Home Theater Forum and Systems banner
1 - 18 of 18 Posts

·
Registered
Joined
·
22 Posts
Discussion Starter · #1 ·
It seems like a slightly complicated topic and neither google or forum search resulted in much conclusive discussion.

Obviously, sealed subwoofers in their 2nd order nature roll off relatively early in comparison to vented (4th order) and bandpass (6th order) enclosures. To compensate this, I've understood people cut the high frequencies, or boost the frequencies a certain amount of decibels at the frequency they want the subwoofer to extend to, and then apply a negative gain equal to the amount of boost to prevent the DSP output or amplifier from clipping. Is this about right?

Second and most important question: is there any general consensus in how much can you EQ?

Here's some theoretical "new -3dB points" for a cheap Hertz Dieci DS300.3 in a small sealed enclosure, at boost amplitudes of +6dB, +9dB and +12dB. The negative gain compensation is not there in the transfer function magnitude window because it's annoying comparing the graphs for the -3dB point if they're applied.


In the SPL window you can see the tradeoff between dynamic range and low-frequency extension.


Group delay rises somewhat relative to the filter amplitude used


Phase warping looks quite severe on the +12dB and +9dB, but I do now know how to correlate this window with perceived sound quality. Clarification and explanations would be greatly appreciated.


I'm wondering how far can you stretch the equalized -3dB point from the "real" -3dB point before you hear negative side-effects of the EQ used? Assuming the user is okay with the headroom loss in maximum SPL that is impossible to avoid, is there a point after which distortion, phase or group delay raises dramatically and EQ turns counter-intuitive? This is a very difficult concept to grasp because simulators do no not offer distortion figures and I haven't found any discussion about it.

Apologize for the potentially confusing post and bad wording, I just really wish someone would be able to clarify this topic.
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #3 · (Edited)
Hi Mike, thanks for the reply!

Here's the excursion plot.


I don't understand why would I need a more powerful amplifier though. I understand I would need double the power to double the SPL, but I'm not trying to get more SPL at lower frequencies like the SPL graph points out, I'm just trying to even out the response.

Here's what I mean, I lose SPL and dynamic range when I compensate the frequency response like this, but I'm not about to overextrude the driver or fry the voice coil/amp because the signal stays below 0dBFS when compensated properly.


What I find out is, do I still get a lot of distortion by doing this?
I'm in the process of building some high-WAF subwoofers in tiny sealed enclosures, and if I can grab myself a low-distortion microphone I will try to run some THD measurements on the equalized and unequalized sub to find out how does this equalizing steepness/amplitude affect distortion. I'll report of course, but in the mean time any information on the topic would be greatly appreciated hence the thread.
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #5 ·
Hello again! I never got much closure on the topic around here, and thought I'd chime in on some stuff I found out, just in case it'd help another audio person later when they might stumble upon this thread.

Apparently the idea I'm describing has been researched in more detail with more expertise by Linkwitz Lab, and it is known as the Linkwitz Transform. I understood it about thusly: every time you extend the -3dB point (F3) lower by an octave (halving the frequency), you lose about -12dB of dynamic range. This is because apparently, sealed subwoofers with a Qtc around 0.707 have a natural second-order roll off (12dB/oct). Makes sense now that someone more mathematically inclined put it into simple words. These Linkwitz Transform circuits can apparently be built using regular components but they seem a bit complex.

I suspect one could get similar results with a simple digital parametric EQ, but I'm not entirely sure.

Now I'm wondering about the effect of room gain in relation to subwoofer roll-off.. Interesting is the uncertainty and lots of differing information about room/boundary gain on the Internet.. This document from Synergetic Audio Concepts seems to get some good data, but doesn't describe the wavelength/frequency in relation to size/material of the walls so one could figure it out on their own room themselves..

The search continues.. If it comes down to that, I will have to get my own measurement microphone and do some own research :D
 

·
Registered
Joined
·
2,074 Posts
seepra, I'm sorry I missed your original posts, I have been fascinated with Linkwitz Transforms for a long time (and I still don't fully or properly understand them :duh: ) but I would have at least chimed in to give you a lead towards some reading.

Since you have seen the Linkwitz Labs stuff, take a look at the miniDSP if you haven't already seen that. I think this is the LT solution du jour, rather than actually building a specific circuit from electronic components.

http://www.minidsp.com/applications/advanced-tools/linkwitz-transform
https://www.minidsp.com/support/forum/suggestion-box/542-re-linkwitz-transform?limitstart=0
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #7 ·
seepra, I'm sorry I missed your original posts, I have been fascinated with Linkwitz Transforms for a long time (and I still don't fully or properly understand them :duh: ) but I would have at least chimed in to give you a lead towards some reading.

Since you have seen the Linkwitz Labs stuff, take a look at the miniDSP if you haven't already seen that. I think this is the LT solution du jour, rather than actually building a specific circuit from electronic components.

http://www.minidsp.com/applications/advanced-tools/linkwitz-transform
https://www.minidsp.com/support/forum/suggestion-box/542-re-linkwitz-transform?limitstart=0
It's okay, I might not have put the most descriptive title for the thread so it might steer people away.

Thanks for the link, the miniDSP and biquad equalizer look amazing for not only fast prototyping but finished products too! The possibilities are endless (besides it's quite badass to think you would have your own D/A conversion for subwoofers alone :D

Can't wait to order my Dayton Audio sub and get cracking! Later on I should definitely get a MiniDSP, but first I'll be busy building my first subwoofer and bracing and assembling the enclosure, I'm sure it's not half as simple as it sounds :D
 

·
Registered
Joined
·
2,074 Posts
Well, it is... And it isn't. Lol. Just read a few detailed build threads on here and pick up on some of the hints and tips that the experienced builders point out. I'll link you to a couple of great ones tomorrow when I'm back at my desk. Just don't rush things, and test fit often if you have time.

There have been a lot of great toys for EQing and tweaking subs... The BFD, Marchand bassis, velodyne SMS 1, others I can't remember off the top of my head... But the miniDSP has to be one of the better ones.
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #9 · (Edited)
Well, it is... And it isn't. Lol. Just read a few detailed build threads on here and pick up on some of the hints and tips that the experienced builders point out. I'll link you to a couple of great ones tomorrow when I'm back at my desk. Just don't rush things, and test fit often if you have time.

There have been a lot of great toys for EQing and tweaking subs... The BFD, Marchand bassis, velodyne SMS 1, others I can't remember off the top of my head... But the miniDSP has to be one of the better ones.
I was considering the Behringer DCX2496 myself, since it could handle a stereo crossover between active mains and active subwoofers, with control over filter steepness even! However, I read some less favourable reviews about it's tendency to run out of processing capability when trying to apply room EQ + crossovers simultaneously, so I gave it a pass. Investing in multiple DSP rack units seems silly too, so I'd rather get a MiniDSP to handle the stuff and give the real winner some more market share than support inferior products. Otherwise the DEQ/DCX/BFD are rather nice, as I've understood they are transparent with high-end DACs like the Benchmark DAC1 (some blind-test by Matrix Audio implied such), but the ergonomics and diversity are lacking, so MiniDSP it is!

PS. I figured that applying -x static gain and +x of parametric boost at some specific Q seems to result in less sacrifice of dynamic range in relation to the Linkwitz transform and using a less steep biquad parameter EQ. I don't know yet are the steeper changes in group delay or phase shift audible, but in a few months that I can get the MiniDSP I should try it out. Would be great to sacrifice less dynamic range for the benefits of Linkwitz transform.

Here's a screenshot of what I mean with parametric EQ vs. Linkwitz biquad parameter, amplitude wise., both are equal distance away from Xmax (like -5% or so) Group delay and phase shift are worse in the flatter parametric EQ, but I suspect they are well under what's audible. We will see in the not-so-long distance future.
 

·
Registered
Joined
·
2,074 Posts
Hey seepra, sorry I'm late with my sub build reference links. There have been lots of good ones on the forum, but a few stood out for me and will definitely impact my next build. Check out a few of them...

1) Mario's sub build in his Cinemar thread. It was a "flat pack" precut panel assembly, but he uses very good construction practices (throughout the whole thread, actually) and provides a lot of detail. Take note on the inserts and socket head cap screws he uses to attach the driver. Also making sure everything is straight and square.
http://www.hometheatershack.com/for...emar-home-theater-construction-thread-81.html

2) dtsdig made some great looking cabinets with curved walls here if you feel like getting more adventurous.
http://www.hometheatershack.com/for...7838-stereo-integrity-18-d2-curved-build.html

3) Alex build a VERY solid dual layer box here.
http://www.hometheatershack.com/forums/sealed-subwoofer-build-projects/66648-dual-opposed-18-a.html

4) There's a great looking butcher block top here on 16hz lover's project.
http://www.hometheatershack.com/for...4168-8ft-long-4-18-si-obsidian-sub-build.html

5) And some very nice bracing here from Trike.
http://www.hometheatershack.com/for...ld-projects/65706-sealed-lms-ultra-build.html

EDIT: 6) One more from baniels with all dowel bracing if you don't feel like cutting windowpanes or panels for the inside.
http://www.hometheatershack.com/for...cts/64052-stereo-integrity-ht-18d2-build.html

There are others, but I'd have to go back a lot farther. Really, just be careful, glue all joints (glue and clamps works fine, some glue and screw, or glue and pin in place with a brad nailer), make sure your cuts are good and your walls are square, and take your time. You'll be fine.
 

·
HTS Moderator , Reviewer
Joined
·
3,381 Posts
It's okay, I might not have put the most descriptive title for the thread so it might steer people away.
If you would like the thread title changed let me know and I'll take care of it for you.
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #12 · (Edited)
Thank you for the awesome links, Owen. And thanks JMan I'll ask if I come up with a better one!

I'm wondering a bit here. Assuming the equalizing boost filter that is not disproportionately narrowband (around Q 2.0-1.0), if I stay well below excursion limit (keeping a safe marking of say, -15% from Xmax), the question is..

is the distortion going to raise from Linkwitz-transform or otherwise extending the -3dB point, or am I only losing headroom? If I'm really only sacrificing headroom, this seems like an amazing solution for everything but the lowest of low infrabass and highest of SPL applications, at least adequate for music material and great for amateur monitoring!

To my understanding, at least in D/A conversion and amplification, distortion raises greatly when you approach the limits of the bandpass. I would imagine the same applies to the natural roll-off point of a sealed subwoofer, even if you never went past excursion or RMS. I have now ordered a measurement microphone, a D/A converter with phantom power, and a JBL GTO1514. I wanted a Dayton Audio, but the greater excursion of the JBL seems to give a lot more confident simulation in this particular application (for the price, I was comparing the 15" and 18" classics to the JBL GTO1514, the reference Daytons would've been a lot better but costly)
 

·
Registered
Joined
·
2,074 Posts
I wish I could help you on the distortion, but I have no idea. I'm sure by extending the low end, it must increase distortion in some amount. I would guess that each driver would handle it differently, depending on the individual characteristics and build quality. Another guess tells me that at reasonable (+3db?) levels, it shouldn't be significant or audible. The harder you push, the more obvious it gets.

Absolutely no scientific facts to back any of that up.
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #14 · (Edited)
I'm bumping the thread with some practical findings after building a 15" sealed sub that is intended to be flat with this principle.



Even though I'm staying well beyond both excursion-related and thermal power handling, the audible distortion at subsonic frequencies becomes obvious easily when extending the response a lot. I take this is partially because of the Fletcher-Munson curve (the fundamental low tone being more difficult to hear than it's high order harmonic distortion components). It really started distracting only at sound pressure levels required by movies. For music, this EQ is alright even though 12dB sounds like a lot of EQ! Extending the -3dB till about 20Hz like in the screenshot above worked at music listening levels and sound pressure levels you would normally consider "within neighbour tolerances" in dorm usage. Since I do live in an apartment soon, the subwoofer and the EQ are a success. The distortion only begins to climb at movie SPLs

THD+N was around 0.8% throughout the range at volumes I personally considered good for music listening. At movie levels, distortion reached 2% at the 20Hz peak. There was lots of higher order harmonics so it sounded rather audible with sine sweeps. With movies and explosions it wasn't as perceptible but knowing it was there made it easier to spot. (I might have cranked the volume up from the reference point until some sort of unintended noises were easily noticeable with the movie material, and then backed down close to the point where I initially measured 2% distortion. The 2% distortion's audibility might've been something else rattling in the room as I've been told 2% shouldn't be audible) It's good to know the distortion climbs rapidly, and severe EQ restricts the subwoofer usage way more than just the -12dB of static gain used to compensate the +12dB Q 1.5 peak at 20Hz.

I'm currently not with my sub (build thread here) but I'm planning on doing some THD+N measurements while I ramp up the volume gradually to see where the sweet spot lies. I wish I had a SPL meter though!
 

·
Registered
Joined
·
102 Posts
I'm bumping the thread with some practical findings after building a 15" sealed sub that is intended to be flat with this principle.



Even though I'm staying well beyond both excursion-related and thermal power handling, the audible distortion at subsonic frequencies becomes obvious easily when extending the response a lot. I take this is partially because of the Fletcher-Munson curve (the fundamental low tone being more difficult to hear than it's high order harmonic distortion components). It really started distracting only at sound pressure levels required by movies. For music, this EQ is alright even though 12dB sounds like a lot of EQ! Extending the -3dB till about 20Hz like in the screenshot above worked at music listening levels and sound pressure levels you would normally consider "within neighbour tolerances" in dorm usage. Since I do live in an apartment soon, the subwoofer and the EQ are a success. The distortion only begins to climb at movie SPLs

THD+N was around 0.8% throughout the range at volumes I personally considered good for music listening. At movie levels, distortion reached 2% at the 20Hz peak. There was lots of higher order harmonics so it sounded rather audible with sine sweeps. With movies and explosions it wasn't as perceptible but knowing it was there made it easier to spot. It's good to know the distortion climbs rapidly, and severe EQ restricts the subwoofer usage way more than just the -12dB of static gain used to compensate the +12dB Q 1.5 peak at 20Hz.

I'm currently not with my sub (build thread here) but I'm planning on doing some THD+N measurements while I ramp up the volume gradually to see where the sweet spot lies. I wish I had a SPL meter though!
You should probably look at using wider bandwidth on the PEQ if it is possible. 1.5Q is less than 1 octave wide. It will surely give you a nice, juicy hot spot on the excursion profile at 20Hz. To make matters worse, your roll off is worse because of it and group delay artifacts (aka ringing) may or may not be present. The filter is rather sharp and a Linkwitz-Transform most definitely it is not! 12 dB boost is too much for this driver under HT, especially with so narrow a filter. 14.5 mm of Xmax is not exactly prodigious and it is only a 15" driver at that. What are you using to equalize?

I don't understand the static gain thing or why 2% distortion is unacceptable at 20Hz given the application. Is the -9 dB static "cut" a compensating filter placed higher up near or beyond the crossover point?

Disortion would be more a problem if it were over 5, approaching 10%. The first harmonic is at 40Hz so I suppose you can hear it using strictly a sine wave, because 20Hz can't really be heard. However, under normal program material I doubt you can discern 2% distortion and it is likely the best you can hope for given the application. Typically, for 2% distortion, the harmonic would be -34 dB from the fundamental. When you think of the myriad of other frequencies playing at a much higher level in comparison, you get a better picture of the ability to discern the distortion at so low a level relative to everything else.

After re-reading I also wonder how you are determining when 20Hz signals are occuring during a movie. Seems you would need real-time analysis to determine this and then how do you discern a 40Hz harmonic of 20Hz, when it is entirely possible that 40Hz is already present in the program content?

That said, it seems I am confirming everything you have observed, but you must realize that given the driver, you really have the best that you can hope for. If you are able, you should get your filter widths out as wide as possible and use one at 20Hz with a gain of about 8dB, with a second filter higher up with the same width cutting back equally +/- 1 or 2 dB. You can simulate the filter effects in Winisdpro. The notch filter thing is not really a good thing for a sealed subwoofer, because you are giving away some of its advantages and creating a hot spot in the FR. Getting low and loud requires multiples or wider-swinging drivers. Larger ones as well.:T
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #16 ·
Sorry if I started calling it actually a Linkwitz-transform at some point, that was most certainly not my intention as I understand it's way different of a filter than just a simple peak. I just called it that because I'm sort of attempting the same end result. Sorry if that was offending! I can edit that out if you think it would be appropriate.

The idea of the negative static gain is, that when I apply +x amount of EQ with any type of a peak filter, I would exceed the 0dBFS point by that many decibels. By using -x amount of gain throughout the range, I don't cause clipping in the digital signal before the DAC or amp. The sum of the filters thus is just attenuation at everything else but when approaching Fc of the filter, I hope that makes sense. I decided to experiment with this instead of an actual Linkwitz-transform because my EQ doesn't support it directly.

I'm using APO Equaliser which is a really handy parametric EQ for Windows (whole OS, not a single program or player), and free! Before I had to travel some I ended up using a +9dB filter at 25Hz, Q 0.9 and liked it better, confirming your suggestion. Now I know I wasn't hearing things, thanks!

When I said "at movie levels" I meant a volume I would use when watching movies but playing sine sweeps, not actually playing a movie. Hearing anything that was not the fundamental was (maybe obviously) quite easy when playing with the volume control. I might have cranked the volume up even further after putting the mic away and watching some movie samples, I tested it a bit louder to find the point where audible side noises occured, and might not have reverted back to the same level as I did when running the sweeps. Admittedly all this was done in a hurry, in a very amateurish atmosphere as the holiday was ending, and I could definitely use some help or pointers with ground-plane measuring THD, possibly IMD if that helps any(?) and frequency response.

Thanks for the very thorough reply!
 

·
Registered
Joined
·
102 Posts
Sorry if I started calling it actually a Linkwitz-transform at some point, that was most certainly not my intention as I understand it's way different of a filter than just a simple peak. I just called it that because I'm sort of attempting the same end result. Sorry if that was offending! I can edit that out if you think it would be appropriate.

The idea of the negative static gain is, that when I apply +x amount of EQ with any type of a peak filter, I would exceed the 0dBFS point by that many decibels. By using -x amount of gain throughout the range, I don't cause clipping in the digital signal before the DAC or amp. The sum of the filters thus is just attenuation at everything else but when approaching Fc of the filter, I hope that makes sense. I decided to experiment with this instead of an actual Linkwitz-transform because my EQ doesn't support it directly.

I'm using APO Equaliser which is a really handy parametric EQ for Windows (whole OS, not a single program or player), and free! Before I had to travel some I ended up using a +9dB filter at 25Hz, Q 0.9 and liked it better, confirming your suggestion. Now I know I wasn't hearing things, thanks!

When I said "at movie levels" I meant a volume I would use when watching movies but playing sine sweeps, not actually playing a movie. Hearing anything that was not the fundamental was (maybe obviously) quite easy when playing with the volume control. I might have cranked the volume up even further after putting the mic away and watching some movie samples, I tested it a bit louder to find the point where audible side noises occured, and might not have reverted back to the same level as I did when running the sweeps. Admittedly all this was done in a hurry, in a very amateurish atmosphere as the holiday was ending, and I could definitely use some help or pointers with ground-plane measuring THD, possibly IMD if that helps any(?) and frequency response.

Thanks for the very thorough reply!
Oh, no offense taken at all. Please excuse the ragged tone of my interrogation!:devil:
I am not sure that you should worry about clipping forward of the amp. I have dialed up all the way with the BFD at 20Hz and nothing really terrible happened besides a very heavy bass sound. Same thing with the equalizer I use in Ubuntu, which has a 31Hz band that allows 12dB of boost. I have measured THD in real time with TrueRTA and it shows all the harmonic frequencies and their levels. A good guide to computing the data can be found here:
http://www.sengpielaudio.com/calculator-thd.htm

To do a "ghetto linkwitz" you need a PEQ that will let you set very wide filters. With the FBQ2496 I used to set a width of 5 octaves, boost 9 dB at 20Hz, the I set another one at 130Hz with the same 5 octave width, cutting about 8dB. The 20Hz boost runs from 3 to 113Hz, so it must obviously be cut back...but cut back correctly. The 130Hz filter would then cut between 23 - 737Hz, so there is very little attenuation of the 20Hz boost center...but every thing from 23 back up the FR is cut back neatly and almost symmetrically. The APO seems pretty configurable if what I'm seeing is correct:

http://sourceforge.net/p/equalizerapo/wiki/Documentation/
 

·
Registered
Joined
·
22 Posts
Discussion Starter · #18 ·
Yes that's APO Equalizer, it's a lovely little tool. A Finnish member at SourceForge has made a really great GUI, actually a few GUI tools that work with it. The newest and most versatile one would be the fPEGGUI-10MC found here .

I'll take your suggestion into account when I manage to get the sub here, as it's currently with my father to toy around with. The APO Equalizer should be capable of using very wide filters so I'll definitely try that.

I don't know about the maximum output level of physical equalizers (would make sense that they can put out dialed up signals), but when you're dealing with digital domain you can't go over -0dBFS without heavy digital clipping. Furthermore, I've understood that typical line level sources are 0.32 V (-7.8 dBu), and since I don't know about the maximum input signals and some pro gear can output +4dBu, I'd rather not risk it if I start using an analog interface at some point. Maybe your equalizer automatically scales the signal to fit inside the DACs dynamic range which would be handy! However I can definitely hear clipping both with my optical output and headphone output if I use too much positive gain without compensation. It is similar to the likes of this video (viewer discretion advised, horrible sound)

I don't mind being interrogated, it's nice that someone has the willingness to help a newb!
 
1 - 18 of 18 Posts
Top