Digital Room Correction

0 Members and 1 Guest are viewing this topic. Read 6910 times.

FredT300B

  • Full Member
  • Posts: 542
Digital Room Correction
« on: 12 Sep 2007, 11:17 am »
I'm under the impression that a significant number of line array builders and buyers are using digital room correction and active crossovers. I'd like to hear from some line array builders and owners who are using this technology. What equipment are you using, what are your impressions, and more specifically, have you done any comparisons of your line arrays with and without the room correction feature?

bwaslo

  • Jr. Member
  • Posts: 60
    • Liberty Instruments
Re: Digital Room Correction
« Reply #1 on: 13 Sep 2007, 04:16 am »
I'm not exactly using a line array, but similar -- a speaker that uses a long BG planar magnetic quasi-ribbon.  Line source, if not line array.

But anyway, I'm using a Behringer FBQ2496 as a digital room correction box (adjusted manually, test gear is needed to do it with that, really).  It frankly makes all the difference in the world.  The high end is a little too soft without it (I listen close, so the high end rolls off a little due to that, from what it would be othewise). 

But the biggest improvements from correction is in the bass (not covered by the lines in my case) and in the sub-woofer range.  The improvements in the bass aren't particular to the system being built around line sources, it is just the facts of life about room acoustics and room modes.  I'm of the (obnoxiously opinionated) opinion that anyone with ANY speaker system who doesn't do correction in the bass is either extremely lucky (as likely as winning the lottery!) or else just kidding himself that he doesn't need it.

ekovalsky

Re: Digital Room Correction
« Reply #2 on: 13 Sep 2007, 05:46 am »
I have the large Nola speakers and quad amp them with TacT equipment in lieu of using the factory passive crossovers and the ancient Dahlquist line level subwoofer crossover.  Each amp receives a full range 24bit, 96kHz digital stream over AES/EBU or S/PDIF.  Signal processing programmed in each amplifier implements crossover, room correction, equalization, and delay.  Finally an amplified analog signal is generated with using direct PCM to PWM conversion.   I guess you could say this is true digital amplification, as opposed to the ubiquitous "class D" designs that accept an analog input.  One can certainly argue the drawbacks and merits of each.  Personally I was thrilled to remove every analog interconnect from my system.  Modifications were made to the all the TacT gear (four S2150 amps and one RCS 2.2X) to minimize jitter and other digital artefacts.  Also the MOSFETs were replaced with superior devices. 

The speakers are a bit unusual in that they feature a short midrange and treble line array (six unspecified Alnico midranges in dipole, nine Raven R1 ribbon tweeters) flanked by an "expanded array" of Seas Excel woofers in individual sealed boxes.  Subwoofer towers have four unspecified 12" woofers per side, each in its own subenclosure and individually ported via front baffle.  You can imagine how difficult, if not impossible it would be to achieve proper time alignment in such a setup using the factory supplied crossvers.  Numerous crossover upgrades have been offered at very high prices by the manufacturer.  I have enjoyed ignoring these knowing I have a far more flexible solution that I can optimize to my room!

While my setup more or less prevents comparison with an without room correction, I will say that proper time alignment of the arrays makes a huge difference in the perceived sound quality.  Line arrays are very sensitive to toe in, as slight changes will significantly alter the time alignment of the midrange and tweeter sections.  I suspect this is why they are often recommended to be set up with no toe in, in an equilateral triangle with the listening position.  One manufacturer (Dali) supplies a floor template to get the toe in exactly right.  With DSP I can set toe in for best response at the listening position then use the adjustable delay to achieve correct time alignment.

Anyway I am a big fan of line arrays, at their best they can more convincingly reproduce live music than any other type of loudspeaker system I have encountered.  I have heard the flagship McIntosh system, the XRT2K, and it is truly phenomenal.  A scaled down version XRT1K is coming out soon at a much more accessible price.  Hopefully one of these days I'll have the opportunity to hear the LS-6/9 and Selah arrays which may be very nearly as good at a more real world price point.


ctviggen

  • Full Member
  • Posts: 5246
Re: Digital Room Correction
« Reply #3 on: 13 Sep 2007, 10:36 am »
Exactly what does time alignment mean in terms of a dipole?  How is it possible to have time alignment when half the signal bounces off the rear/side walls and half has hit your ears well before the bouncing half gets there?  One could possibly time align the front wave from the front drivers, but this wouldn't do anything for rear wave.  Or does time alignment solely consider the front wave? 

Also, how does time alignment work with actual data analysis?  It would seem that to time align a speaker, you'd have to take data using a short window.  However, using a short window means that you won't get reflections and therefore won't deal with the room interactions.  Perhaps that's not a bad thing, but I'm wondering how the TACT actually takes this data.

Also, in terms of a line array, I thought a major benefit was the supposed lack of interaction with the room, at least based on fewer reflections.  Granted, this won't affect the lower octaves, which are primarily room-mode based, but the upper octaves should theoretically need less trimming by a frequency-based room correction system.  Or is that not true? 

ctviggen

  • Full Member
  • Posts: 5246
Re: Digital Room Correction
« Reply #4 on: 13 Sep 2007, 11:07 am »
In terms of line arrays, I find time alignment even more troubling.  Assume you have an 8 foot array of tweeters and your ears are aligned (for lack of a better word) with the middle of the array.  There is going to be a shorter distance between your ears and the tweeter in the middle of the array as compared to the distance between your ears and the tweeter at the end of the array.  This means there is a time misalignment, and unless each driver has time compensation, this time misalignment is inherent to the design of the speaker and no amount of digital correction can correct this.  This is also true for overlapping frequencies produced by tweeters and midranges.  Now, perhaps one could one tweeter and one midrange and "align" them so that certain non-overlapping frequencies would arrive at your ears at the same time, but again this is only for a single set of drivers.  All other drivers relative to this single set would not and cannot (barring time compensation on a per driver basis) be time aligned. 

So, I can't see how time alignment is valid for a line array.  Now, digital frequency correction would still work. 

FredT300B

  • Full Member
  • Posts: 542
Re: Digital Room Correction
« Reply #5 on: 13 Sep 2007, 12:30 pm »
In terms of line arrays, I find time alignment even more troubling.  Assume you have an 8 foot array of tweeters and your ears are aligned (for lack of a better word) with the middle of the array.  There is going to be a shorter distance between your ears and the tweeter in the middle of the array as compared to the distance between your ears and the tweeter at the end of the array...

I believe Danny addressed this issue with some relevant data in another thread. Here's an excerpt from his explanation:

"...the time differential of drivers near the top of the array verses drivers in the middle of the array (where your ear is) is minuscule. From a typical listening distance this 1" to 2" differential is in the order of about a 1/10th of a millisecond or so...

...The only place this will have an effect is in the upper harmonic ranges where wavelengths are very short. Even then as you get a little comb filtering effect causing a slight peak and dip it can easily reversed with a slight variation in listening height. You actually hear an average level due to room related effects. The limited dispersion of some taller tweeters limits interaction in the top octave any way.

A simple side wall reflection can be many times greater with delay of 4 to 6 milliseconds. How much does this smear the sound? Some say it needs to be at least 8 milliseconds of a delay to even hear it as a delay. Some say a little less. Whatever it is it really doesn't matter for this example. The delay of the line source drivers is not even close to that".

AdamM

  • Full Member
  • Posts: 313
    • Robotbreeder.com
Re: Digital Room Correction
« Reply #6 on: 13 Sep 2007, 12:34 pm »
I guess that would be maybe another reason why line arrays like to be in a bigger room?  Moving back would further reduce the distance differential between the middle and end drivers.

bwaslo

  • Jr. Member
  • Posts: 60
    • Liberty Instruments
Re: Digital Room Correction
« Reply #7 on: 13 Sep 2007, 02:59 pm »
I think if you look at it like it is a filter, which is what it equates to, it isn't such a problem.  The concern seems to be that the sound from the drivers aren't all arriving at the same time.  But if the drivers are small enough and close enough together, or are a continuous line source (like a ribbon), then the arrival is effectively continuous, not a series of discrete wavefronts.  In other words, it behaves like a filter that has a high frequency rolloff  -- same as you would get if you combined delayed signals within a FIR filter.  Nothing outrageously ugly about it, and equalizable.  Sure, lack of complete flatness in the frequency response isn't ideal, but the directional characteristic of the array improves the in-room flatness a LOT more than the close-up rolloff hurts it.  Just look at the unsmoothed frequency response of a line driver within a few meters, as compared to from a point source in a room - it will be much better controlled.

The effect does change, though, as you get farther away (because the path differential decreases with distance from the line).  Close up, the highs roll off some (a few dB, like I mentioned my BG lines doing).  Farther away, it flattens out. 

But listening closer has the advantage that you hear more direct sound and less that is just ricocheting off of everything in the room.  The combined, unequally delayed responses of a line are delayed very benignly, when compared to the stronger extremely delayed combined responses you hear (from all the reflections in the room) with a point source speaker (and which could only be ideally corrected digitally for one single listening position in the room).  If the goal is to get more of a single arrival distance of wavefronts, lines or planars are the answer, not the problem!

zobsky

  • Jr. Member
  • Posts: 139
  • Fringe Lunatic - Dallas, Tx
    • My Audio Blog
Re: Digital Room Correction
« Reply #8 on: 18 Sep 2007, 10:51 pm »
I'm under the impression that a significant number of line array builders and buyers are using digital room correction and active crossovers. I'd like to hear from some line array builders and owners who are using this technology. What equipment are you using, what are your impressions, and more specifically, have you done any comparisons of your line arrays with and without the room correction feature?

If not overused, the often despiesed  bass and treble controls can help to bring the sound back in line

Rick Craig

  • Industry Participant
  • Posts: 3680
  • Selah Audio
    • http://www.selahaudio.com
Re: Digital Room Correction
« Reply #9 on: 19 Sep 2007, 03:10 am »
I'm under the impression that a significant number of line array builders and buyers are using digital room correction and active crossovers. I'd like to hear from some line array builders and owners who are using this technology. What equipment are you using, what are your impressions, and more specifically, have you done any comparisons of your line arrays with and without the room correction feature?

I use the DEQX and I have friends and customers that run their arrays with the DEQX, Behringer, TacT, and conventional active crossovers. As long as the DSP units are correctly implemented they do a great job as active crossovers and for applying equalization. For room correction I don't find them to be as beneficial with arrays because a line of woofers can interface better with a room's typical standing waves versus a normal point source speaker. The phase correction can be helpful with the speaker's horizontal coverage; however in some cases I found it to be just "different" in terms of imaging - not necessarily better.

If you have need for low frequency room correction or equalization DSP is great and very flexible - much better than doing it with passive components. Passive components will introduce problems in the bottom octave with insertion loss and phase angles that make it harder for your amp to control the woofers.

I do know of a few people using standard active crossovers and they are happy with the results; however, I think they are too limited to really get the best performance from an array. Most of these units only have symmetrical slopes and don't give you enough control over the shaping of the filters.

ekovalsky

Re: Digital Room Correction
« Reply #10 on: 19 Sep 2007, 08:44 am »
In my setup, only the midrange drivers operate in dipole.  The woofers are sealed box, and the tweeters have their own individually sealed enclosures even though thery are not mounted in a box but rather share the baffle with the midrange units.

I time align using the first arrival, i.e. the front wave.  The rear wave is attenuated (I use absorption panels behind the speakers) and obviously delayed. The process is rather tedius with my TacT/TACS setup.  Besides the differences in physical distance between driver and listener, DSP minimum phase crossovers have lag times that vary with multiple factors, the largest of which is high pass versus low pass operation.  Generally low pass filters produce the greatest lag and this increases as the crossover frequency goes down.  The relationship with slope is more complex and not always predictable.  Anyway my sub towers are the furthest away and have the longest crossover lag so they have no added delay.  Then I measure low frequency pulses sent to the subs and then the woofers of the main towers, the difference in millseconds will be the added delay to the main woofers.  Next I send a mid frequency pulse to the woofers and midranges.  Difference is summed to the previously measured woofer delay and set as the added delay to the midrange section.  Finally, a high frequency pulse is sent to the midranges and tweeters.  The difference is summed to the midrange delay and programmed as the added delay to the tweeters.

Complicating this somehwat is the same test pulses appear different between different driver groups, which is an effect of the applied crossover.  It would be nice if you could just use the impulse peaks to determine the time differences but this has produced rather poor results in my experience.   Instead of using the peak, which is generally what automatic delay calculations are based on, I set my reference points manually on the impulse rise at 10% of peak amplitude.  Worth mentioning is that precision at higher frequencies (i.e. mid-tweeter transition) is excellent but decrases at low frequencies.  This is a function of how narrow or wide the captured waform is. 

Your second point, about differing distances between the ear and the individual drivers in the array, is a good observation.  I have thought about it too as have some manufacturers that make arrays concave to the listening like Grphyon.  But I ultimately decided that there is minimal if any real world manifestation of this problem.  The problem primarily affects tweeters because of their short wavelengths and potential for comb filtering.  But when listening in the near field the tweeter array sends out a cylindrical waveform, meaning that only the tweeter at ear level is majorly contributing to the perceived sound.   The woofers are somewhat more spread out, but the long bass wavelengths prevent any detectable smearing.

By the way a key advantage of dipoles is the (out of phase) front and back sound waves cancel each other at the sides.  This cuts down on side wall reflections.

JonFo

  • Jr. Member
  • Posts: 6
    • My System
Re: Digital Room Correction
« Reply #11 on: 27 Oct 2007, 10:35 pm »
Hi There, first post, so be gentle ;-)

I use the DBX DriveRack 260’s in my system. These speaker processors are the best tool I’ve yet found for tuning a speaker system (across speaker elements) and tuning a set of speakers to each other and the room they are in.
With these, I can pick any number of crossover styles (Linkwitz-riley, Butterworth, and Bessel) four different slopes, full phase angle adjustment in 1 degree increments. I can have low-pass be one slope /style and combine it with a different HP slope and style. I can also add delay at either inputs or outputs (each individually). There are multiple parametric EQ bands on each input, and on each output individually. Very, very flexible.

Using ETF measurement software and my calibrated mic, This allows full adjustment of time-alignment within the speaker elements, and for gain and frequency balance adjustments to get some of the best sound I’ve ever heard from a speaker system.

I go into some detail of how I deploy these DriveRacks in my write-up of my Center Channel ESL hybrid line-array build http://www.martinloganowners.com/~tdacquis/forum/showthread.php?t=2018.

My take is there is no comparing what a speaker processor can do vs any passive components.


Rick Craig

  • Industry Participant
  • Posts: 3680
  • Selah Audio
    • http://www.selahaudio.com
Re: Digital Room Correction
« Reply #12 on: 29 Oct 2007, 01:10 am »
Hi There, first post, so be gentle ;-)

I use the DBX DriveRack 260’s in my system. These speaker processors are the best tool I’ve yet found for tuning a speaker system (across speaker elements) and tuning a set of speakers to each other and the room they are in.
With these, I can pick any number of crossover styles (Linkwitz-riley, Butterworth, and Bessel) four different slopes, full phase angle adjustment in 1 degree increments. I can have low-pass be one slope /style and combine it with a different HP slope and style. I can also add delay at either inputs or outputs (each individually). There are multiple parametric EQ bands on each input, and on each output individually. Very, very flexible.

Using ETF measurement software and my calibrated mic, This allows full adjustment of time-alignment within the speaker elements, and for gain and frequency balance adjustments to get some of the best sound I’ve ever heard from a speaker system.

I go into some detail of how I deploy these DriveRacks in my write-up of my Center Channel ESL hybrid line-array build http://www.martinloganowners.com/~tdacquis/forum/showthread.php?t=2018.

My take is there is no comparing what a speaker processor can do vs any passive components.



Nice work! I would be interested to see how the Extremis woofers measure in the nearfield (1/4" from one of the drivers) from 20-200hz. Based on my measurements of them I think a larger enclosure would give you a smoother response.