ariane and David Reaves is a bloody genious but I don't think it is fair to draw anything from the on-air boxes in the hometown of the processing company in question. I bet cleveland doesnt have too many Orbans on the Air either. ???

Do you think the six stations you mention would remove the Arianes if they installed one of those all singing all dancing vorsis boxes that apparently work so well?

The jetlag is nothing compared to the food poisoning. Thanks Singapore airlines for the double microwaved egg breakfast

 

[menotti1]

No,they wouldn't.Hell,they'd need them worse that ever with that Vorsis 31 bands of pump and suck and it's recording studio style compressors.They had to couple bands together in groups just to get the thing to pass something usable.Actually the stations using it would need an ariane on BOTH ends.

 

[rrsounds]

Unless you need to dump off spectra really fast, (i.e., over a narrow frequency range) IIR filters do a good job.
Their first-order phase delay is negligible, and second order and up can be compensated for.

One thing I don't see much mention of in any books or press is the relevance that the rolloff of an IIR's response is logarithmic, while an FIR's is linear.

Since we humans seem to hear frequencies in a log relationship (musical octaves and all that), it stands to reason a simple garden-variety IIR's response can be considered more 'musical' while your garden-variety FIR's rolloff, not being at all musically related... Well, draw your own conclusions. It probably doesn't really matter.
Just an observation on my part.

Kind Regards,
David

 

[gtomas]

Nothing new here.

Your research about 'other' crossovers and performance can be accessed where? Anyone can make bold non-supported statements. Good engineering practice requires data and proof to support a claim.

That's actually not a bad idea... If I manage to find the time to do it, I guess it would make a fairly interesting white paper with examples of typical bad crossover designs.

I really don't want to name the units that got it wrong, but it will be sufficient to say they are analog to know that I wasn't referring to Omnia even remotely. I'm pretty sure other processing manufacturers also have competent crossover designs.


Regards,
Goran Tomas

 

[rrsounds]

Goran and I are usually on the same page, but I do have to say that almost all of the FM's in New York City employ an AGC over using the stock AGC. And six of those stations use Ariane.

I love the Ariane and David Reaves is a bloody genious but I don't think it is fair to draw anything from the on-air boxes in the hometown of the processing company in question. I bet cleveland doesnt have too many Orbans on the Air either. ???

Don't know about the 'genoius' part, but, yeah, I just walk into the lobby of a New York radio group and tell them I was a CE of a local station in town about 10 years ago, and they all take out their checkbooks.

That's how it works in London, too, right?


Kind Regards,
David

 

[FFoti1]

That's actually not a bad idea... If I manage to find the time to do it, I guess it would make a fairly interesting white paper with examples of typical bad crossover designs.

I really don't want to name the units that got it wrong, but it will be sufficient to say they are analog to know that I wasn't referring to Omnia even remotely. I'm pretty sure other processing manufacturers also have competent crossover designs.

Again, nothing new here. There are some very good analog X-Over methods, along with some that left a bit to be desired. Had you made this post 10 - 15 years ago, there was a story to tell. DSP enabled vastly improved X-Over methods long ago.

-Frank Foti