Interesting 6moons article- The Future of Loudspeaker Design

mac · « **on:** 3 Feb 2005, 02:48 pm »

http://www.6moons.com/industryfeatures/deqx/deqx.html

For some of us the future is already here.

goskers · « **Reply #1 on:** 3 Feb 2005, 04:21 pm »

I have heard that all the cool people are using this DEQX thing.

I am finishing the aluminum and wood work on my amp enclosure then I can finish the wiring. This should be my final step before getting everything to a functional level.

goskers

Russtafarian · « **Reply #2 on:** 3 Feb 2005, 07:58 pm »

If the DEXQ was marketed to the pro/musician audio market, it would list for $1200 and be available on the street for $800. I'm intrigued by the technology but not ready to jump at it's present price.

Interesting thoughts by Srajan regarding interfacing analog sources. I'm not ready to ditch my modest vinyl rig and record collection. Why hasn't anyone developed a hi-rez vinyl specific ADC with cartridge gain, loading, and RIAA equalization done in the digital domain? With the DSP horsepower presently available, this could sound quite good and be priced fairly reasonably.

ctviggen · « **Reply #3 on:** 3 Feb 2005, 08:05 pm »

I have to disagree about the price -- I think the thing that's expensive is the software. The hardware is some cost, but the cost of developing software far outpaces the cost of developing the hardware. When I was a practicing EE, I spent countless hours designing software for a system that used two DSPs, and these were tiny, no-function things at the time. The DEQX far outclasses what I did. They have to recoup their development costs somehow.

OBF · « **Reply #4 on:** 3 Feb 2005, 08:25 pm »

Quote from: Russtafarian

Interesting thoughts by Srajan regarding interfacing analog sources.

That was one part of the article I didn't understand. The DEQX can take a digital source but my understanding is that the 'P' version converts that to analog before applying analog volume control. In this case I don't see how it would be any different than feeding it an analog source except that the TT does not have the DAC stage in front of it so it should be even better.

I guess he's referring to someone who would use the non 'P' version and output 4 digital channels into 2 outboard stereo DACs and a multichannel preamp? I was confused by that point as I thought the standard configuration is to feed the DEQX an analog singal downstream from your DAC and preamp (assuming you're not using the 'P' version with built in DACs and volume).

Well it's not like I'm an expert so maybe there's other ways people are using them, but his description sounded more like a TACT than a DEQX. Is it normal to use the DEQX between your transport and DAC?

R_burke · « **Reply #5 on:** 3 Feb 2005, 08:33 pm »

Quote from: ctviggen

I have to disagree about the price -- I think the thing that's expensive is the software. The hardware is some cost, but the cost of developing software far outpaces the cost of developing the hardware. When I was a practicing EE, I spent countless hours designing software for a system that used two DSPs, and these were tiny, no-function things at the time. The DEQX far outclasses what I did. They have to recoup their development costs somehow.

Assuming a man-year to develop the s/w. The s/w equates to ~$375K at a fully loaded cost of ~$200/hr. If you assume that the h/w is ~$800 of the $3,000 price it would take somewhere around 150 to 200 units sold to break even. I can't imagine a company that wouldn't assume around 100 units a year of sales, so in 2 years we should see a significant price drop - somehow I doubt it, but I can hope

mac · « **Reply #6 on:** 3 Feb 2005, 08:38 pm »

Quote from: Russtafarian

I'm not ready to ditch my modest vinyl rig and record collection.

Neither am I. My turntable sounds sweet feeding the DEQX PDC.

Don't see any reason why RIAA EQ couldn't be handled in the digital domain. Maybe DEQX is listening?

ss397 · « **Reply #7 on:** 3 Feb 2005, 10:23 pm »

the deqx or its equivalent could become another required piece of gear in the rig, just like the cd player. deqx reduces a 20k plus speaker package down to about 2k worth of drivers and nice cabinets so anyone can play, i'm mentally planning my system now.

jackman · « **Reply #8 on:** 3 Feb 2005, 11:04 pm »

I have heard Bamberglabs.com System 5 with Behringer DCX2496 and subwoofers, powered by 6 channels of Theta amps in Phil Bamberg's system and it sounded amazing. Phil used a mic-preamp and Soundeasy to tweek the system for the room and it was probably the best sounding (yet modest) system I have ever heard.

The DCX comes pre-programmed (by Bamberglab) for the speakers and now sells for well under $300. The mic/preamp setup along with Soundeasy are available from BESL for under $500. I know the DCX was made for the Pro market but the sound I heard was very natural and, as I stated earlier, amazing. I liked it so much I purchased a full system (no mic-preamp or Soundeasy yet, that's the next step).

Here is my stupid question, and I'm not trying to be funny. What would the DEQ allow me to do that I can't do with the DCX?

JeffB · « **Reply #9 on:** 3 Feb 2005, 11:20 pm »

What would the DEQ allow me to do that I can't do with the DCX?

Well straight from the Behringer web site,
"The ULTRADRIVE PRO DCX2496 is the ultimate digital loudspeaker management system with an unheard-of audio quality and feature list."

Don't you just love marketing departments.

dwk · « **Reply #10 on:** 3 Feb 2005, 11:47 pm »

Quote from: R_burke

Assuming a man-year to develop the s/w. The s/w equates to ~$375K at a fully loaded cost of ~$200/hr.

One!??!? You can't be serious. How long has Perpetual Tech been promising their Speaker correction module? That is a much easier piece of software than the DEQX stuff, and they've been working on that for years.

I'd say 10 man-years at an opening bid, when you consider that they've gone through different revs of hardware, probably different control architecture prototypes etc. I'm a professional software guy, though DSP is just an interest/hobby, but this type of stuff takes far longer than you probably imagine.

Quote from: jackman

Here is my stupid question, and I'm not trying to be funny. What would the DEQ allow me to do that I can't do with the DCX?

See the other discussions aroound. The short answer is
a) digital output
b)FIR filters rather than IIR filters

Having spent the better part of the last, ummm, maybe 8years tinkering with various ways to do the FIR filter thing with varying degrees of success, the price drop on the DCX2496 prompted me to bite on one. I guess I'll find out how good a substitute it is.

budyog · « **Reply #11 on:** 3 Feb 2005, 11:52 pm »

These machines sound like a new types of equilizers? There has always been ways to adjust the sound to the room. They have just gotten more sofisicated. I do believe that there are situations where they are needed and seem to be a must for subwoofers but certainly not a must for everyone.

mac · « **Reply #12 on:** 4 Feb 2005, 02:30 am »

Quote from: JeffB

What would the DEQ allow me to do that I can't do with the DCX?

I owned one prior to buying a DEQX PDC. There ain't no comparison technically or sonically.

If you want to know the technical details why, I suggest you read the white paper off the DEQX website.

R_burke · « **Reply #13 on:** 4 Feb 2005, 03:45 am »

"I'm a professional software guy, though DSP is just an interest/hobby, but this type of stuff takes far longer than you probably imagine. "

I'm just a Program Manager for a defense firm and if my s/w engineers took more than a man year to solve their DSP issues I'd find new ones. I admit that I'm not familiar with the specifics of this type of DSP, but I am very familiar with the concepts and efforts involved with solving DSP issues and how much coding is involved. Perhaps this type of DSP is more complicated than the DSP involved in Electronic Intelligence (ELINT) and Signal Intelligence (SIGINT) applications, or perhaps I'm used to working with a different level of s/w engineers.

Occam · « **Reply #14 on:** 4 Feb 2005, 07:41 pm »

Quote from: jackman

... What would the DEQ allow me to do that I can't do with the DCX?

Short answer - Nothing, unless you purchase the extra cost option allowing you 6 channels of digital output on the DEQX, if you don't want to use their internal DtoA converters.

Long answer - Iff you add the aditional question "and how well?"

1. I've not heard the DEQX, but I assume the A/D and D/A converters as well as the ic opamps and powrersupplies used on the analog sections are better implemented. The Berhinger uses the JRC4580 opamp in its analog circuitry, and although 'competent', is certainly not 'audiophile approved".
I'm sure the DEQX analog circuitry is substantially better than the 1/10th the price Behringer, but all things are relative -
http://www.6moons.com/audioreviews/overkill/encore.html

"Alas, the same does not hold true for its power supply, analog filter and output stage when inserted into a ne-plus- ultra chain. The DEQX's brilliant designer -- whom I handed Edgar Kramer's Blue Moon Award at the CES for his stock unit -- concurs without sighs. He's already considering an audiophile "cost-no-object" iteration for extreme applications like Overkill Audio's. That's why Derek Wilson bypasses the DEQX's internal DAC with his own external 4-channel unit with outboard power supply. That's why he left me with two DEQXs, one stock, one Overkill's modified version with massively beefed-up power supply. Depending on whom you ask, the power supply in active audio electronics accounts for 70-85% of the sound...."

2. The DEQX's FIR dsp filters are arguably preferable to those IIR filters used in the DCX2496 for the purpose of speaker correction. Above around 400hz, the phase coherence offered by a FIR filter can be audible.

In the realm of room correction, which best takes place in the below 300Hz, the phase coherence offered by a FIR filter is, to the best of my knowledge, not audible. The parametric equilization(s) used for room correction whether it be analog (Rives), FIR (DEQX) or IIR (Behringer) is not an issue of theoretical, aesthetic, or marketspeak 'better', but dependant on the quality and care of implementation.

FWIW

dwk · « **Reply #15 on:** 4 Feb 2005, 08:00 pm »

Quote from: Occam

2. The DEQX's FIR dsp filters are arguably preferable to those IIR filters used in the DCX2496 for the purpose of speaker correction. Above around 400hz, the phase coherence offered by a FIR filter can be audible.

In the realm of room correction, which best takes place in the below 300Hz, the phase coherence offered by a FIR filter is, to the best of my knowledge, not audible. The parametric equilization(s) used for room correction whether it be analog (Rives), FIR (DEQX) or IIR (Behringer) is not an issue of theoretical, aesthetic, or marketspeak 'better', but dependant on the quality and care of implementation.

Sorry, but I think you have it exactly backwards. The benefit of FIR over IIR filters in the mid/treble range is mostly just lobing control and stop-band steepness. There *may* be slight audibility directly related to phase in some cases on some material, etc, but it's not a huge impact.

However, room correction is inherently and unavoidably a non-minimum-phase problem, and any minimimum-phase (ie IIR) correction is at best a partial approach. The types of things that can be done in terms of correction using long FIR filters are so far advanced compared to a simple parametric notch/peak as to make them different things altogether.

In large rooms the benefit might not be worth the considerable increase in complexity, but in smallish rooms where modes are more problematic, it can make a very big difference.

mac · « **Reply #16 on:** 4 Feb 2005, 08:08 pm »

Quote from: Occam

In the realm of room correction, which best takes place in the below 300Hz, the phase coherence offered by a FIR filter is, to the best of my knowledge, not audible.

You might be surprised.

I assume you're familiar with Linkwitz challenge.

Here is what I would like you to do:

Build the circuit. Test it for flat frequency response and equal gain in all three sections. Trim component values if necessary.

For listening via a loudspeaker insert the circuit ahead of your power amplifier. The speaker preferably has no crossover near the 100 Hz region, or is a 2-way with good bass extension, to reduce the possibility that its own phase distortion masks the contribution from the test circuit, or that it pushes the total phase distortion above the threshold of detection. A full-range Quad ESL might be a good speaker candidate.
Quality headphones are probably even more accurate transducers, though observations derived from them may nor translate directly to a loudspeaker in a reverberant environment. Insert the circuit ahead of your headphone amplifier.
Listen in mono.

Listen to your selection of test signals or program material and record any audible differences or lack thereof as you switch between the sections. Record all test conditions so that others can duplicate them, if possible. Note the playback level. Phase distortion may increase the crest factor of the waveform (B above) and lead to non-linear distortion in the speaker and/or the ear producing a change in sound character.

E-mail your observations to me. I will give a summary of the various findings on this page.
I would really appreciate your participation in this investigation. It could help to settle one more issue in knowing what is audible. I am not trying to prove that all allpass crossovers sound the same, they do not in practice, and there are good reasons for it. I merely want greater certainty whether phase distortion is a contributor and, I think, so would you.

Davey was kind enough to provide me with a test CD and I took the challenge. What surprised me was how easily I differentiate between the altered and unaltered music. If you're interested in taking the challenge yourself I'd be happy to send you the CD.

Cheers,
mac.

dwk · « **Reply #17 on:** 4 Feb 2005, 08:16 pm »

Quote from: R_burke

I'm just a Program Manager for a defense firm and if my s/w engineers took more than a man year to solve their DSP issues I'd find new ones.

Well, if you're a Program Manager you should know very well that the coding time is frequently the least of the 'costs' in terms of engineering time. My numbers (which I suspect are low) were assuming that we're starting from concept stage, and thus include prototyping, core DSP routines, microcontroller and PC software interfaces, multiple revs based on different hardware prototypes and undoubtedly many many many listening sessions in a variety of different environments to tune/tweak the algorithms, and the time to make it robust enough to be an idiot-proof consumer product.

Coding a DSP routine in response to a verified and documented set of requirments is indeed a relatively straightforward task. Engineering a commercial product that breaks at least some level of new ground from scratch is not.

Quote

or perhaps I'm used to working with a different level of s/w engineers

great - nice shot. With your management ability and their DSP and coding ability, I expect to see your version out in what - about a year from now?

Marbles · « **Reply #18 on:** 4 Feb 2005, 08:21 pm »

It's OK to argue, but don't screw up this thread with personal attacks.....

ctviggen · « **Reply #19 on:** 4 Feb 2005, 08:34 pm »

To me, the amount of processing and types of analysis required for room correction (and remember that the DEXQ can also do speaker correction) is large. I have an MSEE in signal processing and I couldn't tell you how to do it. So, not only do you have to have someone who can code, but you also have to have someone at least with an MSEE and more likely with a PhD. It's one thing to design a notch filter to get rid of a hump; it's another to get flat response (have you seen Zybar's graphs for the TACT?) from 20Hz to 20kHz. And then there's getting several DSPs to talk to each other, providing a user interface, remotes, menus, etc.