Purpose of flat response below 20Hz - Why does it matter? - Page 9

Originally Posted by DS-21


Wait, 14Hz on human voice? That seems rather obviously to be an artifact of something.

Originally Posted by bossobass


If you take everything below 20 Hz out of this scene, it becomes a complete non-event, certainly nothing like a real pair of low flying 20mm canons strafing you. Conversely, it's a whole new experience as recorded.

Bosso

Originally Posted by mojomike

Cool. Let me know how well that setup works out.

Originally Posted by J_Palmer_Cass

When was the last time you were being strafed by 20 mm canons? Were you in the plane, on the wing (ahh,feel that ULF), or on the ground with a bullet in you head?

You never hear the one that hits you!


Once I did have an occasion to hear a B-24 flying over my house at a very low altitude (maybe 500 feet). Cool sound that you can not duplicate!

Originally Posted by J_Palmer_Cass

You mean that you have no idea on what a close microphone setup can pickup?***Air movement from the mouth can produce very, very high levels of ULF.

Originally Posted by J_Palmer_Cass

Same process applies to anything that has been close microphoned. These same ULF transients may be noticed by musicians playing an instrument, but the same effect is not going to be noticed by a listener in the audience.

Originally Posted by morrischestnut60

I connected everything last night. I have not powered anything up as I don't have to time to do it twice. I am waiting on the filters. However, is there a reason that there are no 40hz high pass filters? I know I have a lot to learn but this stumped me. Is this achieved via a combination of other available filters? I am asking because I realized last night that the ULS-15 crossover starts at 40. I will have to engage it and I want to minimize bass overlap(30-40hz range).

Originally Posted by DS-21

At what level?



Wait, 14Hz on human voice? That seems rather obviously to be an artifact of something.

And more generally, what do your mV values correspond to in terms of level? How far below (if any) are they from the average levels on the tracks? I ask because what I want to know is whether there's anything perceptually relevant.
(I don't care about content at 30kHz, either.)



Most of the time, what people think are "very low frequencies" are not ULF. Do you have numbers and relative levels?

For instance, I once read a study that measured the bass from Tube line under Royal Albert Hall - anyone who's ever heard a concert there will know what I'm talking about - at something around 25 Hz.



No musical program material will be ignored, except for the above-mentioned "sonic spectacular" disks that people only buy (myself included; I own both the Kunzel/CinnPops 1812 and, sadly, the Wellington's Victory) to demo the effects once or twice and maybe beyond that to show off the system to guy friends - never met a girl who gave a fig about bass from ordinance... - not for actual listening.

The point of my inquiry is to determine whether designing a system with the pointed goal of ULF reproduction (below 16Hz organ pedals) is useful for the reproduction of music. Not door slams, not 20mm cannon strafes, not dinosaur stomps, etc.

Originally Posted by DS-21

Good point. I hadn't considered that the mic would be so close to the mouth, though now that you mention it, it is obvious.



That is exactly my point. Nobody on this thread has come close to showing that any stray ULF captured is perceptually relevant to music reproduction.

I am entirely open to changing my mind on subject if anyone provides relevant data to suggest whatever ULF may be captured in the recording process is perceptually relevant to music and not just a recording artifact. Thus far, that data has not been presented.



Why do you want to minimize "bass overlap?"

It is precisely that overlap that leads to smoother upper-bass response! It's what takes a system from low-fidelity mush in the upper bass:

[(single sub in a corner)

to high fidelity:
(properly configured three subwoofer system, using the above subwoofer from 20-120Hz, and the other two subs from their ~40Hz low-end cutoff until 120Hz as well.)

Of course, one does need to take measurements to properly set up a subwoofer.

Originally Posted by morrischestnut60

I dislike overlap for the same reason that I dislike double-bass on receivers. Also, I do not like my dedicated LFE channel above 80. I hate the "echoey" sounds that come from the sub. I also hear dialogue bleeding into the sub during cable viewing on occasions(with LFE above 80). I wish Kenwood made a line of Excelon HT receivers. Life would be so much simplier.

If multiples are supposed to reduce stress on your subs, why the boost at 20hz in the second pic? I understand that the subs crossed at 40 flattened out the midbass but the boost at the lower frequency me. Do you have the first sub doing the entire spectrum(when summed with the other two) which taxes it more?

Originally Posted by morrischestnut60

I dislike overlap for the same reason that I dislike double-bass on receivers. Also, I do not like my dedicated LFE channel above 80. I hate the "echoey" sounds that come from the sub. I also hear dialogue bleeding into the sub during cable viewing on occasions(with LFE above 80).

Originally Posted by pbc

It's not boosted (long story short before you get "yelled at" as I did ). He's simply using the SMS software to measure the response and doesn't actually have the SMS in the chain.

Originally Posted by DS-21

That stuff simply doesn't happen with competent setup.

With incompetent setup, yes, any audio system is a parade of horribles.

Which is why once the gear reaches a rather minimal level of competence, setup dominates.

Originally Posted by morrischestnut60

Thanks for explaining that! I hope DS divulge how he achieved that even response.

Originally Posted by pbc

Click on his name then look for threads he started, probably called "Geddes" or Multiple sub setup, or something like that. Or search his name for threads he's posted in as I think he's posted it a plethora of times now in various threads...

Originally Posted by morrischestnut60

You have one sub running 20-120 and two running 40-120. What are you using to eq them since you only use the SMS to measure the system response? I assume you level match via your receiver. Dumb it down for me. You only have to do it once.

Originally Posted by DS-21

At what level?

Wait, 14Hz on human voice? That seems rather obviously to be an artifact of something.

And more generally, what do your mV values correspond to in terms of level? How far below (if any) are they from the average levels on the tracks? I ask because what I want to know is whether there's anything perceptually relevant.

The point of my inquiry is to determine whether designing a system with the pointed goal of ULF reproduction (below 16Hz organ pedals) is useful for the reproduction of music. Not door slams, not 20mm cannon strafes, not dinosaur stomps, etc.

Originally Posted by DS-21

The credit goes to Dr. Earl Geddes, who came up with the setup method I use. It's an iterative process. Except for the part about moving the mic in an arc (taking multiple measurements and letting hardware or software average them is better, see here for more on the method: http://mehlau.net/audio/multisub_geddes/

I do my processing (level/delay-"phase"/EQ) with a cheap box called the miniDSP. There are three bands of parametric EQ in that setup. Most of the magic, and most of the work, is in getting the levels and delays right.

To save you the trouble of searching for more setup details, pretty much everything is here: http://www.avsforum.com/avs-vb/showthread.php?t=1340062

The two subs that are only running to ~40Hz are not highpassed, that's just what their passive radiators are tuned to. In the revised version (forthcoming) all of the subwoofers will be closed boxes, and all will contribute to ULF. Thus far, unloading hasn't been an issue.

For the record, I've used the same methods to achieve similar results in other systems as well. Ones with greater scale. This one is just the most recent.

It works even better (subjectively, not necessarily measured) when the mains are running full range as well! Unfortunately, the scale of these mains does not allow that. Here's the response of the first system I did using Dr. Geddes' setup methods, pre-miniDSP, in and old home, when the SMS-1 was the only processor I had. It used ZERO EQ, just three mains fullrange and three subs with relative levels and "phase" (really delay) controls used to optimize the response.

(The main sub was tuned to the high-20s, hence the rolloff.) I think the subs came in at either 120 or 140 Hz, but don't remember.

Originally Posted by kwarny

Thanks for that post Bosso. I have seen spectral analysis for cymbals and was looking for those for other instruments. You saved me a few hours today .

I have a Mac and went through at least 10 different programs yesterday before I liked one for spectral analysis but it does not accept 24 bit songs. It took me less than half an hour to find one and compare it to the others. Took me more time to convert my files from FLAC to WAV and then run an analysis. I still need to find a way to save the graphs as pictures. There is content below, it just depends if the mixer added a filter (some are brick walls).

I have some tracks from the same recording company. I've compared an unfiltered track to one of their filtered track. Both have content to at least 6 Hz. There is a drum test in a large live room also. The content when he announces the distance from drums to microphone has a peak of 1 mV at 14 Hz. When he is actually playing the drums, there are recorded values of 4.5 mV at 7 Hz. Another track with a drummer has almost no content below 20 Hz (.2 mV peak) but when the bass drum is used in the second half there is content at 9.6 mV at 7 Hz. Those are just some examples for instrument related.

As for outside noise, isn't that what connects a person to the event? Recordings of real events contain massive amounts of content below 20 Hz. Whether or not the person listens at a volume loud enough or if the noise floor in the environment is low enough is another matter as is the recording.

A larger 3 dimensional soundstage is generally attributed to the extension of the speakers. Does someone know at approximately what Hz this no longer pertains to?

Originally Posted by ReneV

Spectacular post, bossobass. Thank you!

Have you come across measurements other than dB/Hz and Hz/s that you have felt should be taken into account when reproducing sound?

Do you do anything in particular to accurately reproduce along the Hz/s axis, similar to what the *flat* part of the subject advocates for dB/Hz?

Do you know if there is something equivalent to Equal Loudness Curves for Hz/s that good recordings align themselves with?


edit: replaced dB/s with Hz/s throughout; apologies for any confusion.

* Amplifiers are a voltage source, in theory, they should be able to produce any current on demand, in order for the voltage not to collapse under load. Well, that's the theory. In practise, SS amplifiers come very close to simulating an ideal voltage source but only to a certain current limit, after which the voltage collapses very rapidly, often with entire parts of the waveform missing, not just being clipped. This is normally due to the protection circuits. These are of course designed to kick in at current appreciably larger than those needed to drive the rated RESISTIVE load (usually at least double). That being said, there are great differences in how amplifers deal with overload of this kind. This is important for the topic because the current LIMIT of a SS amplifier almost as a rule depends on the actual output voltage at any given time. Amplifiers using bipolar junction transistors in the output stage (i.e. 'regular' transistors) have to cater for one cumbersome peculiearity of the latter - the transistor cannot withstand high voltage and high current at the same time, even if voltage x current = amount of heat produced is below the rated limit. This is important because speakers are not a resistive load. In a resistive load, current increases proportionally with output voltage. As output voltage increases, voltage across transistors decreases. So, current through transitor increases, and voltage on transistor decreases - hence the rule that you cannot have high voltage and high current across the transistor at the same time is easy to satisfy.

But, as I said, loudspeakers are not resistive loads, they are reactive loads. In a reactive load, the current waveform can lead or lag voltage. An extremely reactive load will have it's maximum current in the precise moment when there is the maximum voltage across the amps transistors, which would make it very difficult for the amp to survive, if it were not for it's output protection. And this is where output protection and current capability come into play in the reproduction of transients.

High power output (i.e. voltage capability) does not per se guarantee high current output. This fact was heavily abused during the receiver power wars, because power was measured on a dummy load, which is normally purely resistive. For a realistic load, protection would kick in way below rated power, because a realistic load is reactive.

Originally Posted by kwarny

When he is talking it could be an artifact. That is why I listed the recording when no instrument is played. Every time the bass drums are hit (in both songs the ULF level is increased). The ULF content of the first track is at least the same or greater than the less than 100 Hz (4.5 mV) while at ~ 200 Hz it is about 12.5 mV.

Originally Posted by J_Palmer_Cass

I never said that I hate your subs (or any other system designs). I was just stating the obvious. Those are rather large subwoofers.

I am kind of disappointed that you are afraid to push them when you suspect that there may be higher level ULF below tune. Why not high pass those speakers at the proper frequency to protect those drivers?

Originally Posted by J_Palmer_Cass

Ah yes, that underwater type sensation. Like at the end of this clip.

http://www.youtube.com/watch?v=wru7XVriBdQ&feature=fvsr




Anyhow, how large is your room and how loud do you play that movie to get that effect? In other words, whre do you fit in with this protocall as far as room size and listening level is concerned? Is -10 to -15 below calibrated reference the level that you listen to get this effect?


http://www.audioholics.com/education...ofer-room-size

Originally Posted by Steveo1234

Alrighty then,. Out of curiousity:


Kanye west - robocop

Originally Posted by DS-21

Good point. I hadn't considered that the mic would be so close to the mouth, though now that you mention it, it is obvious.



That is exactly my point. Nobody on this thread has come close to showing that any stray ULF captured is perceptually relevant to music reproduction.

I am entirely open to changing my mind on subject if anyone provides relevant data to suggest whatever ULF may be captured in the recording process is perceptually relevant to music and not just a recording artifact. Thus far, that data has not been presented.



Why do you want to minimize "bass overlap?"

It is precisely that overlap that leads to smoother upper-bass response! It's what takes a system from low-fidelity mush in the upper bass:


(single sub in a corner)

to high fidelity:

(properly configured three subwoofer system, using the above subwoofer from 20-120Hz, and the other two subs from their ~40Hz low-end cutoff until 120Hz as well.)

Of course, one does need to take measurements to properly set up a subwoofer.

Originally Posted by kwarny

Does bass on a recording help develop a three dimensional soundstage?

Originally Posted by bossobass

DS-21 has experienced this and has mostly confused the phenomenon with a low inductance-induced increase in transient response from the driver, which has little to do with dynamic tracking of transients. He's come to the point where he is preaching that any sub driver with higher than .6 mh inductance is incapable of fidelity, which is not necessarily (or even generally) true.

Originally Posted by morrischestnut60

Glad you returned. I found the thread. You gave me the embedded link but I missed it. Awesome stuff. Quite surprised that you mentioned the mains because I started wondering about my crossovers reading the thread. My setup is

Boston VR3s main(raised from full to 60hz after audyssey)?
Boston VR920 center(raised to 80 after audyssey)?
Boston VRMX x 4(crossed at 80)
Dual ACI Titans (will fully commit and let rolloff)?
ULS-15 (will fully commit and run full)
Denon 1908(Audyssey XT)

Originally Posted by morrischestnut60

I did override the Audyssey crossovers after calibration.

Originally Posted by morrischestnut60

Can I get by without the MiniDSP and Mics($800) if I get the velodyne(since it will provide the test tones and measurement as well)?

Originally Posted by sputter1

The SMS is good for a quick and dirty setup but it's not very telling.
At 1/3 octave it's way too coarse.
REW is far more accurate and use the SMS to correct what REW is telling you.
You'd be surprised at what the SMS isn't showing. (ask me how I know )

Originally Posted by DS-21

In general, yes. I think we've all observed the "turn the subs off, feel the soundscape collapse" phenomenon. Though the degree of improvement also depends on how well the bass is set up in the room.

However, that doesn't actually answer the question I've presented, which is quite narrow: the perceptual effect of extreme ULF, not bass generally. It does not necessarily follow that a system with a low-end cutoff of 16Hz will sound less spacious than one with a cutoff at 3Hz.



First, that is a really gross exaggeration of my position. The rule of thumb that seems to track my observed reality is 1mH Le per Ω of Re. (Assuming a driver that uses shorting rings well to control inductance linearity over stroke; obviously, drivers lacking shorting rings entirely are low fidelity units.) Above that, and things start getting bad. Below that, inductance probably isn't a factor.

But if your "dynamic tracking" theory is right - I'm very happy to accept that it may be - how come using the same amp a high-inductance sub will sound thick and a low-inductance sub will sound like music? Even if the low-Le sub is in fact rather inefficient broadband. The Aurasound NS18, for example, has a 1W/1m rated sensitivity of just 86dB.

[NOTE TO MODERATORS: would it be possible to break off the post from which I quoted below and my reply into a new thread? It seems way off the stated topic.]



I'll be honest, I don't know any of those speakers, and a quick search didn't tell me what the drive unit complements are. I assume they are vented, in which case the first thing I would recommend doing is plugging the ports with foam plugs, or failing that rolled up towels or something. The mostly 2d-order rolloff of a leaky closed box is much easier to integrate with subs than a 4th order rolloff with unloading below that.



I don't know. That depends more than anything else on the cone area and volume displacement of your mains. I wouldn't run anything a mains with less than a 10" woofer (or dual 8's) full range in a "typical" room (scale that up if your room is very large or leaky, and perhaps down if it's smaller). You don't want the mains to overload.



Well...you can get (depending on placement constraints, room construction, etc.) between maybe 80-100% of the way there with just continuously-variable level/lowpass/phase controls and good measurements. If all your subs have those, and not just phase inversion, then it might be worth trying to calibrate without EQ. If not, the miniDSP is cheap and easy to use. And there are other solutions as well that provide individualized EQ for each channel. That's just the one I believe to currently be the most cost-effective considering OSX compatibility, capabilities, price, and ease-of-use.

I would not buy a Velo SMS-1 today. I use it because it's quick to use, with the MIC-5 take spatially averaged measurements in one go - without the MIC-5, I bought mine years ago when it was the only game in town. Today, it's simply too expensive (even at the current going rate of $300-400; when it was introduced it routinely sold for double that!) and too limited.

I would instead start with the Audyssey mic and FuzzMeasure Pro., which will generate tones and take measurements. (There is also not-a-Mac software that can do similar things. I have never used any of it.) You then take 5-6 measurements and the software will average them. Later on, if you want better data, buy a FireWire sound card (I use the M-Audio FireWire Solo, but don't know what's come out in the last several years) and a calibrated mic. But honestly for most consumer uses, the Audyssey mic and a Mac should be enough. (I cannot vouch for the mic-in ports of not-a-Macs, because I don't use them.) Since I discovered how closely data from just plugging the Audyssey mic into my MacBook tracks the data with the better setup, I've basically stopped using the better measurement setup.



You're simply wrong. Well, mostly wrong. I agree with you if one's talking about the SMS-1 without the MIC-5 spatial averaging kit, or another means of effecting spatially averaged measurements with the SMS-1.

I suspect you're chasing phantom high-Q artifacts of improper measurement procedure (single-point rather than spatial average) instead using valid measurements to focus on things that actually matter.

I've confirmed my results with the SMS-1 using spatially-averaged measurements through more sensitive equipment on several previous occasions (see supra) and the results are simply not materially different. So I feel no need to continue to do so. When used properly - with the MIC-5 to produce a spatial average, and the five mics set up around the primary listening position - the SMS-1 provides sufficient data resolution for home system setup. If I were a professional, I would likely use something else, if only because I'd want the graphs I showed my clients to look prettier than phone pics snapped off a TV screen. But for consumer use, the problem with the SMS-1 is that it's overpriced and lacks capability (only global EQ/phase/level, when it should be able to provide that for each channel in addition to the global control).

Originally Posted by DS-21

It does not necessarily follow that a system with a low-end cutoff of 16Hz will sound less spacious than one with a cutoff at 3Hz.

First, that is a really gross exaggeration of my position. The rule of thumb that seems to track my observed reality is 1mH Le per Ω of Re. (Assuming a driver that uses shorting rings well to control inductance linearity over stroke; obviously, drivers lacking shorting rings entirely are low fidelity units.) Above that, and things start getting bad. Below that, inductance probably isn't a factor.

But if your "dynamic tracking" theory is right - I'm very happy to accept that it may be - how come using the same amp a high-inductance sub will sound thick and a low-inductance sub will sound like music? Even if the low-Le sub is in fact rather inefficient broadband. The Aurasound NS18, for example, has a 1W/1m rated sensitivity of just 86dB.

Originally Posted by bossobass

See, this is where the problem is. You make these declarations, but what are they based upon?

Originally Posted by bossobass

Sensitivity of the driver at 1k Hz is irrelevant to the point.

Originally Posted by bossobass

My dynamic tracking statement isn't based on theory. Regardless of the sub driver used, if the amplifier used can't provide enough concentrated power to amplify the input signal of a transient without clipping, the dynamics of the source will not be properly tracked. Period.

Originally Posted by bossobass

Can you show me the data that suggests a 'low inductance sub' sounds different from a sub using a driver with no shorting ring, all else being equal?

Originally Posted by bossobass

If you take the Tumult MKI and cross it at 100 Hz (an octave above where it was designed to be crossed),

Originally Posted by bossobass

If you cross the MKI properly at 50 Hz and then compare it to the MKII (with shorting ring and flat to 500 Hz) with the MKII properly crossed at 100 Hz (properly in both cases assumes a system designed for each of those scenarios)... there will be no difference in fidelity.