Okay Jreave#
Here is my second attempt. I now did the measurments:
I completely understand the physics in play here. I just didn't know how to measure it. But this technique makes sense.
You are right, without crossovers on the drivers the two curves align almost perfectly! See below:
I also tried again with crossovers. This is not as pretty, but again the best result seem to be around the same value as without crossovers above.
So, 0,38 inch = 9,65 mm offset of mid behind tweeter. I think this makes snese on my 4 degree titlted baffle. I also compared the summation with this 9 mm delay to other values I tried before and this is deffinitively the best summation!
(Crossed at 2000 Hz. I don't know what causes that phase jump at 3,5kHz and 7kHz. The 7 k one is also there on the measurment of the tweeter alone.)
Thanks 😀
Next up, same exercise for the woofers! 🙂
Here is my second attempt. I now did the measurments:
- From 300 Hz without crossovers.
- 1 m distance between mid and tweeter
- on axis on drivers
- 4 ms gating - first reflection at 4,4 ms
I completely understand the physics in play here. I just didn't know how to measure it. But this technique makes sense.
You are right, without crossovers on the drivers the two curves align almost perfectly! See below:

I also tried again with crossovers. This is not as pretty, but again the best result seem to be around the same value as without crossovers above.

So, 0,38 inch = 9,65 mm offset of mid behind tweeter. I think this makes snese on my 4 degree titlted baffle. I also compared the summation with this 9 mm delay to other values I tried before and this is deffinitively the best summation!

(Crossed at 2000 Hz. I don't know what causes that phase jump at 3,5kHz and 7kHz. The 7 k one is also there on the measurment of the tweeter alone.)
Thanks 😀
Next up, same exercise for the woofers! 🙂
Last edited:
digitalthor#
Actually my midrange driver has higher sensitivity (91,5 dB) than my tweeter (90,4 dB). So I think the gain difference of the two amplifires is around 4,5 dB. That would probably normaly result in almost no gain difference due to a naturaly more sensitive tweeter as you say.
Juhazi#
I know that equalization of the drivers is important for proper summation, but for the cause of determining accoustical offset it shouldn't matter. I also realized that the cone breakup of the MW19P at 3kHz which I spend a lot of time trying to reduce may not really be an issue with a crossover of 2kHz or below. So i should learn not to overdoo it!
Okay thanks for the graph. Is there any predefined time between peaks that are supposed to be right?
I am wondering what my DSP can contribute with here or not. My test show that I have a accoustical offset of 9 mm. But the tweeter peak is much earlier than the midrange peak (as on your graph). Would it be better or worse/"wrong" to delay the tweeter one or more full wavelengths(at crossover) to align the two peaks?
Actually my midrange driver has higher sensitivity (91,5 dB) than my tweeter (90,4 dB). So I think the gain difference of the two amplifires is around 4,5 dB. That would probably normaly result in almost no gain difference due to a naturaly more sensitive tweeter as you say.
Juhazi#
I know that equalization of the drivers is important for proper summation, but for the cause of determining accoustical offset it shouldn't matter. I also realized that the cone breakup of the MW19P at 3kHz which I spend a lot of time trying to reduce may not really be an issue with a crossover of 2kHz or below. So i should learn not to overdoo it!
Okay thanks for the graph. Is there any predefined time between peaks that are supposed to be right?
I am wondering what my DSP can contribute with here or not. My test show that I have a accoustical offset of 9 mm. But the tweeter peak is much earlier than the midrange peak (as on your graph). Would it be better or worse/"wrong" to delay the tweeter one or more full wavelengths(at crossover) to align the two peaks?
Time difference between peaks in step response depends on xo-frequency/wavelength of a cycle - I suppose there is a formula... LR2 difference is half cycle (peaks to opposite direction), LR4 full cycle. This is how John Atkinson tells it.
SPL curve and phase go hand-in hand. Ideal crossover supposes that phase is constant unless xo curve changes it. Phase error harms summation of signals, but I don't know how much it affects step response/timing too.
Main reason for doing eq before xo is to make eq easier by following straight line.... Other benefit is that it makes easier to change xo order or frequency after that. Doing eq first, based on measurement, takes care of baffle step compensation too, automatically.
SPL curve and phase go hand-in hand. Ideal crossover supposes that phase is constant unless xo curve changes it. Phase error harms summation of signals, but I don't know how much it affects step response/timing too.
Main reason for doing eq before xo is to make eq easier by following straight line.... Other benefit is that it makes easier to change xo order or frequency after that. Doing eq first, based on measurement, takes care of baffle step compensation too, automatically.
Last edited:

I remember the 1st time I did this for myself and got it right - I was just amazed at how well it actually worked. Who knew science could be so cool?
I've noticed those phase aberrations in REW as well and I don't know the cause. But it's part of the reason that I go elsewhere to extract minimum phase from my frd files.
Juhazi#
True.. it is easier to align a straight line, this was also why i raised the tweeter for theese measurments. I just got afraid of dooing more harm than good when equalising my drivers. If you look at this measurement of my raw midrange driver gated at 4 ms then what would you do?
My baffle is 225mm wide so i would put a bafflestep correction at around 500Hz which seems to match the measurment. But what else? would you do anything to dip from 2k to 3k and the the mayhem above 3k (crossing between 1700 and 2000)? Or rather not?
In the low end i would use a LR-transform to flatten it to 50hz (crossing between 100 and 200 Hz).
True.. it is easier to align a straight line, this was also why i raised the tweeter for theese measurments. I just got afraid of dooing more harm than good when equalising my drivers. If you look at this measurement of my raw midrange driver gated at 4 ms then what would you do?

My baffle is 225mm wide so i would put a bafflestep correction at around 500Hz which seems to match the measurment. But what else? would you do anything to dip from 2k to 3k and the the mayhem above 3k (crossing between 1700 and 2000)? Or rather not?
In the low end i would use a LR-transform to flatten it to 50hz (crossing between 100 and 200 Hz).
#Jreave
Thanks for teh guidance, then we will see if it is just as easy with the side firing woofer :S
Sience IS cool! Especially when you understand it 😛
Hmmmm so a software issue maybe?
Thanks for teh guidance, then we will see if it is just as easy with the side firing woofer :S
Sience IS cool! Especially when you understand it 😛
Hmmmm so a software issue maybe?
^^
- For BSC set Shelf2 at 400Hz for +5dB.
- Another Shelf2 around 4kHz different direction, Q of 3 or 4 and gain -10dB. This is not so critical to get straight because the driver is beaming and this is quite far out of passband.
- BoostCut around 1khz, Q 1, -2dB This is very delicate range, small changes can be heard!
- Lowest tail eq carefully with Shelf or LT - do it later on if needed, measurement must have IR gating of at least 20ms
- Gain -6dB (in left column) I don't think it is a good idea to use +5,5dB unity gain for tweeter!
- Delay by simulation for a start (set as microseconds, shown also as mm)
This is just guessing, take new measurement after uploading the settings. Then set new parameters or add filters, upload and measure again etc. When it looks fine, save settings!
This is my MR183w midrange settings
- BSC shown, at 850Hz
- Boost/cut at 135Hz, below passband to attenuate peak at driver's resonance
- Highpass2 at 300Hz
- Lowpass2 at 3000Hz
- BoostCut at 6500Hz to tame cone resonance above passband
Because I use LR2 filters, it is very important to do precise eq one octave past the xo point - xo 300/3000Hz means straight response from 150Hz to 6000Hz, which is a real challenge and I don't quite reach that with these settings! With higher order xo this is easier and mid's and tweeter's low end won't get so stressed (lower distortion)
- For BSC set Shelf2 at 400Hz for +5dB.
- Another Shelf2 around 4kHz different direction, Q of 3 or 4 and gain -10dB. This is not so critical to get straight because the driver is beaming and this is quite far out of passband.
- BoostCut around 1khz, Q 1, -2dB This is very delicate range, small changes can be heard!
- Lowest tail eq carefully with Shelf or LT - do it later on if needed, measurement must have IR gating of at least 20ms
- Gain -6dB (in left column) I don't think it is a good idea to use +5,5dB unity gain for tweeter!
- Delay by simulation for a start (set as microseconds, shown also as mm)
This is just guessing, take new measurement after uploading the settings. Then set new parameters or add filters, upload and measure again etc. When it looks fine, save settings!
This is my MR183w midrange settings
- BSC shown, at 850Hz
- Boost/cut at 135Hz, below passband to attenuate peak at driver's resonance
- Highpass2 at 300Hz
- Lowpass2 at 3000Hz
- BoostCut at 6500Hz to tame cone resonance above passband
Because I use LR2 filters, it is very important to do precise eq one octave past the xo point - xo 300/3000Hz means straight response from 150Hz to 6000Hz, which is a real challenge and I don't quite reach that with these settings! With higher order xo this is easier and mid's and tweeter's low end won't get so stressed (lower distortion)
Attachments
Last edited:
I don't know what causes that phase jump at 3,5kHz and 7kHz.
I've noticed those phase aberrations in REW as well and I don't know the cause.
That's just the display wrapping once the phase barely passes -180, isn't it? IOW, the phase graph is actually a large U.
Juhazi#
Thanks for your input! I gave it a try.
The raw midrange response:
The response with your suggested EQ settings:
I made some adjustments:
The total EQ is now:
Summation of tweeter and driver with the 9.5mm delay is pretty nice i think!:
Did I miss anything?
Thanks for your input! I gave it a try.
The raw midrange response:
The response with your suggested EQ settings:
I made some adjustments:

The total EQ is now:
- +5dB low shelf at 400Hz
- High shelf at 4kHz with Q of 2 and gain -6dB
- -1dB cut at 880Hz with Q of 3
- -2dB cut at 1700Hz with Q of 1
- -2.5dB cut at 3800Hz with Q of 9
- (And a low shelf boost of +6dB at 55Hz to extend it flat to 40Hz without crossover)
- LR4 at 150Hz and 2000Hz

Summation of tweeter and driver with the 9.5mm delay is pretty nice i think!:

Did I miss anything?
HammerSandwich#
Yes it is just making that vertical line and shift to make the phase more readable. But why is there a phase shift there? There is no crossover there. What else can cause the phase shift?
Yes it is just making that vertical line and shift to make the phase more readable. But why is there a phase shift there? There is no crossover there. What else can cause the phase shift?
Great! 😀
Yeah tell me about it 😛
How would you do it?
So far I did near field measurments of my midrange to get the desired slope downwards.
This was the EQed midrange with its natural low end rolloff:
Then I added a low shelf as described to extend it to 40Hz:
Then the theoretical LR4 filter gives exactly what you would expect:
I then did the same on the woofer.
In near field everything is perfect. But is that a valid way of doing it?
How would you go about this? Gating cannot be used any longer right?
Yeah tell me about it 😛
How would you do it?
So far I did near field measurments of my midrange to get the desired slope downwards.
This was the EQed midrange with its natural low end rolloff:

Then I added a low shelf as described to extend it to 40Hz:

Then the theoretical LR4 filter gives exactly what you would expect:

I then did the same on the woofer.
In near field everything is perfect. But is that a valid way of doing it?
How would you go about this? Gating cannot be used any longer right?
Hmm, double side-firing woofers near floor.
First I would take nearfiel measurement of a single driver to see how it performs. But this response is almost useless to set eq and xo.
Next is to let the speaker stand on the floor like it is meant to be. Put it in the middle of a large room, place mic on the floor in front of it at 1m distance. Open all doors from the room, then sweep, look at response with different IR gating (6-500ms) and smoothing to find out where reflections and modes are. move the mic and look at response again, what cahges aare there? Based on this start setting eq. You should not try to eq room mode bumps and dips! Then set xo and estimated delay.
Now the tricky part begins, move the mic up so that it has equal distance to mid and woofers. Now play W+M and try to get spl levels and step resoponse/timing right.
After this, put the speaker on the intended position in your room and place mic at 90cm from floor near your listenig sweet spot. Mic should be at least 1m from back wall! Listen to it first with music, is it basically right? Then measure, use different mic locations to see reflections and modes. Then fine-tune eq and check step may times.
Right delay is difficult to find, in small rooms wall reflections easily corrupt step response calculation ( a reflected sound might be louder that straight sound). Listen to again and repeat everything next day! For me this took 2 weeks around Christmas...

First I would take nearfiel measurement of a single driver to see how it performs. But this response is almost useless to set eq and xo.
Next is to let the speaker stand on the floor like it is meant to be. Put it in the middle of a large room, place mic on the floor in front of it at 1m distance. Open all doors from the room, then sweep, look at response with different IR gating (6-500ms) and smoothing to find out where reflections and modes are. move the mic and look at response again, what cahges aare there? Based on this start setting eq. You should not try to eq room mode bumps and dips! Then set xo and estimated delay.
Now the tricky part begins, move the mic up so that it has equal distance to mid and woofers. Now play W+M and try to get spl levels and step resoponse/timing right.
After this, put the speaker on the intended position in your room and place mic at 90cm from floor near your listenig sweet spot. Mic should be at least 1m from back wall! Listen to it first with music, is it basically right? Then measure, use different mic locations to see reflections and modes. Then fine-tune eq and check step may times.
Right delay is difficult to find, in small rooms wall reflections easily corrupt step response calculation ( a reflected sound might be louder that straight sound). Listen to again and repeat everything next day! For me this took 2 weeks around Christmas...
Last edited:
Ooh where did you find that pretty speaker! 😀


Thanks for the tips!
I didn't have really great success with the bass tuning yet.. So far my best result (Judged by listening to music) I got from not applying any EQ to the woofers what so ever, just the crossover. And then dampen the peaks measured in listening position. But still not perfect!
I will try to follow your guide, even though I don't have a big room.. I have a small(3,6 x 4ish m) almost empty room though. Maybe that will work?
Problem with my listening position is that I am sitting in a couch really close to the back wall :/ I know this is not ideal, at all, but it is how it is.


Thanks for the tips!
I didn't have really great success with the bass tuning yet.. So far my best result (Judged by listening to music) I got from not applying any EQ to the woofers what so ever, just the crossover. And then dampen the peaks measured in listening position. But still not perfect!
I will try to follow your guide, even though I don't have a big room.. I have a small(3,6 x 4ish m) almost empty room though. Maybe that will work?
Problem with my listening position is that I am sitting in a couch really close to the back wall :/ I know this is not ideal, at all, but it is how it is.
In you room this will be difficult! What material are walls, brick/concrete , gypsum board with insulation, wood? Material affects how much of bass energy goes through the wall.
If you listen next to wall, you must let low end response to rise 5-10dB (plus modes) in measurements (below 100Hz). Your brains will compensate for this and reverberant/decayed sound energy is more important to how it sounds to you. With hard wall there will also be destructive reflection nulling the response sharply at some frequency, just like what happens with the front wall. SBIR calculator
Be patient and try different placements of speaker and mic, this way you will learn how the speaker and room interact. One trick to test is to play sine wave of certain problematic frequency from a single speaker and walk around the room to hear how amplitude changes. Remember that also height position matters!
If you listen next to wall, you must let low end response to rise 5-10dB (plus modes) in measurements (below 100Hz). Your brains will compensate for this and reverberant/decayed sound energy is more important to how it sounds to you. With hard wall there will also be destructive reflection nulling the response sharply at some frequency, just like what happens with the front wall. SBIR calculator
Be patient and try different placements of speaker and mic, this way you will learn how the speaker and room interact. One trick to test is to play sine wave of certain problematic frequency from a single speaker and walk around the room to hear how amplitude changes. Remember that also height position matters!
Multiple subs 😉 Actually, if you tune it right - one sub can make a big difference - of course in combination with both your mains.Problem with my listening position is that I am sitting in a couch really close to the back wall :/ I know this is not ideal, at all, but it is how it is.
Juhazi#
Walls are brick walls with wallpaper/tapestry/wall covering (don't know the right word).
Floors are wood, so is ceiling.
Okay, thanks for the tip.
There is sadly no real option to move my speakers around, even though the placement is the opposite of ideal. This is a sketch of my living room (the circles 1, 2 and 3 are listening positions - All measurements are estimates, only a sketch. But the right speaker is close to the wall!):
The corner placed speaker (right speaker) is in a really bad spot I know, but I have nowhere else to put it. The other speaker (left) is placed in a way no simulation software can handle (and I don't know how it actually affects sound).
Would I benefit from removing the woofer facing the side wall and only have one woofer pointing towards the TV? Then I won't have the force cancellation anymore though.
I can do this by (on both speakers of course):
Removing one woofer (would create a open back/dipole/ported configuration)
Removing one woofer and block the hole (sealed enclosure again but twice the volume pr driver.)
Disconnect the woofer but leave it installed. (Passive radiator design).
However, I was hoping I could simply design the speakers for this actual placement (knowing it would be a redo of crossover if they where ever to be moved - !).
DigitalThor#
With the information above, would you think there would be any benefit from having one (or two) of the same 10 in subs in my mains (Shallow SB subs - only kind of 10" driver that fits under my low couch -15 cm of air) under the couch?
Walls are brick walls with wallpaper/tapestry/wall covering (don't know the right word).
Floors are wood, so is ceiling.
Okay, thanks for the tip.
There is sadly no real option to move my speakers around, even though the placement is the opposite of ideal. This is a sketch of my living room (the circles 1, 2 and 3 are listening positions - All measurements are estimates, only a sketch. But the right speaker is close to the wall!):

The corner placed speaker (right speaker) is in a really bad spot I know, but I have nowhere else to put it. The other speaker (left) is placed in a way no simulation software can handle (and I don't know how it actually affects sound).
Would I benefit from removing the woofer facing the side wall and only have one woofer pointing towards the TV? Then I won't have the force cancellation anymore though.
I can do this by (on both speakers of course):
Removing one woofer (would create a open back/dipole/ported configuration)
Removing one woofer and block the hole (sealed enclosure again but twice the volume pr driver.)
Disconnect the woofer but leave it installed. (Passive radiator design).
However, I was hoping I could simply design the speakers for this actual placement (knowing it would be a redo of crossover if they where ever to be moved - !).
DigitalThor#
With the information above, would you think there would be any benefit from having one (or two) of the same 10 in subs in my mains (Shallow SB subs - only kind of 10" driver that fits under my low couch -15 cm of air) under the couch?
Difficult room and positioning. Can you rotate clockwise 90¤? Now R speaker bets bass boost from two walls and also midrange has lots of early refections from sidewall. L speaker gets very little boost from wall. 90¤ turned setup would be symmetrical.
I think that anyway you should make the speakers to have double active woofers and forget the subwoofer. With dsp you can do different eq corrections to L/R. Still stereo image surely won't be symmetrical through spectrum. Multiple subwoofers might be better, but require enormous effort and trials to play really well - I don't recommend that in this case.
p.s. Trick to WM xo : Put the speaker in the door opening to corridor or another room, this way backside radiation and reflection will get eliminated almost totally! Also wall boost is smaller. Helps a lot to set timing/delay.
I think that anyway you should make the speakers to have double active woofers and forget the subwoofer. With dsp you can do different eq corrections to L/R. Still stereo image surely won't be symmetrical through spectrum. Multiple subwoofers might be better, but require enormous effort and trials to play really well - I don't recommend that in this case.
p.s. Trick to WM xo : Put the speaker in the door opening to corridor or another room, this way backside radiation and reflection will get eliminated almost totally! Also wall boost is smaller. Helps a lot to set timing/delay.
Last edited:
I know this is a difficult room. I cannot do much about the layout. This is an apartment and at some point in the not too far future we will move to a house. So until then I don't expect perfect results. Just as good as possible under given circumstances.
That is a good tip with the doorway. Didn't think about that way of making the setup symetrical.
How big a difference do you think it will make once i get the second speaker build? Won't it be a little like multiple subs? (2 compared to 1).
That is a good tip with the doorway. Didn't think about that way of making the setup symetrical.
How big a difference do you think it will make once i get the second speaker build? Won't it be a little like multiple subs? (2 compared to 1).
I'm a big believer in symmetrical front speaker placement so if you absolutely cannot alter the room configuration (which I'll assume to be true otherwise you probably would have done it already), I would add a sort of moveable/foldable 'wall' of some type to the left of the speakers. Something as shown in the pic below (that would be with it open). I'd then center the front setup in the new space. I did something like this in an old place of mine using a folding closet door that I pulled from a spot where it wasn't needed, but you can probably build something better than that with your skills.
Also in a small space like that, I've found absorption panels on the wall behind the couch to be very helpful.
Also in a small space like that, I've found absorption panels on the wall behind the couch to be very helpful.
Attachments
- Home
- Loudspeakers
- Multi-Way
- Small active 3-way floor stander with a punch