Compared to the wavelength at low bass frequencies, it's generally negligible.
Yes, I can't give you a comprehensive answer but this thread shows an analog shelf filter, but this thread shows one with a 10dB boost at 90Hz whereas we're talking about a 10dB cut at 20Hz. It's essentially the same thing.
https://www.diyaudio.com/community/...eqing-transmission-lines-and-the-like.310150/
I'm guessing all you need to do is change the values.
Since I meant a reflex box it would be necessary to equalise two pole-pairs to a given target function. So two biquad functions à la Linkwitz are needed or one fourth oder EQ (like a 4th order state-variable filter for instance). Unless one existing pole can be used as-is.
I would never ever attempt to turn a closed box into a fifth order highpass. I am sometimes crazy but not THAT crazy.
Fifth order instead of fourth would just add some subsonic protection ti a reflex box.
Regards
Charles
Oh right, sorry I misunderstood.Since I meant a reflex box it would be necessary to equalise two pole-pairs to a given target function. So two biquad functions à la Linkwitz are needed or one fourth oder EQ (like a 4th order state-variable filter for instance). Unless one existing pole can be used as-is.
I would never ever attempt to turn a closed box into a fifth order highpass. I am sometimes crazy but not THAT crazy.
Fifth order instead of fourth would just add some subsonic protection ti a reflex box.
Regards
Charles
It can be done on a similar fashion.
But like I said, it only works well on paper.
Practically speaking a BR system will have a much bigger error in its response.
This happens because of the non-linear BL(x) and Cms(x) and the fact that it's an higher order system.
So there won't be much left from that beautiful response.
But yeah adding an additional HP filter helps at least with the cone excursion.
Or when you can tune low enough, you can get a response that is kinda like a closed box + HP with a low Q.
Shelf filter seems to work fine for sealed too.
Here's same driver in same size box, but sealed now and using a 19Hz, 2.1Q, -19dB shelf. And is the only filter being used.
It's the power screen on the right. The left is the ported version previously posted, for comparison.
So sealed cost about 7dB for close to the same f-3.
Excursion and group delay for sealed.
I'm getting more and more convinced, that when bottom end acoustic rolloff is the same between sealed vs ported...well, the dang subs are the same.
The size/width of the knee is as important as the rolloff below the knee, I think.
When knees and response below match perfectly, sealed will have no more of an advantage with group delay than ported.
I mean heck...acoustic response is acoustic response, right?
Also seems to me, whatever filters and techniques we use to get to any given acoustic response....shelving, high-pass, PEQs, Linkwitz transform...the combination of any or all of those filters.....
........they all end up after being summed together to be the same NET filter in the end.
Seems it just comes down to using whatever filter resources we have,
and don't fret over the idea that a LT transform is gonna work better than rolling your own with named filters, for example.
Here's same driver in same size box, but sealed now and using a 19Hz, 2.1Q, -19dB shelf. And is the only filter being used.
It's the power screen on the right. The left is the ported version previously posted, for comparison.
So sealed cost about 7dB for close to the same f-3.
Excursion and group delay for sealed.
I'm getting more and more convinced, that when bottom end acoustic rolloff is the same between sealed vs ported...well, the dang subs are the same.
The size/width of the knee is as important as the rolloff below the knee, I think.
When knees and response below match perfectly, sealed will have no more of an advantage with group delay than ported.
I mean heck...acoustic response is acoustic response, right?
Also seems to me, whatever filters and techniques we use to get to any given acoustic response....shelving, high-pass, PEQs, Linkwitz transform...the combination of any or all of those filters.....
........they all end up after being summed together to be the same NET filter in the end.
Seems it just comes down to using whatever filter resources we have,
and don't fret over the idea that a LT transform is gonna work better than rolling your own with named filters, for example.
Last edited:
More or less, it can get quite convoluted, and sometimes you have to keep in mind that you don't clip a certain stage.
(especially with analog active filter, most DSP's have about 24dB headroom).
Also, you can kind of mimic certain filters with other types.
For example use a param EQ as a HP filter, with the added benefit that the groupdelay won't be as bad.
Well VituixCAD for example, has a IIR block where it can find just the right set of parameters.
Which is even better and easier!
Does it give out biquads or a set of specific filters (like REW's auto EQ)?Well VituixCAD for example, has a IIR block where it can find just the right set of parameters.
Which is even better and easier!
I HATE biquads lol.
It's the same thing?
I guess unless you wanna work with FIR filters, but IIR filters always come in blocks (aka biquads)
I have no idea what REW has, barely use that program.
I meant biquad vs a named filter .......like a PEQ for example: 100Hz, +3dB , Q=2
I realize that named filter is converted into a biquad by the dsp software/platform. I just don't like to have to fool with a biquad string itself.
And yep, FIR is all I'll use anymore...other than maybe for system high-pass.
The sum is minimum phase but the no delay?
The box compliance and port mass behave as bandpass filter working in parallel with the filter representing the driver which are summed external to the box.
Below box tuning the output of the port approaches zero delay, above tuning the box compliance and port mass invert the phase.
To achieve this the box first has to compress to then get the mass moving which takes time, once moving the box has to decompress before it can act on the mass to reverse its direction of movement. The output of the port is therefore delayed.
So at and above port tuning and starting at t=0, the first half cycle is output from the driver only, the port has not had time to get moving.
The implication is that the initial half cycle will produce less SPL overall at and above port tuning than the following cycles (steady state signal).
Ok, will rephrase that, the amplitude peak of the port output is delayed with respect to the drivers amplitude peak output at and above resonance
This may affect the transient performance of the reflex system as opposed to a sealed or front loaded horn system, ie. the impact felt on the body from the wave front effectively being compressed in output compared to a closed rear chamber system.
This may affect the transient performance of the reflex system as opposed to a sealed or front loaded horn system, ie. the impact felt on the body from the wave front effectively being compressed in output compared to a closed rear chamber system.
Maybe just call it group delay, that makes it even easier.
Although that's a little obvious, any higher order system always adds group delay. (aka phase shift)
In other words, you will have the same amount of group delay with a closed box system that's modified with a bunch of (high-pass) filters to get the same response.
Perry,
As you say, you can’t get around the one cycle phase/time delay of a bass reflex (phase inversion) system, a 25Hz Fb BR at minimum introduces ~ 40ms delay at Fb compared to a sealed box response, as can be seen in Josh Ricci’s measurements of sealed and BR cabinets using the same B&C 21SW152 drivers:
https://data-bass.com/#/systems/5b11bfe651412e00047d6693?_k=re5awp
By 50 Hz the group delay of the BR has dropped to near that of the sealed box.
Adding a 25Hz BW24 filter then adds another 15ms at Fb, the group delay still dropping to that of the sealed system by ~ 50Hz.
You say that the virtue of the shelf filter is it has 25-33% of the phase shift the BW24 filter adds at Fb , so could reduce the 15ms delay to ~ 3.75ms.
So, using the “least compromised” shelf filter (which won’t provide the out of band protection of a BW24…) the 75ms group delay could perhaps be dropped to 63.75ms, “a different date”
.I’m glad that a tiny fraction of one cycle difference at 25 Hz does not sound like a “a different date” to me, though do wish my hearing and sense of timing was better...
Art
Hi Art,
This is the "back of napkin" way I think about it:
BR is 24dB/octave or 4th order. 180 degrees out of phase at cutoff. at 20Hz that is 25msec.
Adding a 4th order high pass protection makes it 8th order. 360 degrees out of phase at cutoff. At 20Hz that is 50msec
Using a shelf instead makes it 6th order. 270 degrees out of phase at cutoff. (Give or take a few.) At 20Hz that is 37.5msec.
So that cuts the phase delay by 25%. I think that's a meaningful improvement. At very low cutoff frequencies, that's a bare minimum of 10ms improvement.
And the 24dB filter in the real world does not always give better protection, because that steep jag in a Q=2.5 or Q=3 shelf filter gives you a larger drop where it matters the most which is just below driver resonance. The amount of music material in the lowest part of the shelf where it flattens out is usually trivial. A 16dB drop in that range is more than enough.
This is a 10dB shelf, Q = 3, notice the drop between 25Hz and 13Hz is 24dB, and flattening out at -10dB at 5Hz and below is just enough to do the job. In a home stereo there's no music material that low. Probably much different in a PA system, where microphone pops might go down to DC :^>
This is the "back of napkin" way I think about it:
BR is 24dB/octave or 4th order. 180 degrees out of phase at cutoff. at 20Hz that is 25msec.
Adding a 4th order high pass protection makes it 8th order. 360 degrees out of phase at cutoff. At 20Hz that is 50msec
Using a shelf instead makes it 6th order. 270 degrees out of phase at cutoff. (Give or take a few.) At 20Hz that is 37.5msec.
So that cuts the phase delay by 25%. I think that's a meaningful improvement. At very low cutoff frequencies, that's a bare minimum of 10ms improvement.
And the 24dB filter in the real world does not always give better protection, because that steep jag in a Q=2.5 or Q=3 shelf filter gives you a larger drop where it matters the most which is just below driver resonance. The amount of music material in the lowest part of the shelf where it flattens out is usually trivial. A 16dB drop in that range is more than enough.
This is a 10dB shelf, Q = 3, notice the drop between 25Hz and 13Hz is 24dB, and flattening out at -10dB at 5Hz and below is just enough to do the job. In a home stereo there's no music material that low. Probably much different in a PA system, where microphone pops might go down to DC :^>