Use Linkwitz Transform to reduce size of Enclosure?

[Aucosticraft]

I was simulating different closed enclosure for the same driver using WInIsd and adding Linkwitz Transform to extend its lower frequency response. While doing this I thought if it is possible to reduce the size of enclosure and extend lower frequency response of the system.
say for first design I get 50liters of enclosure with f3 of 50Hz and design Qtc = 0.7 and implement Linkwitz Transform to get F3 of 20Hz.

now next suppose I design smaller enclosure, 25liters, f3= 100Hz, Qtc = 1.5
and implement Linkwitz Transform to get F3 of 20Hz.

so can I say that with the help of LT we can reduce the size of enclosure keeping the lower frequency response same? what all things would be affected? which one of two will sound better?

 

[chris661]

What size driver are you using?
There's no point in using eq to get more bass, if you run out of linear displacement before it'll play 20hz loud enough. Having used it in the past, I wouldn't advise going lower than an octave below the uneq'd response, as the amount of gain required to go further makes an even bigger problem of excursion.
Next up, despite efforts to flatten the response, a resonant system (q=1.5) will still exhibit that resonance, so a large peak before rolloff would best be avoided.

Regards,
Chris

 

[sreten]

Hi.

You can create any frequency response you like actively, the real issue
is if the maximum SPL related to the volume displacement of the driver
is suitably matched by the driving amplifier power and the bass alignment.

Simply put if you model it the LT will need more amplifier power as long
as the driver can take it, noting the max SPL of any driver in terms of
excursion capability is completely independent of the box response.

More power essentially allows a smaller box to reach fixed SPL limits.

rgds, sreten.

An LT changes the box Q, its simply not true its still there.

 

[Aucosticraft]

Dear chris661 and sreten, thanks for the reply. Also wish you a happy new year in advance.

@chris661
I am using 8 Inch driver.
To handle excursion problem I have chosen the woofer with high excursion. also after simulation when I saw the its excursion at the f3 frequency, it was within the limit.

@sreten
I ll keep in mind the guidelines you provided. also i understand it don't change any of the box characteristic. I was trying to know that if two boxes designed with different f3. ( one lower and other higher) and then equilized system with LT to get same f3 of both, will sound same?

 

[chris661]

Here's some reading for you.
ESP - The Linkwitz Transform Circuit
Sounds like you've taken the displacement required into consideration. In that case, experiment with the circuit (try different Q of the response - some say Q=0.5 is better than Q=0.7).
Looking briefly at some sims, an 8" driver will need to move 9mm each way to reproduce 20Hz at 86dB. This wouldn't be audible, but will eat up a lot of headroom. I'd set a target for more like 30Hz, but that's just me...

Sreten, if you eq a response to be flat, does a resonance disappear? Then surely, a system with a resonance in it's frequency range could exhibit that resonance from time to time?

Chris

 

[Aucosticraft]

just few graph which i forgot to post.

@ Chris

I will read the link provided by you. thanks.

 

[Don Hills]

...
Sreten, if you eq a response to be flat, does a resonance disappear? Then surely, a system with a resonance in it's frequency range could exhibit that resonance from time to time?
Chris

No, EQing a resonance to flat makes it disappear, both in the frequency and time domain.

For example, take a 6 dB resonance with a Q of 0.7.
Starting from no signal, excite it with a signal at its resonant frequency and plot the level over time. It will rise from the original level to a level 6 dB higher, over a time period determined by the Q. (High Q = longer rise time.)

Now take a 6dB notch with a Q of 0.7 and repeat the excitation. The level will drop to a level 6 dB lower, over a time period determined by the Q.

Now put them in series and repeat. The rise in level over time of the peak will be exactly countered by the drop in level of the notch.

 

[AndrewT]

LT works at low levels of LF gain.
By the time gain gets to 20dB the speaker will be incapable of playing bass that can keep up (SPL wise) with the mid and treble.
At 15dB of LT gain you will be struggling to get a speaker to play well with this much signal being sent to it.
Try to limit your LT gain to 10dB or certainly less than 12dB.

The further you push the F-3dB frequency down, the more LT gain your speaker will need.

If you need 50W to play your Mid and Treble drivers which are both rated at 88dB/W @ 1m then +12dB of LT gain requires 800W to get the same SPL from an 88dB/W @ 1m bass driver.
If the bass driver is 85dB/W @ 1m the bass driver will need 1600W to match the 50W driving the Mid and Treble. I see that as a big problem.
Now get your self a high efficiency PA bass driver that cannot go down low but has lots of Power capability and lots of Sd.
Lets say you find a 96dB/W @ 1m driver.
Adding on the 10dB of LT gain to get the extension you need is equivalent to 86dB/W @ 1m.
Now you need just 80W to the Bass driver to match the SPL of the 50W to Mid and Treble.

 

[chris661]

No, EQing a resonance to flat makes it disappear, both in the frequency and time domain.

For example, take a 6 dB resonance with a Q of 0.7.
Starting from no signal, excite it with a signal at its resonant frequency and plot the level over time. It will rise from the original level to a level 6 dB higher, over a time period determined by the Q. (High Q = longer rise time.)

Now take a 6dB notch with a Q of 0.7 and repeat the excitation. The level will drop to a level 6 dB lower, over a time period determined by the Q.

Now put them in series and repeat. The rise in level over time of the peak will be exactly countered by the drop in level of the notch.

According to winISD, you'll still get phase shift around resonance. Anyway, in terms of level, you're right, but won't the system tend to ring at that frequency? This would show up on things like impulse measurements. Isn't a too-small box the cause of boomy bass, with lots of "overhang"? Even when eq is applied, the problem doesn't disappear.

Chris

 

[sreten]

Sreten, if you eq a response to be flat, does a resonance disappear? Then
surely, a system with a resonance in it's frequency range could exhibit
that resonance from time to time?

Chris

Hi,

One version of the LT has poles and zeros that entirely cancel
the real poles and zeros and then adds lower poles and zeros.

Though I do admit you have a point. The above is small signal, the
reality is large, and that fact might become apparent sometimes,
though the actual differences will still be relatively small to most.

It will be the real resonance affecting large signal behaviour, and that's
probably why some extreme active speakers never sounded that good.

rgds, sreten.

 

[Aucosticraft]

Here's some reading for you.
ESP - The Linkwitz Transform Circuit
Sounds like you've taken the displacement required into consideration. In that case, experiment with the circuit (try different Q of the response - some say Q=0.5 is better than Q=0.7).
Looking briefly at some sims, an 8" driver will need to move 9mm each way to reproduce 20Hz at 86dB. This wouldn't be audible, but will eat up a lot of headroom. I'd set a target for more like 30Hz, but that's just me...

Sreten, if you eq a response to be flat, does a resonance disappear? Then surely, a system with a resonance in it's frequency range could exhibit that resonance from time to time?

Chris

Chris, I went through the link, Although I had read the link some time b4 but it makes more sense reading it now. Thanks for the timely pointer. I took displacement into consideration because whatever SPL I can achieve is what I have to compromise with. I would further post my experiments here.

One thing we all realize that no design or topology is perfect answer when it comes to speaker design. one benefit is achieved at the cost of other.
So yes. this thread has helped me a lot to understand about LT. what you say if tuning to f3 @ 30Hz makes sense. for two reasons. first 20 hz can't be distinctly heard. second , Gaining some benefit of spl. also I think the very low frequencies but in audioable range makes bass sound boomy if room is not acoustically treated. If i am not mistaken the room gain typically is between 50 Hz to 20Hz for most of the home listening area.

But yes I would like to taste how it would sounds practical by setting f3 at different frequencies. But please tell me apart from technical glitches how does this sound like. i mean compared with vented design or something else.

And yes, cannot thank enough to all who has shared valuable experience enriching this thread. wish to hear more from you. Thanks.

 

[chris661]

Changing the f3 using a LT will give subjective results. You'll find a happy medium between driver distortion (running out of Xmax etc), and bass extension. You'll also find that some f3s will sound better in your room than others. Setting a shallower rolloff (Q=0.5) will probably help to integrate the subwoofer's rolloff with room gain, as (usually) the lower the frequency, the more room gain you get. Putting the subwoofer next to a wall will help re-inforce the bass, at the possible expense of boomy-ness. If it does sound boomy, play around with the f3 and Q of the system. This will help it to combine smoothly with the added gain of being put near a wall (quarter space instead of half space, being on the floor).
So, you'll find that certain settings will suit you and your room. I can't tell you what to use, though I can suggest different things.

You're welcome!
Chris

 

[Aucosticraft]

Yes I agree... LT is like using graphic equalizer and setting the frequency gain to attain lower f3. what would you say what should be the Xmax that should be considered while designing. I think it should be 75% of what is recommended by the mfg. cause while simulating it on WinIsd we see limit at that single frequency. but real music wave has lot of other smaller wavelength frequency riding it. what do you say?

 

[Henry8]

If you need 50W to play your Mid and Treble drivers which are both rated at 88dB/W @ 1m then +12dB of LT gain requires 800W to get the same SPL from an 88dB/W @ 1m bass driver.
If the bass driver is 85dB/W @ 1m the bass driver will need 1600W to match the 50W driving the Mid and Treble. I see that as a big problem.
.

This is a textbook quotation, (and this is the sort of stuff which may put people off using a LTX system) in the real world the level of signal in the music at say 30HZ is low, compared to 50Hz+, so even with boost levels of 12dB normal amp outputs (100-150W) are more than enough.
A more essential need is a warp filter if you use vinyl as a source.

 

[Henry8]

50liters of enclosure with f3 of 50Hz and design Qtc = 0.7 and implement Linkwitz Transform to get F3 of 20Hz.

A point to note: Lets assume you use an LTX to acheive -3db at 20Hz with a q of 0.707 (classic butterworth rolloff), but the actual response will not follow this curve at lower frequencies, as your power amps -3db point will now become significant (say even if its -3db is at 5Hz).

 

[AndrewT]

This is a textbook quotation, (and this is the sort of stuff which may put people off using a LTX system) in the real world the level of signal in the music at say 30HZ is low, compared to 50Hz+, so even with boost levels of 12dB normal amp outputs (100-150W) are more than enough.
A more essential need is a warp filter if you use vinyl as a source.

A point to note: Lets assume you use an LTX to acheive -3db at 20Hz with a q of 0.707 (classic butterworth rolloff), but the actual response will not follow this curve at lower frequencies, as your power amps -3db point will now become significant (say even if its -3db is at 5Hz).

I have never written a text book.
I do not have a text book on Linkwitz transforms nor on EQed speakers.

If you listen to enough different source material at the same volume level, then eventually you will find that the peak voltage across the whole frequency spectrum is the same, or near enough.
On that basis it is useless to say that 20Hz peak Voltage is always much less than 50Hz peak voltage. Don't delude your self.

On one hand you suggest a steep cut high pass filter to attenuate LF and then criticise an amplifier that rolls off early F-3dB ~5Hz.
Power amplifiers that need to reproduce wideband audio sources should extend to F-1dB ~2Hz to 4Hz. If a particular source needs a high pass filter than that should be applied before the signal arrives at the Power Amplifier.

 

[chris661]

Yes I agree... LT is like using graphic equalizer and setting the frequency gain to attain lower f3. what would you say what should be the Xmax that should be considered while designing. I think it should be 75% of what is recommended by the mfg. cause while simulating it on WinIsd we see limit at that single frequency. but real music wave has lot of other smaller wavelength frequency riding it. what do you say?

You're using the Linkwitz Transform on a subwoofer, so it won't see the higher frequencies, because it'll be crossed over. For excursion calculations, they can be safely ignored.
If I was you, I'd try to keep excursion minimal (a couple of mm p/p for normal levels - mine needed to be run at less than that, they had 3mm of Xmax), that way you'd have some headroom for when it gets turned up. Remember that +6dB required twice the excursion at the same frequency. An octave lower (half the frequency) requires 4x the excursion.
So, if you have a driver moving at 2mm p/p at 50Hz, for the same volume setting at 25Hz, you'd require 8mm of excursion. Now add a LT circuit to make 25Hz and 50Hz the same level (will require ~12dB of gain), you'd make it move another 4x, so 32mm of excursion will be required. Clearly, this must be approached with care.
Mechanical damage occurs very quickly, whereas thermal overload takes time. That said, don't discount power ratings altogether. Just bear in mind what you'll be putting the driver through.

Chris

PS - 30Hz with a Q of 0.7 or 40Hz with a Q of 0.5 would likely be good starting points.

 

[Lee in Montreal]

I was using for a while a Kef Kube 200. It is a Linkwitz transform device that changes the lower curve from about 80 to 20Hz. Great tool, but with limitations. It makes a good speaker sound very deep and low. But be carefull for what you are asking for. It comes at a price. Much harder on the driver as the correction curve is usually around 12-15db. Much longer cone excursion that will limit the overall SPL, and big amp required. Fine if you listen your music at volume not too high.

[IMGDEAD]http://sound.westhost.com/ltc-f4.gif[/IMGDEAD]

 

[sreten]

If you listen to enough different source material at the same volume level,
then eventually you will find that the peak voltage across the whole
frequency spectrum is the same, or near enough.
On that basis it is useless to say that 20Hz peak Voltage is always
much less than 50Hz peak voltage. Don't delude your self.

Hi, quite simply no you won't and only one person is deluded here, rgds, sreten.

Its your pet theory and its simply wrong, for all real recordings, as said before.
It does not stand up to any sensible and practical analysis of the real case.

 

[kipman725]

no Andrew is correct I can think of many tracks where the greatest amplitude signals are <50Hz I can also think of a few where the greatest amplitude signals are <20Hz necessitating the need for high pass filters on most systems or some kind of dynamic limiting. A 11Hz fundamental signal going though 12dB of gain from a circuit similar to a LT is not pretty in terms of cone excursion (cone slams into mechanical limits).