Is this what a linkwitz transform is for?

Discussion in 'Archived Threads 2001-2004' started by Charlie G, May 8, 2002.

  1. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    I'm looking to build a sub (or subs) to fill in the bottom end on my speakers, which are 103-104db/watt/meter efficient 100hz exponential horns. It's not quite so much a sub as a 'woofer', since there is some response in the >100hz range.
    I'm rather limited on space, so I was fiddling with some 18" drive units and found that with two 18's in a 6 cu foot sealed box, (or one each in 3 cu foot) I get some interesting numbers. Down low, output is obviously limited by excursion (only 7mm each way on these high-eff woofers). But, I still have plenty of output at say, 30hz (>110db anechoic) The problem is, Qts is above 1, so response continues to rise quite a bit.
    So, the subs would have the output I need down low, but the reponse continues to rise as frequency does, up to >125db anechoic, at max power. So I just kinda need to cut off the high freqs, level everything off at about the 30hz level, and let it roll off below that.
    I have two concerns:
    Is this what a linkwitz transform is for? Even if it lets me simulate a larger box, even if the Qts were actually down around 0.6, the response still rises too much, I still need to shelve it off at about the level it is at at 30hz.
    I figure with a Behringer, and some simple opamp filter if needed, I can probably get the response to what I want, but this also concerns me:
    Apart from the frequency response (which will be corrected), what are the cons of such a high Qts? I think this refers to the 'damping' of the box, does it not? If the box is 'underdamped' will there be excessive distortion or anything like that?
    EDIT: Just ran some more numbers.. if I give each woofer its own 4-5 cu foot cabinet (around a 21" cube, heavily stuffed), I get response around 112db anechoic @ 30hz w/ 125 watts into each woofer. This also lowers the Qtc to about 0.72, which I like a whole lot more.
    It's unlikely I'd EVER play music this loud, and these numbers are *right* on the excursion limits for the woofers, but it's nice to have some headroom. Means the woofers are just putting along at normal listening levels, which is good [​IMG]
    Thanks
    Charlie
     
  2. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    I think the response I need is similar to the shelf EQ's used for baffle-step correction, only it needs to be 'tighter', as in transition over a smaller frequency range.

    Basically I need a filter with 0db from DC (or low, whatever), up to around 30hz, then transition to -9db up around 80hz, and stay around there from 100hz on up.

    Could I get pretty close to that using like 2-3 filters on the behringer?

    Thanks

    Charlie
     
  3. Greg Monfort

    Greg Monfort Supporting Actor

    Joined:
    May 30, 2000
    Messages:
    884
    Likes Received:
    0
    Trophy Points:
    0
    I have two concerns:
    Is this what a linkwitz transform is for?
    =====
    Yes, and lower it to the Qtc = 0.5 needed for good blending to horns.
    =====
    I figure with a Behringer, and some simple opamp filter if needed, I can probably get the response to what I want,
    =====
    Yep. Some 'hornys' now just put an amp on each horn and use digital EQ for XOs and frequency shaping.
    =====
    but this also concerns me:
    Apart from the frequency response (which will be corrected), what are the cons of such a high Qts?
    =====
    You mean Qtc. [​IMG] You can use EQ to lower it.
    =====
    I think this refers to the 'damping' of the box, does it not?
    =====
    Correct.
    =====
    If the box is 'underdamped' will there be excessive distortion or anything like that?
    =====
    Yes, it's not my choice, but some prefer it though. Ideally, with a sealed design you choose an a alignment that is the inverse of the room's gain for a ~flat in-room response.
    =====
    EDIT: Just ran some more numbers.. if I give each woofer its own 4-5 cu foot cabinet (around a 21" cube, heavily stuffed), I get response around 112db anechoic @ 30hz w/ 125 watts into each woofer. This also lowers the Qtc to about 0.72, which I like a whole lot more.
    =====
    If space is an issue, I wouldn't bother. Find the cab volume that yields the desired lowest frequency amplitude then use a LRT to re-EQ it to whatever 'floats yer boat'.
    The best non horn alignment would be an EBS (extended bass shelf). Here a really large cab tuned very low is used to get a 'shelf' that when combined with the XO's roll off yields a highly damped FR very similar to a horn's. For instance a JBL 2242 would need 20ft^3 tuned to 15-16Hz. Got room for two refrigerator sized cabs in your room? I have one in each corner. [​IMG]
    GM
     
  4. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    I heard Mike Bates's system at the Midwest Audiofest, and he has real nice horns and two HUGE woofer boxes. They sure sound great.
    Even going to bass horns was pushing it in terms of space. I really don't have room for many cubic feet.
    The thing is, I don't really want any more output than the woofer is capable of down low, which would be impossible anyway [​IMG] I just need to fix the frequency response.
    Although from what I can see, the change in Qtc that the LR offers is desireable. On top of that I'd need some kind of shelving EQ to fix the rising response (even a Qtc of 0.5 gives an F3 up near 53hz). But that is OK, because the level it achieves at 30hz (106db for a single 0.8 Qtc cab) should be plenty. It's not like I'm trying to boost the low end above what the driver is capable of or anything.
    Thanks
    Charlie
     
  5. Greg Monfort

    Greg Monfort Supporting Actor

    Joined:
    May 30, 2000
    Messages:
    884
    Likes Received:
    0
    Trophy Points:
    0
    Ok, if I understand you correctly then use the LRT calculator to sim a Q = 1 response that yields the desired response higher up, IOW set the F3 as low as necessary (if it's within the gain limit of course).

    If the F3 is low enough, then the 'ringing' will decay out before it can become audible, the same as happens with a low tuned EBS, leaving you with a ~ 0deg acoustic phase response just like a horn's.

    GM
     
  6. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    Here's the best way I can describe it:
    Using unibox, it draws me a nice line that is the output of the woofer at maximum excursion, passes through the point 30hz, 106dB. I figure for music, 30hz extension is plenty, and I'll probably have some under that. With ultimately two woofers and/or enclosures, that'll give me peak 112db anechoic... who knows what exactly in room.
    Now, in a Qtc = 0.8 box, this 30hz point is the -12db point, referenced to up around 100hz where it is 118db or so. At max power, which is no longer excursion-limited, output has curved away from the excursion limit, and at 100hz is close to horizontal. Here I am limited by power handling of the driver. This curve (106 @ 30, 118 @ 100) is at 200w, the RMS handling of the driver, in a Qtc = 0.8 box.
    I figured that if I can shelve the response, and/or pull down the high end, and keep it at about 106db from 30hz on up, that will be great. This is essentially what I want to do. The lower part of the curve (200w, Qtc 0.8), up until around 50hz, runs RIGHT along the excursion limit for the driver. I think this pretty much sets my max output down low in stone, unless I go to ported.
    I'm not sure I understand what you mean I should do with the LRT, exactly. I'd set my 'in' values to the 0.8 Qtc and the 61.24hz Fb (or is that using Fs of the driver? That's 26.6hz), right? And the desired values I'd set to... Qtc 1?
    EDIT: Ah ha! I think I understand... set the linkwitz with the input values, and desired output Qtc.. (why 1, as opposed to 0.5~0.6?) Then set the desired freq to the one that gives me about 12db of gain in the low end. Then after (or before) the circuit, have 12db of attenuation. Voila! The high end is shelved down by 12db.
    Wait... no it isn't... OK, I give up. I don't get it [​IMG]
    Thanks!
    Charlie
     
  7. Greg Monfort

    Greg Monfort Supporting Actor

    Joined:
    May 30, 2000
    Messages:
    884
    Likes Received:
    0
    Trophy Points:
    0
    Hmm, you can't sim a LRT response in a regular box program, unless Unibox has this feature. Does it?

    For instance, using the JBL 2242, a Qtc = 0.7 has a ~F3 = 80Hz in 2.11ft^3. Plugging this into a LRT calculator, I hit the suggested 20VDC gain limit with a ~F3 = 26Hz/Qtc = 0.7.

    So in this size cab the Xmax limit is ~105dB at ~26Hz, but since you can vary Qtc, you can dial in how much flat BW from this point you want based on how much room gain you get or how much 'ringing' you find acceptable. All we're doing here is trading upper BW efficency for a different alignment.

    Really, the only EQ you should need would be for room modes unless of course the LRT gain limits you from achieving your goal with your particular driver.

    GM
     
  8. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    Well, I have a separate spreadsheet that shows me the curve of the linkwitz transform. The thing is, if I want the higher freqs 12db down from where they are, then with the linkwitz I would want about 12db gain at 30hz or so, right? (which would subsequently be attenuated 12db, leaving 30hz at 0db, and 100hz at -12db)

    The problem is, with the linkwitz transform and an output Qtc around .5-.6, I can't get a gain of 12db at 30hz. If I crank the output Qtc way up to like 1, then I can get the 30hz range up the neccesary 12db relative to the 100hz range. So I'm not going to be able to completely shelve the high end with a linkwitz transform, at least not if I want the Qtc low.

    Charlie
     
  9. Greg Monfort

    Greg Monfort Supporting Actor

    Joined:
    May 30, 2000
    Messages:
    884
    Likes Received:
    0
    Trophy Points:
    0
    Ok, gotcha, then looks like you'll either need some additionsl EQ, try a different driver, or just use EQ to do it all. I was hoping you could set Q = 1 down around 15-16Hz where it would decay away enough by 30Hz.

    GM
     
  10. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    I'll probably use the linkwitz (which gives about 6db gain at 30hz, at a setting of F=30hz, Qtc = .5). The remaining 6db can be handled alright by a shelving EQ.
    My last concern is the linkwitz offers more gain below 30hz (continues up to about 8-9db at 20hz, and 12db down near DC). These levels could cause the speaker to exceed Xmax. Although like I said, I don't really expect to be using all that headroom, it does concern me. Perhaps I'll have the only sealed subs that need a subsonic filter? [​IMG]
    Hold on here a sec...
    When you say, set Q=1, you mean have the filter not adjust the Qtc, right? As in, if I have a 0.6 box, keep the linkwitz set to 0.6? If I do THAT, I get the 12db gain I need with the freq set to 15hz. Of course that means actually building a 0.6 Qtc box. It still works if I set it up for 0.7 or 0.8 (I get 12db gain at 30hz). Of course that means I still have a 0.7 or 0.8 sub, and don't get all the advantages of the Linkwitz, doesn't it? Also this further expands the super-high-gain-below-30hz problem.
    If I set f(0)= 60.24hz, Q(o)= 0.8, and f(p)= 15hz, Q(p) = 0.6, it is CLOSE. I get about 10-11db gain at 30hz. At Q(p)= 0.5, I get about 10db. Gain just plain skyrockets below 30hz, though.
    Thanks
     
  11. Greg Monfort

    Greg Monfort Supporting Actor

    Joined:
    May 30, 2000
    Messages:
    884
    Likes Received:
    0
    Trophy Points:
    0
    >Hold on here a sec...

    When you say, set Q=1, you mean have the filter not adjust the Qtc, right? As in, if I have a 0.6 box, keep the linkwitz set to 0.6?

    ====

    There ya go!

    ====

    >If I do THAT, I get the 12db gain I need with the freq set to 15hz. Of course that means actually building a 0.6 Qtc box.

    ====

    Correctomundo.

    ====

    > It still works if I set it up for 0.7 or 0.8 (I get 12db gain at 30hz). Of course that means I still have a 0.7 or 0.8 sub, and don't get all the advantages of the Linkwitz, doesn't it?

    ====

    Well, you can only do so much with a given driver.

    ====

    > Also this further expands the super-high-gain-below-30hz problem. If I set f(0)= 60.24hz, Q(o)= 0.8, and f(p)= 15hz, Q(p) = 0.6, it is CLOSE. I get about 10-11db gain at 30hz. At Q(p)= 0.5, I get about 10db. Gain just plain skyrockets below 30hz, though.

    ====

    Plugging your numbers in I get a 24.15dB DC gain, so since it's somewhat higher than the 20dB max recommended, probably not viable. f(p) = 19Hz appears to be the max. Unfortunately, like most things, speaker building is about juggling tradeoffs.

    GM
     
  12. Charlie G

    Charlie G Extra

    Joined:
    Jan 17, 2001
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    0
    Thanks a lot, man.

    I'll probably do what I can with the linkwitz, and then do a simple shelf EQ, that or take care of the last bit in a parametric or graphic EQ.

    Charlie
     

Share This Page