Velodyne Digital Drive Plus 18 (DD18+) Subwoofer Review

[mike c]

Piling on isn't cool either.

did you want me to just ban him?

i'm sorry if i want to clean up AH of rude know-it-alls. (go ahead and read through his previous posts here on AH)

even if he really knew-it-all (which he really doesn't), none of his points get through because of how he pushes them on people.

 

[Sputter]

did you want me to just ban him?

i'm sorry if i want to clean up AH of rude know-it-alls. (go ahead and read through his previous posts here on AH)

even if he really knew-it-all (which he really doesn't), none of his points get through because of how he pushes them on people.

Lol, of course not. Matt spoke to him though.

DS worded his post badly. If everyone got banned for a strongly worded post we'd all be banned.

 

[mike c]

Lol, of course not. Matt spoke to him though.

DS worded his post badly. If everyone got banned for a strongly worded post we'd all be banned.

yeah we would. but this really isn't his first time to do so, it seems he speaks like everyone were idiots and he was the almighty.

i guess what i'm trying to say is ... that was his last warning.

 

[gene]

I guess serious thought is "misinformation" to the guy who wrote that "toppled-MTM's don't suck so bad despite what people who can hear think, and how closely that tracks their abysmal measured performance, really, I promise" article...



Interesting claim. You do realize that a 1-octave measurement would from 5-200 Hz would consist of all of 6 data points (5, 10, 20, 40, 80, 160), right?



While that is true, you've provided no evidence to support your position. Saying that a firm is "very vague about this" is not information.

I have compared my SMS-1's output to 1/3-octave smoothed measurements using FuzzMeasure, taken with the same mic that has not been moved. My conclusion from those comparisons is that the system is accurate enough to make pulling out the mic preamp, etc. not worth it. The SMS-1 (assuming the MIC-5 is also used) is easier to use. (Incidentally, these days I only use my SMS-1 for measurement. The processing side of the device is just too limited; it cannot even individually adjust level and phase for its three outputs. So I measure with the SMS-1 fed by a MIC-5 with a configuration to properly take a spatial average at the listening position*, adjust the levels/delays/PEQ with a miniDSP, and then put the SMS-1 and MIC-5 back in the closet)

*A representative configuration for a proper spatial average at the listening position looks like this, minus the kittens:

Notice that I am taking a spatial average around one listening position, not trying to use flawed single-point measurements to average multiple spots.



That's all quite silly. For one thing, if it were true, why would Velodyne offer the MIC-5 kit? They're not exactly targeting consumers with that, now are they?

A more reasonable inference is the lazy person will throw on the auto-EQ, and the more interested person will, using the measurement system Velodyne provided with the sub, manually adjust the parametric EQ.



I believe Dr. Toole has only written one book, Sound Reproduction. At least, that is the only book that comes up in an Amazon search for "Floyd Toole."

Had you read my post with a modicum of care, you would have noted that I provided cites from that very book.



If you want to chase phantom problems, be my guest. But while you waste time to no audible benefit, I'll be enjoying the music thank you very much...

You like to quote Dr. Floyd Toole's book but did you actually read it? I doubt you did else you wouldn't be making the argument that measuring with 1/3rd octave resolution is "enough". I need not reinvent the wheel when masters like Floyd Toole have already done so. If you really want to know why its not enough, read his book instead of just referencing it.

Spatial averaging without applying weighing of importance to the measurements is pretty useless but it does seem impressive to set mics up around the room doesn't it?

I prefer to measure the primary 2-3 seats and correct for those. I then spot check the other seats to make sure I didn't cause any harm.

i have detailed this quite extensively in the following article which incidentally was peer reviewed by Dr. Floyd Toole:

Home Theater Multiple Subwoofer Set-Up & Calibration Guide — Reviews and News from Audioholics

As for MTM center channels being flawed, I actually did write an article about how they can work just fine if they are properly designed and the listeners sit no more than 20deg off axis. I also provide measurements to prove it too:

Vertical vs Horizontal Center Channel Speaker Designs

You display a rude and arrogant position and you are on your last leg of my patience on this forum. Consider this your only warning.

 

[DS-21]

On a more serious note which DS-21 brings up, I have never used an SMS-1 so I can't comment on it but the thought that perhaps it is based on the same software and performs the same did enter my mind. Again I have never used one though.

Josh, I'm curious: did you ever hook the DD18+ up to a TV through the composite or S-video output - I just looked it up to see if it still has them, and it does - to see if the graph is similarly offset?

I ask solely because it would be interesting to know if it displays with the same offset on on a TV as it does on a computer screen. When my next new amp for my second sub shows up,* assuming the amp actually works for me (which it will if it's not DOA, as it's an updated version of the amp I currently use for the primary sub) I'll take pictures of the SMS-1 screenshots and compare them to what I get hooking my new Art Dual Pre - new MacBook doesn't have FireWire, which I didn't realize until I got it home, so I needed a new mic preamp... - up to the MIC-5 and taking steady-state measurements of that physical spatial average in FuzzMeasure, as well as taking five measurements in the same "bubble" with a single mic hooked up to the Dual Pre and averaging them in FuzzMeasure.

*Not really germane to this thread, but let's just say that I'm going to recommend everyone who uses the miniDSP to set levels/delay/EQ for a DIY multisub step get the balanced version, because the low output of the unbalanced unit seriously limits amp choice, and it's also IMO unacceptably noisy when one uses a line driver to compensate for the low output. (Amps with lower input sensitivity, such as the Dayton rack sub amps and I expect most plate amps in commercial subs, work just fine with the unbalanced miniDSP; so would, I expect, Emotiva amps, because they seem to design in low input sensitivity to get people to think they sound better than other amps.) Also, it seems that some of the modern lightweight Class D pro amps generate a lot of spurious HF noise when plugged into a power strip that turns outlets on/off based on the state of the AVR, though not when plugged directly into the wall.

Having measured my subs via SMS and then measuring with REW there was a wide difference between the two graphs, enough for me to use REW for measuring and SMS to apply filters based on REW.
***
I would say that some rooms might be tougher to get a flat response, therefore the difference between SMS and REW would increase. Conversely, the easier room could be a closer match to what SMS and REW 'sees'.

I suspect that you're onto something with those last two sentences. A room-multisub system that has fairly low variance in true response will lead to smaller variances in measurements taken with different measurement techniques, while a room-sub system with higher true variance would lead to a larger variance in measurements taken with different techniques. That could lead to differing viewpoints.

But just to confirm, when you talk about wide differences between REW results and SMS-1 results, you're talking primarily about differences that appear to be due to smoothing, right? That is to say, it didn't look to you as if the SMS-1 display was offset compared to the REW graph, did it?

Also, consider that in a spatially-averaged measurement, there is a degree of "smoothing" (really convergence to the true response) from the averaging of multiple measurements anyway. That helps take away a lot of the phantom grass that people who don't take spatially-averaged measurements but have measurement systems capable of displaying highly precise (if ultimately inaccurate) single-point measurements and powerful parametric EQ devices often waste time "correcting."

I think what DS may be saying (it's fun putting words into people's mouths), is that 1/3 Octave is enough from a measurement perspective to adjust for what matters.

Just to clarify, 1/3 octave smoothing of a steady-state measurement. I wouldn't - and don't - play successive discrete 1/3 octave test tones, measure with an SPL meter, and plot the result in a spreadsheet.

Would be interesting to know whether the SMS-1 is in fact measuring at a higher resolution and presenting the FR at 1/3 Octave, or measuring at 1/3 octave as well.

Given that the test tone is a sweep rather than discrete tones, I find it doubtful that it's just sampling discrete 1/3-octave tones and plotting a curve that connects them. I have no inside knowledge one way or the other, though.

You like to quote Dr. Floyd Toole's book but did you actually read it?

Obviously, yes.

Spatial averaging without applying weighing of importance to the measurements is pretty useless but it does seem impressive to set mics up around the room doesn't it?

Gene, this time I even gave you a picture! As should be clear from that picture, I do not advocate "set[ting] mics up around the room" [emph. added]

Rather, what one should do is take a proper spatial average puts the mics in a bubble around the intended listening position. See the picture in my previous post for an example of a proper setup to ascertain the true response with much less error at one listening position..

One can of course measure at multiple listening positions. That means taking a spatial average around a "bubble" centered around each alternate listening position.

Furthermore, on what grounds would one weight one measurement over another? They're all an equal part of the spatial average for that listening position. After all, the point of taking multiple measurements is to correct for errors inherent in a single sample. See Geddes and Blind, "The Localized Sound Power Method," 34 JAES 3 (Mar., 1986), at 167.

That said, I asked Dr. Geddes about my approach of taking the spatial average with multiple mics deployed at once and averaged with a mixer, as opposed to using one mic in multiple positions and averaging the resultant graphs in software. His response was as follows:

There is a small preference for doing spatial averages on data in dB, but this is not a critical factor. Theoretically the averaging should be in dB, but studies have shown difference in only fairly extreme cases. When averaging linear signals, the phases can be such that nulls can appear when no real null exists. SO if you do average the mix signals with a mixer the nulls may be wrong, but everywhere else the data is fine. However, software averaging is orders of magnitude cheaper to do, so I would think this to be the best approach.

Next time I do a subwoofer setup - soon, I expect; see supra - I want to compare using the Velodyne MIC-5 (which is, after all, a simple mixer/mic-preamp and 5 mics) to taking five individual measurements. I hope the MIC-5 comes out identical, because it makes the process much quicker. Though my comparison cannot be totally valid because I only a calibration file for one mic. The other five are not calibrated, and even if they were how could one use calibration files when the data are mixed upstream of the measurement program?

Now, I could see, if someone really cared about the sound at multiple positions, constructing an algorithm that combines all of the measurements one takes, weighting the spatial average for each position differently in the summation. That would see to me to require some bespoke program, though perhaps FuzzMeasure or one of the packages could do such a thing. I've not explored it, as I'm not personally that concerned with response all around the room.

i have detailed this quite extensively in the following article which incidentally was peer reviewed by Dr. Floyd Toole:

Home Theater Multiple Subwoofer Set-Up & Calibration Guide — Reviews and News from Audioholics

It's worth noting that your article is inconsistent with Harman's modern approach to real rooms, Sound Field Management, so perhaps an update is in order. The Welti & Devantier work you cite is based on idealized rectangular rooms, not real domestic living rooms with varied openings, wall shapes, large furniture, etc.

SFM employs randomized (as opposed to symmetrical) placement of multiple subwoofers (with no bias against odd numbers) and uses an algorithm to set independent levels, delay/phase, and to some degree parametric EQ for each one. For more on SFM, please see Toole, § 13.3.6, at 230.

As for MTM center channels being flawed, I actually did write an article about how they can work just fine***

...

I apologize for introducing a digression about that...article into this thread. Whatever my thoughts about it, it is entirely out of the scope of this thread, which is Velodyne's excellent-performing DD18+ subwoofer generally, and insofar as my participation is involved, an apparent software bug in either the measurement or the display system of the DD18+ that seems to shift the measurement taken by the device over by 1/2 octave.

Josh Ricci uncovered the bug in his review and discussed it in some detail. But until I mentioned it Josh's work seems to have gone unnoticed amid much kvetching about comparative irrelevancies such as pricing.

Hopefully someone at Velodyne read Josh Ricci's superb review (his approach should be the model for all reviews of loudspeakers, not just subs, albeit with polar maps for larger-bandwidth devices) and is thinking about where in the measurement or display system the bug might reside and how to fix it.

 

[Ricci]

DS-21,

I did not try running the DD18+ EQ system through the TV or a monitor. It was much easier to use my laptop since it was already there set-up with REW and I needed it to control the DD18+ outdoors for testing as well.

I'm not sure of what was happening with the EQ system but that was what jumped out at me immediately was that the major response features appeared to be shifted. As far as I know Velo hasn't commented on it. I do wonder what sort of resolution is actually there in their software.


Sputter,

Do you have any pics of the difference between the SMS-1 and REW? I'm sort of interested in this now.

 

[Sputter]

Sputter,

Do you have any pics of the difference between the SMS-1 and REW? I'm sort of interested in this now.

Sorry Josh, I didn't keep any direct comparisons. (i never thought to)
I do need to do a close mic graph of one of my 15H's, i'll do a comparison then. With Christmas on our tails it'll be a few days before I can.

Jim

 

[gene]

Quote:
Originally Posted by Dr. Earl Geddes
There is a small preference for doing spatial averages on data in dB, but this is not a critical factor. Theoretically the averaging should be in dB, but studies have shown difference in only fairly extreme cases. When averaging linear signals, the phases can be such that nulls can appear when no real null exists. SO if you do average the mix signals with a mixer the nulls may be wrong, but everywhere else the data is fine. However, software averaging is orders of magnitude cheaper to do, so I would think this to be the best approach.

That is very true and hence reaffirms my point about averaging multiple measurements. I do agree that we typically want to correct only low Q bass problems but you still need high resolution data to accurately do this. I make a series of closely spaced measurements and use LMS which can apply weighting to the measurements and even do minimum phase transforms when needed. I don't know of too many other measurement packages that offer math functions so unfortunately I am stuck with the antiquated yet still powerful LMS measurement system.

Next time I do a subwoofer setup - soon, I expect; see supra - I want to compare using the Velodyne MIC-5 (which is, after all, a simple mixer/mic-preamp and 5 mics) to taking five individual measurements. I hope the MIC-5 comes out identical, because it makes the process much quicker. Though my comparison cannot be totally valid because I only a calibration file for one mic. The other five are not calibrated, and even if they were how could one use calibration files when the data are mixed upstream of the measurement program?

Years ago I toyed with the Mic 5 and SMS system and found too much variability between sweeps to make any sort of accurate corrections. Their sweep times are too fast and the resolution is too low so at one moment you can see a null then all the sudden its gone in the next sweep. Longer sweep times would be nice but the Velodyne system currently isn't set up for this. If enough users like yourself email Velodyne and ask them for longer sweeps and higher measurement resolution, they may finally listen. I voice this each time I review an SMS based subwoofer product but they claim I am a minority in a large sea of users that simply don't know what to do with higher resolution measurements. They do have a valid point, most end users aren't skilled enough to do accurate and meaningful compensation.

Now, I could see, if someone really cared about the sound at multiple positions, constructing an algorithm that combines all of the measurements one takes, weighting the spatial average for each position differently in the summation. That would see to me to require some bespoke program, though perhaps FuzzMeasure or one of the packages could do such a thing. I've not explored it, as I'm not personally that concerned with response all around the room.

LMS can do this. You should be concerned for at least 2 seats (yours and your significant other). The worst measuring seat in the house goes to the Mother-in-Law

It's worth noting that your article is inconsistent with Harman's modern approach to real rooms, Sound Field Management, so perhaps an update is in order. The Welti & Devantier work you cite is based on idealized rectangular rooms, not real domestic living rooms with varied openings, wall shapes, large furniture, etc.

Actually Welti and Devantier spent extensive time reading my article and helping me run sims to come up with some of the subwoofer positioning options I listed in the diagrams. I've also spoken to Welti on numerous occasions about this topic and he not only liked my article but he also created a massive simulation analysis for me which I eventually plan on publishing when I can find the time to frame an article around it.

My article gives recommendations for rectangular rooms but most of what I write about combats odd shape room problems too. My room for example is "L" shaped so I have a lot of experience dealing with this first hand.

The goal of any multi sub system should be to get all the subs working as best as they can together and integrated with the main channels. Only at that point should you even consider applying a global EQ to all the subs. This is a last resort and often only minimum EQing is needed if speaker placement, listening seat placement and calibration is done correctly.

SFM employs randomized (as opposed to symmetrical) placement of multiple subwoofers (with no bias against odd numbers) and uses an algorithm to set independent levels, delay/phase, and to some degree parametric EQ for each one. For more on SFM, please see Toole, § 13.3.6, at 230.

I am well aware of SFM but I am not writing an article to promote a particular companies solution. I try to keep things as generic as possible. SFM isn't even commercially available for home theater products. Right now its only offered via JBL Synthesis as far as I know. Incidentally I've heard many Harman systems setup with ARCOS and SFM and haven't been impressed as of yet.

see: http://forums.audioholics.com/forums/loudspeakers/75973-jbl-synthesis-cedia-2011-demo-evaluation.html

Hopefully I will be able to analyze their system in my own listening space one day to get a better picture for how well it works.

I apologize for introducing a digression about that...article into this thread. Whatever my thoughts about it, it is entirely out of the scope of this thread, which is Velodyne's excellent-performing DD18+ subwoofer generally, and insofar as my participation is involved, an apparent software bug in either the measurement or the display system of the DD18+ that seems to shift the measurement taken by the device over by 1/2 octave.

Josh Ricci uncovered the bug in his review and discussed it in some detail. But until I mentioned it Josh's work seems to have gone unnoticed amid much kvetching about comparative irrelevancies such as pricing.

Hopefully someone at Velodyne read Josh Ricci's superb review (his approach should be the model for all reviews of loudspeakers, not just subs, albeit with polar maps for larger-bandwidth devices) and is thinking about where in the measurement or display system the bug might reside and how to fix it.

Yes I still believe its a measurement resolution issue. If I get time, I will do my own measurements on my DD-15+ comparing their legacy display to a TV monitor, vs computer vs my LMS system. I am not surprised by Josh's results but the end goal for all of us is to help Velodyne make their system even better. I will engage Velodyne again on this issue after the holidays.

 

[DS-21]

DS-21,

I did not try running the DD18+ EQ system through the TV or a monitor. It was much easier to use my laptop since it was already there set-up with REW and I needed it to control the DD18+ outdoors for testing as well.

Makes sense. I was just curious.

(The smiley, for the record, was simply because as I was writing it occurred to me that Velo may have dropped the TV display option, so I checked their website to see if they still offered it.)

That is very true and hence reaffirms my point about averaging multiple measurements.

How? Dr. Geddes' point is directly contrary to what you're arguing. The paper he co-authored on sound power measurements (see supra) also does not endorse any weighting of the individual observations. To do so is just bad statistics!

What he's saying is that averaging is better in dB through software than in linear signals through a mixer, not that any one measurement should be weighted over the others. Especially when one's measurements reach into the statistical region (~200Hz), which I believe is proper procedure, and I suspect you do too.

I make a series of closely spaced measurements and use LMS which can apply weighting to the measurements

On what basis do you weight the measurements? No single point can, after all, account for the complex interactions of the ears and the bones of the face and chest in perceiving bass.

Years ago I toyed with the Mic 5 and SMS system and found too much variability between sweeps to make any sort of accurate corrections.

I have years of experience with the combination, and my experience is different from yours. It's quite consistent from sweep to sweep, and even from rep to rep with days between them. I've never seen the disappearing null issue you claim, unless of course I've applied some processing (a delay change or something - I only apply boost at the low end to extend the response, not in the modal region, though that may well be some vestigial audiophool prejudice) that could have an effect on the null.

Their sweep times are too fast and the resolution is too low so at one moment you can see a null then all the sudden its gone in the next sweep.

I do agree with you that longer-duration sweep would be useful. That said, several years ago when I set FuzzMeasure to do the same length and frequency sweep with 1/3 octave smoothing, it essentially confirmed the results from the SMS-1. Since then, I've used the SMS-1 because it's quicker to use with the continuous sweep than computer-based measurements are.

If enough users like yourself email Velodyne and ask them for longer sweeps and higher measurement resolution, they may finally listen.

Last time I spoke to anyone at Velodyne, the paraphrased answer was a fairly terse "we have no plans for an upgrade or an update to the SMS-1." Admittedly, my wishlist for the box is different from yours, but I think that's too bad. A self-contained bass optimization system that includes both a measurement system and level/delay/EQ control control strikes (with some automated capabilities for the marketers to talk up) me as a fundamentally right idea. But I assume the SMS-1 didn't do too well for them. For that matter, I don't know a single other consumer either who purchased a MIC-5 or who's written about using the SMS-1 with multiple mics hooked up to a different mixer...

You should be concerned for at least 2 seats (yours and your significant other). The worst measuring seat in the house goes to the Mother-in-Law

In our flat, I think the kittens probably care more about sound quality than my SO does.

But as it were, the sound is quite consistent across one couch, and the bass is very elevated on the daybed. (Why? Two reasons: it's against a wall, and between my primary and third subwoofers.)

Actually Welti and Devantier spent extensive time reading my article and helping me run sims to come up with some of the subwoofer positioning options I listed in the diagrams.

What the sims don't tell you is that, broadly speaking, placement of any individual sub is far less important that simply having enough pressure sources in the room to smooth out room modes. And the big weakness of Welti and Devantier's work is that, until SFM they focused rather myopically on symmetrical placements. That is to say, they simply didn't test placements that lacked at least one axis of symmetry.

My article gives recommendations for rectangular rooms

Without giving the only really useful advice about subwoofer placement: put them where they will fit.

Really, the notion that one has to fit modal-region pressure sources into some sort of fixed pattern is not only obsolete but also counterproductive if the goal is smooth response. (Below the modal region, placement matters even less, and unless the room is basketball gym-sized the subs sum in-phase basically no matter where they are.)

The goal of any multi sub system should be to get all the subs working as best as they can together and integrated with the main channels.

Perhaps a semantic, and pedantic, nitpick, but I would reverse those. In fact, I don't care at all how they work together without the mains. Modern best practices[/i] focus on integrating one sub with the mains, and then sequentially blending the other subs in.

Only at that point should you even consider applying a global EQ to all the subs. This is a last resort and often only minimum EQing is needed if speaker placement, listening seat placement and calibration is done correctly.

Agreed. The overwhelming majority of the magic comes from getting the relative levels (which is why the "gain-matching" folks are trading away smoothness of response for more output, but I digress) and the phase/delay right. EQ employed should be minimal, and IMO focused on setting the low-end corner via boost or correcting something like a falloff in output due to two subs rolling off sooner than the third.

The best measuring*-and-sounding bass system I've had thus far (a Tannoy B475 and two DIY JBL 2235H-based subs in a room that had three masonry walls and three cardboard-n-spit walls) literally required no EQ.

*The "if it doesn't go down to 3Hz, it's crap" brigade would disagree, because the LF corner of the system was in the mid-20s.

I am well aware of SFM but I am not writing an article to promote a particular companies solution.

When you cite Harman people to the exclusion of others, you're tacitly endorsing Harman's approach.

SFM isn't even commercially available for home theater products. Right now its only offered via JBL Synthesis as far as I know.

One can go through the same iterative process manually. Just look at the PDF I linked to, supra.

Yes I still believe its a measurement resolution issue.

I just don't see how a resolution issue can make, as Josh wrote a few posts above, "the major response features appeared to be shifted." [emph. added]

A software bug in the display rendering seems to be the most logical cause of the behavior Josh observed, followed by perhaps a software bug in the measurements themselves.

I am not surprised by Josh's results

Except for the apparent bug in the measurement display, I'm not either. Velodyne's always made good subs, and it is entirely reasonable to assume that the same power applied to a driver of equal or greater throw in a bigger box will lead to more output with less compression.

I will engage Velodyne again on this issue after the holidays.

It'll be interesting to see what they have to say. Josh's review shows that they have a very solid product on their hands, except for the one bug with the measurement display.

 

[gene]

Quote:
Originally Posted by gene View Post
I make a series of closely spaced measurements and use LMS which can apply weighting to the measurements
On what basis do you weight the measurements? No single point can, after all, account for the complex interactions of the ears and the bones of the face and chest in perceiving bass.

Quote:
Originally Posted by gene View Post
Years ago I toyed with the Mic 5 and SMS system and found too much variability between sweeps to make any sort of accurate corrections.
I have years of experience with the combination, and my experience is different from yours. It's quite consistent from sweep to sweep, and even from rep to rep with days between them. I've never seen the disappearing null issue you claim, unless of course I've applied some processing (a delay change or something - I only apply boost at the low end to extend the response, not in the modal region, though that may well be some vestigial audiophool prejudice) that could have an effect on the null.

Quote:
Originally Posted by gene View Post
Their sweep times are too fast and the resolution is too low so at one moment you can see a null then all the sudden its gone in the next sweep.
I do agree with you that longer-duration sweep would be useful. That said, several years ago when I set FuzzMeasure to do the same length and frequency sweep with 1/3 octave smoothing, it essentially confirmed the results from the SMS-1. Since then, I've used the SMS-1 because it's quicker to use with the continuous sweep than computer-based measurements are.

Quote:
Originally Posted by gene View Post
If enough users like yourself email Velodyne and ask them for longer sweeps and higher measurement resolution, they may finally listen.
Last time I spoke to anyone at Velodyne, the paraphrased answer was a fairly terse "we have no plans for an upgrade or an update to the SMS-1." Admittedly, my wishlist for the box is different from yours, but I think that's too bad. A self-contained bass optimization system that includes both a measurement system and level/delay/EQ control control strikes (with some automated capabilities for the marketers to talk up) me as a fundamentally right idea. But I assume the SMS-1 didn't do too well for them. For that matter, I don't know a single other consumer either who purchased a MIC-5 or who's written about using the SMS-1 with multiple mics hooked up to a different mixer...

Quote:
Originally Posted by gene View Post
You should be concerned for at least 2 seats (yours and your significant other). The worst measuring seat in the house goes to the Mother-in-Law
In our flat, I think the kittens probably care more about sound quality than my SO does.

But as it were, the sound is quite consistent across one couch, and the bass is very elevated on the daybed. (Why? Two reasons: it's against a wall, and between my primary and third subwoofers.)

Quote:
Originally Posted by gene View Post
Actually Welti and Devantier spent extensive time reading my article and helping me run sims to come up with some of the subwoofer positioning options I listed in the diagrams.
What the sims don't tell you is that, broadly speaking, placement of any individual sub is far less important that simply having enough pressure sources in the room to smooth out room modes. And the big weakness of Welti and Devantier's work is that, until SFM they focused rather myopically on symmetrical placements. That is to say, they simply didn't test placements that lacked at least one axis of symmetry.

Quote:
Originally Posted by gene View Post
My article gives recommendations for rectangular rooms
Without giving the only really useful advice about subwoofer placement: put them where they will fit.

Really, the notion that one has to fit modal-region pressure sources into some sort of fixed pattern is not only obsolete but also counterproductive if the goal is smooth response. (Below the modal region, placement matters even less, and unless the room is basketball gym-sized the subs sum in-phase basically no matter where they are.)

Quote:
Originally Posted by gene View Post
The goal of any multi sub system should be to get all the subs working as best as they can together and integrated with the main channels.
Perhaps a semantic, and pedantic, nitpick, but I would reverse those. In fact, I don't care at all how they work together without the mains. Modern best practices[/i] focus on integrating one sub with the mains, and then sequentially blending the other subs in.

Quote:
Originally Posted by gene View Post
Only at that point should you even consider applying a global EQ to all the subs. This is a last resort and often only minimum EQing is needed if speaker placement, listening seat placement and calibration is done correctly.
Agreed. The overwhelming majority of the magic comes from getting the relative levels (which is why the "gain-matching" folks are trading away smoothness of response for more output, but I digress) and the phase/delay right. EQ employed should be minimal, and IMO focused on setting the low-end corner via boost or correcting something like a falloff in output due to two subs rolling off sooner than the third.

The best measuring*-and-sounding bass system I've had thus far (a Tannoy B475 and two DIY JBL 2235H-based subs in a room that had three masonry walls and three cardboard-n-spit walls) literally required no EQ.

*The "if it doesn't go down to 3Hz, it's crap" brigade would disagree, because the LF corner of the system was in the mid-20s.

Quote:
Originally Posted by gene View Post
I am well aware of SFM but I am not writing an article to promote a particular companies solution.
When you cite Harman people to the exclusion of others, you're tacitly endorsing Harman's approach.

Quote:
Originally Posted by gene View Post
SFM isn't even commercially available for home theater products. Right now its only offered via JBL Synthesis as far as I know.
One can go through the same iterative process manually. Just look at the PDF I linked to, supra.

Quote:
Originally Posted by gene View Post
Yes I still believe its a measurement resolution issue.
I just don't see how a resolution issue can make, as Josh wrote a few posts above, "the major response features appeared to be shifted." [emph. added]

A software bug in the display rendering seems to be the most logical cause of the behavior Josh observed, followed by perhaps a software bug in the measurements themselves.

Quote:
Originally Posted by gene View Post
I am not surprised by Josh's results
Except for the apparent bug in the measurement display, I'm not either. Velodyne's always made good subs, and it is entirely reasonable to assume that the same power applied to a driver of equal or greater throw in a bigger box will lead to more output with less compression.

Quote:
Originally Posted by gene View Post
I will engage Velodyne again on this issue after the holidays.
It'll be interesting to see what they have to say. Josh's review shows that they have a very solid product on their hands, except for the one bug with the measurement display.

DS-21;

If you have Dr. Floyd Toole's book "Sound Reproduction", I suggest you read Chapter 13. This chapter is very clear about optimal positioning of subs in a room to smooth out bass response. As far as my understanding of ARCOS & SFM, the system will actually measure the room and suggest optimal locations for the subs. Keith Yates actually makes a living doing this using FEA to find the best locations for subs in any shaped room.

Random placement to create "pressure sources" will not give you as smooth a response and will involve you using more EQ than necessary if you were instead to focus more on subwoofer placement.

I don't have time to go point for point with you on the forum, especially with XMAS coming, but I may address this more formerly in a future article.

Low Q bass issues don't change much with only a few inch shift in mic position so I don't subscribe to what you are saying here. I weigh importance of measurements per seat. In my theater I have 2 money seats, the front row center for 2CH listening and the back row center for multi channel listening. My primary goal is to get the best response in those two locations. I then check the other seats to make sure I didn't do any harm to those. The bass in all 5 seated locations of my theater is very consistent with a bit more emphasis in the back row since its closer to the back wall. I prefer this for multi channel hence why I built a riser/resonator platform to even make this effect more dramatic. I use very little global EQ in my system b/c I took the time to find the best subwoofer locations in my odd shaped room and also tweaked level and delay to each sub. Getting the combined output of your subs to properly integrate with your main channels is CRITICAL and often overlooked since many calibrators don't do fullrange sweeps of the subs + mains. I discuss this in my article and it has been looked at by many experts in the field smarter and more knowledgeable than myself. Chris at Audyssey is on the very same page as I am on this point and I am happy that Audyssey has finally updated their system to do a global EQ correction of all subs simultaneously. Chris and I went over this type of approach in detail at my place when I was setting up my theater 5 years ago. It was incredible to do this experiment with him first hand and it taught us both a great deal.

 

[DS-21]

If you have Dr. Floyd Toole's book "Sound Reproduction",

Seriously, Gene? Have you read the posts to which you've replied? I've cited Dr. Toole's book several times on this thread. One generally cannot cite page and section references of a book that s/he does not have.

I suggest you read Chapter 13. This chapter is very clear about optimal positioning of subs in a room to smooth out bass response.

Um, Gene, I cited a section of chapter 13 above. I think you need to re-read it. To give you a guided summary by someone who gets paid to interpret intricate and detailed texts for a living, the chapter goes through the interesting but ultimately limited-relevance Welti and Devantier studies (only symmetrical placements studied, at one elevation, in idealized rectangular rooms, etc.), before leading to SFM. If you have the book, turn to page 234 to see a discussion of SFM applied to Dr. Toole's own room - and note that the cleanest placement, at the bottom of fig. 13.18 at 235, is asymmetrical.

Keith Yates actually makes a living doing this using FEA to find the best locations for subs in any shaped room.

I know that. It's interesting stuff.

I'm sure that there are some placements that are better than others. But the magic set of placements is going to be different in every room, and it is toweringly unlikely to be a symmetrical placement. Otherwise, Yates wouldn't be able to make a living doing this stuff!

My point is simply any distribution of pressure sources around a room will be better than clumping all the bass in one pressure source by using a single sub and highpassing the mains in the modal region. Yes, one should experiment with placement to the extent one can. But the fact of the matter is that subwoofer placement can only drive the configuration of the room so far. One needs to balance it with aesthetics, flow, SO approval, and so on. The only thing one really loses by having to use EQ or set the level of one or more of the subs lower is a bit of power headroom due to the EQ cuts required. But that's only an issue for people who have marginal headroom to start. IMO, as long as the primary sub is on paper capable of adequate headroom for one's SPL demands by itself, one won't fall below that floor, so it'll work out even if EQing is required at the end.

Random placement to create "pressure sources" will not give you as smooth a response and will involve you using more EQ than necessary if you were instead to focus more on subwoofer placement.

First, "pressure source" is just a shorthand for "a main running full-range or a subwoofer."

Second, I suggest you try it before you knock it. If you competently apply something like the Geddes method - see link, supra - you'll be pleasantly surprised. I know I was, and I had tried some of the ideas from Welti's original PowerPoint prior.

The only places I've heard more natural-sounding bass than from a multisub system set up along Geddes' lines have names like Musikverein, Staatsoper, Grosser Saal der Philharmonie, Royal Concertgebouw, Royal Albert Hall, Woodruff Symphony Hall, Carnegie Hall, etc.

Though last spring I heard a well-executed double-bass array for the first time, and found it to have the same natural bass character as the Geddes-style systems I've set up in my own homes and two homes of friends. The tradeoffs are that the DBA requires a fairly sealed rectangular room, very specific - though easy to do - placements, a lot more volume displacement for a given SPL; on the plus side it does allow one to use fewer amp channels than a Geddes or SFM style system. The DBA I heard had all 8 15" Dayton Reference woofers - 4 in a specific pattern on the front wall, 4 in a specific pattern on the back wall - were all driven off of a single Behringer EP2500 amp, 4 fronts on one channel and four rears on the other.

I'm also intrigued by Wayne Parham's idea of using two subs as "flanking subs" on the floor and blended with the mains up to 200Hz or so, basically following on Ray Allison's ideas, and then two more randomly dispersed in the room with appropriately lower crossovers a la Geddes. I expect that in my next home I'm going to experiment with that approach, because it provides the advantage of more pressure sources in all the modal region without the localization issues. I don't know if any of your subs are good enough to do that - few commercial subs that can also go low are really good up to 200Hz - but if they can it might be worth trying since you have four of them.

I weigh importance of measurements per seat.

And you're not getting good ones unless you take a spatial average for each position. Now, if you use your subs too low (the conventional wisdom of an ~80Hz lowpass on the sub and corresponding high-pass on the mains is just a recipe for bad-sounding bass), perhaps you're right that the spatial average is not necessary. (One other nice thing about a proper multisub setup is that one can use the subs higher without localization becoming a factor.) But I'm talking about a more sophisticated setup designed to get the smoothest modal-region response, and thus the most natural-sounding reproduction of acoustic bass sounds in music. For "theater," I suppose it doesn't matter much as long as it booms.

Getting the combined output of your subs to properly integrate with your main channels is CRITICAL and often overlooked since many calibrators don't do fullrange sweeps of the subs + mains.

YES YES YES YES YES! YOU'RE ABSOLUTELY RIGHT

Every approach that does not take into account the interaction between mains and subs is so severely flawed as to be unworthy of mention. If nobody takes anything else away from this discussion, they should take this this point away.

Chris at Audyssey is on the very same page as I am on this point and I am happy that Audyssey has finally updated their system to do a global EQ correction of all subs simultaneously.

I use and like Audyssey - I was even an early adopter of their Alpine-made MultEQ XT car box - but the problem is what they do before pining the subs together. Subs should not be matched and then EQ'ed as one, but should be sequentially integrated into the mains, a la Geddes and SFM. The approach I've found to work best is to manually set the subs sequentially up with the mains, and then run Audyssey.

 

[jinjuku]

then[/i] run Audyssey.

With you there 100%. I manually EQ each sub in position and then let Audyssey sum it and then make any manual tweaks to user preference.

 

[Sputter]

With you there 100%. I manually EQ each sub in position and then let Audyssey sum it and then make any manual tweaks to user preference.

You'd still have to EQ once you sumed the subs together. I'm not seeing the advantage by doing that each first.

 

[gene]

Seriously, Gene? Have you read the posts to which you've replied? I've cited Dr. Toole's book several times on this thread. One generally cannot cite page and section references of a book that s/he does not have.



Um, Gene, I cited a section of chapter 13 above. I think you need to re-read it. To give you a guided summary by someone who gets paid to interpret intricate and detailed texts for a living, the chapter goes through the interesting but ultimately limited-relevance Welti and Devantier studies (only symmetrical placements studied, at one elevation, in idealized rectangular rooms, etc.), before leading to SFM. If you have the book, turn to page 234 to see a discussion of SFM applied to Dr. Toole's own room - and note that the cleanest placement, at the bottom of fig. 13.18 at 235, is asymmetrical.



I know that. It's interesting stuff.

I'm sure that there are some placements that are better than others. But the magic set of placements is going to be different in every room, and it is toweringly unlikely to be a symmetrical placement. Otherwise, Yates wouldn't be able to make a living doing this stuff!

My point is simply any distribution of pressure sources around a room will be better than clumping all the bass in one pressure source by using a single sub and highpassing the mains in the modal region. Yes, one should experiment with placement to the extent one can. But the fact of the matter is that subwoofer placement can only drive the configuration of the room so far. One needs to balance it with aesthetics, flow, SO approval, and so on. The only thing one really loses by having to use EQ or set the level of one or more of the subs lower is a bit of power headroom due to the EQ cuts required. But that's only an issue for people who have marginal headroom to start. IMO, as long as the primary sub is on paper capable of adequate headroom for one's SPL demands by itself, one won't fall below that floor, so it'll work out even if EQing is required at the end.



First, "pressure source" is just a shorthand for "a main running full-range or a subwoofer."

Second, I suggest you try it before you knock it. If you competently apply something like the Geddes method - see link, supra - you'll be pleasantly surprised. I know I was, and I had tried some of the ideas from Welti's original PowerPoint prior.

The only places I've heard more natural-sounding bass than from a multisub system set up along Geddes' lines have names like Musikverein, Staatsoper, Grosser Saal der Philharmonie, Royal Concertgebouw, Royal Albert Hall, Woodruff Symphony Hall, Carnegie Hall, etc.

Though last spring I heard a well-executed double-bass array for the first time, and found it to have the same natural bass character as the Geddes-style systems I've set up in my own homes and two homes of friends. The tradeoffs are that the DBA requires a fairly sealed rectangular room, very specific - though easy to do - placements, a lot more volume displacement for a given SPL; on the plus side it does allow one to use fewer amp channels than a Geddes or SFM style system. The DBA I heard had all 8 15" Dayton Reference woofers - 4 in a specific pattern on the front wall, 4 in a specific pattern on the back wall - were all driven off of a single Behringer EP2500 amp, 4 fronts on one channel and four rears on the other.

I'm also intrigued by Wayne Parham's idea of using two subs as "flanking subs" on the floor and blended with the mains up to 200Hz or so, basically following on Ray Allison's ideas, and then two more randomly dispersed in the room with appropriately lower crossovers a la Geddes. I expect that in my next home I'm going to experiment with that approach, because it provides the advantage of more pressure sources in all the modal region without the localization issues. I don't know if any of your subs are good enough to do that - few commercial subs that can also go low are really good up to 200Hz - but if they can it might be worth trying since you have four of them.



And you're not getting good ones unless you take a spatial average for each position. Now, if you use your subs too low (the conventional wisdom of an ~80Hz lowpass on the sub and corresponding high-pass on the mains is just a recipe for bad-sounding bass), perhaps you're right that the spatial average is not necessary. (One other nice thing about a proper multisub setup is that one can use the subs higher without localization becoming a factor.) But I'm talking about a more sophisticated setup designed to get the smoothest modal-region response, and thus the most natural-sounding reproduction of acoustic bass sounds in music. For "theater," I suppose it doesn't matter much as long as it booms.



YES YES YES YES YES! YOU'RE ABSOLUTELY RIGHT

Every approach that does not take into account the interaction between mains and subs is so severely flawed as to be unworthy of mention. If nobody takes anything else away from this discussion, they should take this this point away.



I use and like Audyssey - I was even an early adopter of their Alpine-made MultEQ XT car box - but the problem is what they do before pining the subs together. Subs should not be matched and then EQ'ed as one, but should be sequentially integrated into the mains, a la Geddes and SFM. The approach I've found to work best is to manually set the subs sequentially up with the mains, and then run Audyssey.

DS-21;

I had Floyd Toole look over this thread personally and here is what he had to say:

Hi Gene,

Now that SFM is "out there", who knows what is being claimed for it.

I see that DS-21 says "
What the sims don't tell you is that, broadly speaking, placement of any individual sub is far less important that simply having enough pressure sources in the room to smooth out room modes. And the big weakness of Welti and Devantier's work is that, until SFM they focused rather myopically on symmetrical placements. That is to say, they simply didn't test placements that lacked at least one axis of symmetry.

This is simply not true. It is entirely possible to place several subs in a room and make the result intolerable if they should, by chance, amplify one or two modes that negatively affect key listeners. Multiple subs need to have some form of "organization" - either by placement in a room of known geometry, or by signal processing, in which case there can be some flexibility in both the room geometry and the placement. And, in both instances, good results are limited to specific listening areas or positions. There are no "universal" cures. The original research reported by Welti was constrained to rectangular rooms. In such rooms, if one is to have any serious hope of controlling specific modes, one must employ symmetrical placements. It wasn't "myopic", it was a useful starting point for a many listeners on limited budgets. Later we addressed the matter of non-symmetrical rooms, or symmetrical rooms in which only non-symmetrical sub arrangements were possible. In these cases, which must be considered individually, SFM attempts to make the best of whatever, possibly arbitrary, sub arrangement exists. To improve on this, one can go through an iterative trial and error process to find more optimum arrangements. Or, one can engage room modeling, and get closer to the optimum arrangement first time. In the latter case, though, it is likely that a final SFM tuning will be useful, because room modeling programs make simplifying assumptions while SFM deals with reality.

Swept 1/3-octave smoothing of a high rez measurement is much more useful than old fashioned fixed-center-frequency 1/3-octave filters, yielding the "how many bars are there" staircase display. However, if one has high rez data, and one wishes to eliminate "grass", it may turn out that 1/3-octave smoothing is too much - it is something to be experimented with. There is NOTHING sacrosanct about 1/3 octaves.

I thought I saw a comment about multi mic measurements using a mixer - if so, that is an absolute no-no. The data from each location must be captured separately and the results averaged. But then one gets a result that applies to no listener. It is better to display the multiple measurements so that one can see the variations - and then, look at the average to see what it really means.

As you can see, you are not properly presenting and interpreting his research. SFM is a tool to help open up more placement flexibility but it doesn't negate proper positioning of the subs in the room.

Spatial averaging the measurements like you are doinghas some serious issues. Using 1/3rd octave measurements is not revealing or accurate especially if they aren't originally taken in high resolution.

I suggest you re-read the articles I have authored on this topic and spend some time digging deeper into the subject matter. No professional installer I know of is doing your approach and using 1/3rd octave resolution measurements to do any sort of meaningful bass EQing.

Further discussion on this topic should continue in a dedicated thread b/c its taking away from the original purpose of this thread which is to support the review of the Velodyne DD-18+.

 

[jinjuku]

You'd still have to EQ once you sumed the subs together. I'm not seeing the advantage by doing that each first.

Notice I said manual tweaks are last. I use Audyssey to get everything in the neighborhood and then optionally tweak to taste.

The subs are externally eq'd outside of Audyssey.

It's like cooking I follow the recipe and then tweak it out

 

[jinjuku]

To improve on this, one can go through an iterative trial and error process to find more optimum arrangements. Or, one can engage room modeling, and get closer to the optimum arrangement first time. In the latter case, though, it is likely that a final SFM tuning will be useful, because room modeling programs make simplifying assumptions while SFM deals with reality.

This is what I was getting at. Some times you have to put your time in and hope that the placement doesn't get in the way of other important things. This is true in an odd shaped room.

Think about it. The odd shape room is normally a multifunction room in a lot of peoples home. The perfect dimensioned room is taking into account you have an owner that has the funds and means at their disposal to put in a proper room. Part of these papers, while not overtly stating it, have a certain target audience. They aren't necessarily writing to the guy that is running out and picking up two BIC H-100's.

 

[cjwhitehouse]

One difference between the old DD subs (including the SMS-1) and the new DD+ subs is that with the old subs, the sweep tone was generated within the sub. It was therefore always in sync with the on-screen display and the associated moving filter applied to the microphone input. With the new DD+ subs, with the exception of the SELF-EQ mode, the sweep is generated from an external source CD. A key part of the new process then involves the sub syncing itself to the source sweep. If that doesn't work perfectly you get measurement errors that result in incorrect EQ results.

I recently acquired a pair of DD18+'s and raised the issue with Velodyne that the sync'ing process does not always seem to be entirely reliable, resulting in variable results. They acknowledged that they too had seen this issue and it is a known problem, particularly where the raw room response for the sub position is very poor to start with. Their recommendation was to do the crawl test first to get a better basic starting position and then the syncing process should be more reliable.

This may explain at least part of the measurement issues that you observed.

 

[Ricci]

CJ,

What you proposed is certainly a possibility. I would have to suspect that many users would experience the same sort of problem and may not in fact even be aware, since most would not double check with other measurements. I've never been a fan of using test signals that are not generated from within the software for this reason.

 

[lsiberian]

How do I miss these classic threads. A thread on a Velodyne sub going 10 pages is quite surprising.

 

[cjwhitehouse]

CJ,

What you proposed is certainly a possibility. I would have to suspect that many users would experience the same sort of problem and may not in fact even be aware, since most would not double check with other measurements. I've never been a fan of using test signals that are not generated from within the software for this reason.

It's not entirely clear to me why Velodyne made this particular change. Yes, it eliminates one additional cable between sub and amp/processor but at the expense of introducing another technical challenge and the potential for poor results from those users not prepared to experiment and persevere with manual EQ and/or use of separate measurement tools. The acronym K.I.S.S. comes to mind.