Minimum-phase filters are limited to 6dB/oct steps, but you can use shelwing filters to achieve what you are looking for here: just set the Q to some very low value and play with the cutoff frequency.
When using linear-phase filters you can enter any multiple to 1dB/oct value, and choose the cutoff frequency you want. It will naturally be -6dB at that frequency, and 0dB at DC (which is always faaaaar away with a log scale
). So for example if you want -6dB at 20kHz then choose that as the cutoff frequency, and then play with the slope to get it "start" where you want it to.
Having a straight line magnitude response starting at some defined frequency like you show in your screenshot could be implemented with linear-phase. I would rather have a more natural and softer knee, personally, but it is doable. It would be a "difficult" target to achieve with a limited number of taps though and the knee would always end up softer.
I could look into that in some future version.
When using linear-phase filters you can enter any multiple to 1dB/oct value, and choose the cutoff frequency you want. It will naturally be -6dB at that frequency, and 0dB at DC (which is always faaaaar away with a log scale
). So for example if you want -6dB at 20kHz then choose that as the cutoff frequency, and then play with the slope to get it "start" where you want it to.Having a straight line magnitude response starting at some defined frequency like you show in your screenshot could be implemented with linear-phase. I would rather have a more natural and softer knee, personally, but it is doable. It would be a "difficult" target to achieve with a limited number of taps though and the knee would always end up softer.
I could look into that in some future version.
Last edited:
Thanks,
It's not necessary,(to waste your time).
It can be done with min.phase EQ (with low shelving,this is a bit tricky).
-1dB/oct gives 3.32dB by decade.(log(10)/Log(2)).
So,for 2 decades,it gives 6.6dB
0.5 dB/oct would be ok.
No worry,it was a suggestion.
and talking/writing about FIR possibilities.
PS:when i talking 3dB/6dB slope leads to a misunderstanding.
I was talking to the end of audio band (-3dB or -6dB) for any "cuttoff" frequency.(slope is not"constant" whatever the corner frequency.
It's not necessary,(to waste your time).
It can be done with min.phase EQ (with low shelving,this is a bit tricky).
-1dB/oct gives 3.32dB by decade.(log(10)/Log(2)).
So,for 2 decades,it gives 6.6dB
0.5 dB/oct would be ok.
No worry,it was a suggestion.
and talking/writing about FIR possibilities.
PS:when i talking 3dB/6dB slope leads to a misunderstanding.
I was talking to the end of audio band (-3dB or -6dB) for any "cuttoff" frequency.(slope is not"constant" whatever the corner frequency.
Last edited:
Let’s say you have a high efficiency midbass in a half fiberglass half metal enclosure (car kick panel perhaps) no clue on the volume but a guess would be about .1cuft
And let’s say that enclosure rings really badly , like between 100-300hz somewhere
The ringing is so horrific it’s simply not usable.
So you cut a hole in the back of the enclosure and vent it to the outside of this hypothetical car. You drill a 3.5” hole and vent it to the outside. Some of the ringing is gone but some remains , you proceed to now make it a very leaky to inside of the car by drilling more holes in it. The ringing subsides for the most part and becomes usable to some degree, but still has a nasty bump that sounds hollow let’s say at 150hz
You’ve done what you can to make it work but you are not able to change the enclosure any more, it has to be installed and has to sound better.
You do some close mic measurements. With the speaker out (let’s say it’s. 2118h we all know that driver) it plays flat and rolls off at 200, your measurements show a nice smooth phase response and as it rolls off the phase goes the other way (naturally of course)
You put the speaker in the enclosure and remeasure, it looks mostly the same except the phase is much steeper. And at 80 has several wraps
Do you eq the response flat so the wraps go away because of the added gain and so it stays on the graph or is that going to mess up your next measurement to find out what phase to move to help that box ringing calm down........
I ask because if the ringing is a artifact of the enclosure and the ringing is so severe the measurements are hard to read.
How to determine that correctly?
Thanks
And let’s say that enclosure rings really badly , like between 100-300hz somewhere
The ringing is so horrific it’s simply not usable.
So you cut a hole in the back of the enclosure and vent it to the outside of this hypothetical car. You drill a 3.5” hole and vent it to the outside. Some of the ringing is gone but some remains , you proceed to now make it a very leaky to inside of the car by drilling more holes in it. The ringing subsides for the most part and becomes usable to some degree, but still has a nasty bump that sounds hollow let’s say at 150hz
You’ve done what you can to make it work but you are not able to change the enclosure any more, it has to be installed and has to sound better.
You do some close mic measurements. With the speaker out (let’s say it’s. 2118h we all know that driver) it plays flat and rolls off at 200, your measurements show a nice smooth phase response and as it rolls off the phase goes the other way (naturally of course)
You put the speaker in the enclosure and remeasure, it looks mostly the same except the phase is much steeper. And at 80 has several wraps
Do you eq the response flat so the wraps go away because of the added gain and so it stays on the graph or is that going to mess up your next measurement to find out what phase to move to help that box ringing calm down........
I ask because if the ringing is a artifact of the enclosure and the ringing is so severe the measurements are hard to read.
How to determine that correctly?
Thanks
Last edited:
Well I figured it out by trail and error
I used a combination of linear phase eq and minimum phase eq
Mostly linear phase eq
Which is wierd , I tryed linear phase eq before and it sounded echoed
However , the kick panel pod sounds echoed, so it makes sense to me now why
Phase and mag measurement is much better.
I can do a close measurement and see the crossover and the phase no more wrapping and doing wierd stuff
Oh and I also had my axis off was zoomed in way too much
But now even zoomed it’s looking good
Listening to it the harmonics sound so much better and natural.
Wow I love rephase even more now. What an amazing tool!
(slope is not "constant" whatever the corner frequency.).Depending of the choice of the "tilt"
In fact,-3dB (or -6dB) at 20 KHz in every case.
Purpose is to "finish" at -3dB(or -6dB) with any sloping region.
I like the idea and will definitely look into implementing such a feature in a future version
Thanks pos
The issue is the cavity walls. There so close to the basket and the air volume is tiny. Even vented to the outside it adds a echoed ringing
I’ve done all the dampeners (wouldn’t have posted this if I haven’t exhaust all efforts.)
Trying to understand what makes this happen.
Go mount a midrange in a coffee can and put dampeners on the coffee can , it will still sound like it’s in a can. Dampeners lower the resonance frequency by adding mass. At some point it’s still in a tiny little chamber, the only option left is remove the mount and run it as a baffle-less dipole (which actually sounds better than this I’ve tried it).
I got the measurements to look better, at least I can do something with them.
I was wondering what is the norm for how a measurement should look when a speaker is in way too small enclosure. What happens to the speaker and what can typically be done in dsp/fir to mitigate some of its effects.
I’m kinda hoping for some ideas I haven’t thought of. I’ve read everything I can about it. I think I’m past that part and am venturing into uncharted territory with fir in car. So the most info I can possibly get to get me thinking the right direction is invaluable to me and so much appreciated!
Danke schoen.
The issue is the cavity walls. There so close to the basket and the air volume is tiny. Even vented to the outside it adds a echoed ringing
I’ve done all the dampeners (wouldn’t have posted this if I haven’t exhaust all efforts.)
Trying to understand what makes this happen.
Go mount a midrange in a coffee can and put dampeners on the coffee can , it will still sound like it’s in a can. Dampeners lower the resonance frequency by adding mass. At some point it’s still in a tiny little chamber, the only option left is remove the mount and run it as a baffle-less dipole (which actually sounds better than this I’ve tried it).
I got the measurements to look better, at least I can do something with them.
I was wondering what is the norm for how a measurement should look when a speaker is in way too small enclosure. What happens to the speaker and what can typically be done in dsp/fir to mitigate some of its effects.
I’m kinda hoping for some ideas I haven’t thought of. I’ve read everything I can about it. I think I’m past that part and am venturing into uncharted territory with fir in car. So the most info I can possibly get to get me thinking the right direction is invaluable to me and so much appreciated!
Danke schoen.
So no doubt you are getting reflections from the can coming back through the cone. Call it can honk, a relative of horn honk, which has been diagnosed as mouth reflections. Gunness filtering (google is your friend) would be an elegant solution but its shall we say challenging for a DIYer. If Pos would help, it might approach feasibility. The basic idea is to cancel the reflections by injecting a delayed copy of the signal into the input of equal magnitude and opposite polarity. Easy to say, hard to do.
As you shrink the size of the metal can you approach a sealed back midrange. Perhaps you should try to seal the back of your midrange. This will minimize the delay of the reflections that come back through the cone so that the frequencies at which they interfere are out of band.
As you shrink the size of the metal can you approach a sealed back midrange. Perhaps you should try to seal the back of your midrange. This will minimize the delay of the reflections that come back through the cone so that the frequencies at which they interfere are out of band.
Last edited:
Thanks nc535
Okay okay I’m listening.
Yes this is exactly what it seems to be doing now that you put it like that.
I never really thought of it like that before, but yes that’s is good.
I could easily (if it’s easier and less messy on the signal ) put another small let’s say 4” sealed back midrange in side the kickpod with the other speaker. Play it in reverse and use dsp......
I would have to drill a hole to put mic go to inside of pod to get the FR data
But could tune the other midrange inside of the pod to cancel the back sound.
Bad idea?
.....I don’t know what to say. Thank you!
I will definitely be searching and studying about making Gunness filters. This is great tho
I know what to at least consider now
Okay okay I’m listening.
Yes this is exactly what it seems to be doing now that you put it like that.
I never really thought of it like that before, but yes that’s is good.
I could easily (if it’s easier and less messy on the signal ) put another small let’s say 4” sealed back midrange in side the kickpod with the other speaker. Play it in reverse and use dsp......
I would have to drill a hole to put mic go to inside of pod to get the FR data
But could tune the other midrange inside of the pod to cancel the back sound.
Bad idea?
.....I don’t know what to say. Thank you!
I will definitely be searching and studying about making Gunness filters. This is great tho
I know what to at least consider now
Uhmm, I think auto linearization of phase, if alone, would be really bad,
as in REALLY bad
Auto magnitude, more OK.
Just in case you don't know, rePhase will import REW's auto EQs.
IMHO, the beauty of rePhase, apart from being donate-ware, is that it IS manual...makes us learn what we are doing and be responsible for our corrections ......before maybe moving on to more auto type adjustments.
Dirac , (which is “auto”) mainly If auto is what you want maybe.
I prefer rephase is I had to pick one over the other
I’ve have excellent results with Dirac and rephase together , using rephase to do my crossovers and any horn related issues , and Dirac to do my left and right matching
And than a fine tune with rephase.
Soon I’m getting off the Dirac and going strictly rephase.
I’m waiting for parts from mini to do a pair of opendrc 2x2 s into a pair of 2x4hds.
I used a Dirac as a aid to teach me what fir can sound like
But actually once I started just making trial and error measurements and filters in rephase I could learn what is happening enough to limp around.
I get stumped sometimes , it’s learning progressions , I just this week had a really big advancement in my learning. I was doing something wrong for a while , now that I got it under control I had to re do every fir but it’s excellent now.
On another note
I added a 2” speaker in my kick pod, playing it in reverse and sealed the basket with 100 layers of electrical tape and a 25w 47ohm resistor on the 2” to lower the power quite a bit
Put the mic in the pod took some measurements of the 8 and the 2
Got the 2” to cancel the ringing and horn loading fairly good now. I just tryed matching the difference of the 8 in and out of the pod. And reversed it in polarity , all done in rephase! Worked so good. I can’t believe. Had to make a few filters till I felt it was nailed down
The 8 measurements are soooooo much better now.
The 8 is a tiny bit quieter but it should be , the ringing is gone. Lot better
Better depth , it’s working. Now to prefect this.
This is a very good idea. I am so surprised how well it works
That hollow sound is gone the echoing at 800 is reduced so much.
Extremely pleased. Thank you guys for the help
You can see the 2” at the bottom inside the kick pod
I prefer rephase is I had to pick one over the other
I’ve have excellent results with Dirac and rephase together , using rephase to do my crossovers and any horn related issues , and Dirac to do my left and right matching
And than a fine tune with rephase.
Soon I’m getting off the Dirac and going strictly rephase.
I’m waiting for parts from mini to do a pair of opendrc 2x2 s into a pair of 2x4hds.
I used a Dirac as a aid to teach me what fir can sound like
But actually once I started just making trial and error measurements and filters in rephase I could learn what is happening enough to limp around.
I get stumped sometimes , it’s learning progressions , I just this week had a really big advancement in my learning. I was doing something wrong for a while , now that I got it under control I had to re do every fir but it’s excellent now.
On another note
I added a 2” speaker in my kick pod, playing it in reverse and sealed the basket with 100 layers of electrical tape and a 25w 47ohm resistor on the 2” to lower the power quite a bit
Put the mic in the pod took some measurements of the 8 and the 2
Got the 2” to cancel the ringing and horn loading fairly good now. I just tryed matching the difference of the 8 in and out of the pod. And reversed it in polarity , all done in rephase! Worked so good. I can’t believe. Had to make a few filters till I felt it was nailed down
The 8 measurements are soooooo much better now.
The 8 is a tiny bit quieter but it should be , the ringing is gone. Lot better
Better depth , it’s working. Now to prefect this.
This is a very good idea. I am so surprised how well it works
That hollow sound is gone the echoing at 800 is reduced so much.
Extremely pleased. Thank you guys for the help
You can see the 2” at the bottom inside the kick pod
Last edited:
Import of settings from REW auto eq is only a partial workaround for lack of the feature in rephase for several reasons. REW refuses to fill any dip below the target line then complains when you set the target low enough that it doesn't need to. REW lacks flexibility in target generation - its only aimed at room eq, not XO generation. Rew limits resources applied to the eq and seems aimed at IIR filters rather than FIR.
That target line is REWs major benefit re' filters. But if I'm going to have to go into rephase and manually correct for REW's short comings as well as to linearize the phase, then why not do it all in rephase to begin with?
Thus my feature request: can you give us the ability to import a target response (in addition to the existing ability to import a measurement) that acts as a visual aid during manual equalization?
That target line is REWs major benefit re' filters. But if I'm going to have to go into rephase and manually correct for REW's short comings as well as to linearize the phase, then why not do it all in rephase to begin with?
Thus my feature request: can you give us the ability to import a target response (in addition to the existing ability to import a measurement) that acts as a visual aid during manual equalization?
Filling of dips depends on your settings for individual and overall filter boost. Generally a bad idea nonetheless.
REW warns you if much of the response is above the target. Not sure how that constitutes a complaint.
REW has a speaker driver target setting that allows selection of Bessel, Butterworth or Linkwitz-Riley lowpass and high pass filters up to 8th order. It also allows a "house curve" to be added to any of the targets that can be any shape you like and is defined by a simple text file.
Yes it is.
In meantime there is below trick for stop bands, a "COMPENSATE" HP filter LR 4th order at 100Hz is set and a "COMPENSATE" LP filter BW 2nd order at 25kHz is set which should be close to target for your new line arrays, now it should work import a measurement and EQ everything flat as a pancake, when that EQ work is done disable the two compensate filters and what stands back is the target curve.
Attachments
clever, perhaps too much so I'd almost rather draw my target by hand on an overlay for the screen.
Just posted latest results done the hard way in my line array thread
https://www.diyaudio.com/forums/full-range/337956-range-line-array-wall-corner-placement-35.html#post5872856
manual eq to get in the ballpark and implement crossover to subs in MiniDSP
rew-rephase auto-eq to refine and apply house curve and balance channels
rephase phase flattening
several times around that loop to get it as close to right as it is
I'm not sure there is an easy way; probably be a lot easier for me next time though.
I'd almost rather draw my target by hand on an overlay for the screen. Just posted latest results done the hard way in my line array thread
https://www.diyaudio.com/forums/full-range/337956-range-line-array-wall-corner-placement-35.html#post5872856
manual eq to get in the ballpark and implement crossover to subs in MiniDSP
rew-rephase auto-eq to refine and apply house curve and balance channels
rephase phase flattening
several times around that loop to get it as close to right as it is
I'm not sure there is an easy way; probably be a lot easier for me next time though.
On another note
I added a 2” speaker in my kick pod, playing it in reverse and sealed the basket with 100 layers of electrical tape and a 25w 47ohm resistor on the 2” to lower the power quite a bit
Put the mic in the pod took some measurements of the 8 and the 2
Got the 2” to cancel the ringing and horn loading fairly good now. I just tryed matching the difference of the 8 in and out of the pod. And reversed it in polarity , all done in rephase! Worked so good. I can’t believe. Had to make a few filters till I felt it was nailed down
The 8 measurements are soooooo much better now.
The 8 is a tiny bit quieter but it should be , the ringing is gone. Lot better
Better depth , it’s working. Now to prefect this.
This is a very good idea. I am so surprised how well it works
That hollow sound is gone the echoing at 800 is reduced so much.
Extremely pleased. Thank you guys for the help
You can see the 2” at the bottom inside the kick pod
This is very interesting
Can you draw a picture of this setup with responces and corrections?
signal - rew - amp - 2"
- rew - amp - 8"
As I understand it the 2" is cancelling the reflections at about 800Hz?
Maybe in an own thread. I know it is a lot of work, but the solution is fantastic!