Dear fellow forum members,
I am currently preparing for another day of measuring my diy B&W CM8 S2 with FA-123 hypex plate (also converted to sealed model by removing bass reflex port).
Apart from experimenting with placing absorbing materials in the middle instead of the sides (the 'burrito approach'), I would also like to experiment with resetting the crossover frequencies and type of linkwitz riley filter (LR2, 4, 8 & 16).
To my (very limited) understanding, when working with LR filters, it's important to have the cones (tweeter, mid and woofer) properly alligned.
In order to do that, I have to measure the distance between the cones and a fixed point (e.g. the front of the speaker). However, the cones of the drivers of the B&W CM8 S2 are very different in design.
Where exactly do you need to measure?
See https://www.homecinemamagazine.nl/wp-content/uploads/2015/05/BW-CM-S2-review-7-770x449.jpg for a photo of the speakers.
In the past I've used the 'invert' option on the plate amps to determine the delay in the tweeter by seeing a dip at the crossover frequency (using LR4 filters). However, this was done after designing the crossovers and I'd like to implement the tweeter delay before the design.
Any tips? Thanks!
I am currently preparing for another day of measuring my diy B&W CM8 S2 with FA-123 hypex plate (also converted to sealed model by removing bass reflex port).
Apart from experimenting with placing absorbing materials in the middle instead of the sides (the 'burrito approach'), I would also like to experiment with resetting the crossover frequencies and type of linkwitz riley filter (LR2, 4, 8 & 16).
To my (very limited) understanding, when working with LR filters, it's important to have the cones (tweeter, mid and woofer) properly alligned.
In order to do that, I have to measure the distance between the cones and a fixed point (e.g. the front of the speaker). However, the cones of the drivers of the B&W CM8 S2 are very different in design.
Where exactly do you need to measure?
See https://www.homecinemamagazine.nl/wp-content/uploads/2015/05/BW-CM-S2-review-7-770x449.jpg for a photo of the speakers.
In the past I've used the 'invert' option on the plate amps to determine the delay in the tweeter by seeing a dip at the crossover frequency (using LR4 filters). However, this was done after designing the crossovers and I'd like to implement the tweeter delay before the design.
Any tips? Thanks!
IF you have the possibility to stagger the drivers instead of mounting them to the same plane, a flat front baffle, the "point source" for each can be considered the voice coil to cone (or dome) joint.
So if, say, you have 2 drivers, a woofer and a tweeter and the woofer cone is , say, 8 cm deep and the tweeter is, say, 1 cm deep, you can mount the tweeter 7 cm further back than where it would seat if screwed to the main front panel.
PS: just checked your picture, to me those are VERY conventional cone speakers, just with a funny plug glued to the magnet pole piece
If used as is, you will NOT correct for the time delay difference by tweaking the passive crossover; now if you use an active crossover and separate amplifiers, you can slightly delay the signal sent to tweeter.
The easiest solution would be to place the tweeter in its own small enclosure and sit it on top of the cabinet, a few cm to the back.
Measure and do the Math to find how many.
So if, say, you have 2 drivers, a woofer and a tweeter and the woofer cone is , say, 8 cm deep and the tweeter is, say, 1 cm deep, you can mount the tweeter 7 cm further back than where it would seat if screwed to the main front panel.
PS: just checked your picture, to me those are VERY conventional cone speakers, just with a funny plug glued to the magnet pole piece
If used as is, you will NOT correct for the time delay difference by tweaking the passive crossover; now if you use an active crossover and separate amplifiers, you can slightly delay the signal sent to tweeter.
The easiest solution would be to place the tweeter in its own small enclosure and sit it on top of the cabinet, a few cm to the back.
Measure and do the Math to find how many.
Last edited:
It is possible to tweak the crossover to compensate for a time difference and achieve the same results as a Linkwitz filter.
Hi Allen,
Thanks for your response. With the FA-123 plate amps from Hypex, you can easly delay the tweeter response. There are multiple ways of doing this using REW (e.g. Time Alignment between mid + tweeter -- measuring step response using REW | DiyMobileAudio.com Car Stereo Forum) but a nice start would be to know the difference between the two coils.
Thanks for your response. With the FA-123 plate amps from Hypex, you can easly delay the tweeter response. There are multiple ways of doing this using REW (e.g. Time Alignment between mid + tweeter -- measuring step response using REW | DiyMobileAudio.com Car Stereo Forum) but a nice start would be to know the difference between the two coils.
Indeed. The strategy you describe in your first post makes me think you want to reference the distances to the listening position?
There are many ways you can do that. For example you could measure each driver at a fixed distance from the baffle, and then calculate the different distances to the listening position.
Be careful about time adjusting drivers that are working together, this can upset their acoustic behaviour.
There are many ways you can do that. For example you could measure each driver at a fixed distance from the baffle, and then calculate the different distances to the listening position.
Be careful about time adjusting drivers that are working together, this can upset their acoustic behaviour.
Thanks for your responses.
Since I'll be listening at tweeter height at approximately 2.8 m from the speaker position and since I'll be using a NAD C658 with DIRAC (full licence) as a pre-amp, my mission is somewhat simpler (I hope).
What I'm basically trying to do is to find the right tweeter (and/or woofer/mid) delay to implement various LR filters. I'm thinking of using the following steps (Using REW to Determine Time Delays Between Drivers - Technical/Modifications - The Klipsch Audio Community) but I'm trying to find a starting delay for the tweeter (there is approximately 2.5 cm between the coils of the tweeter/mid and tweeter/woofer).
In the past months of using my filters (build with the help of this forum), I found out that I corrected way to much small dips and bumps and after removing them the system sounds better.
Therefore (and since I have some extra spare time the coming weeks) I decided to do a new fresh round of filter designing given that I now a little bit better what my preferences are.
My plan is to adjust the speaker isolation (burrito technique) since I maybe now overdid the isolation on the inner sides of the speakers. Then I'd like to remeasure the frequency responses of the drivers and design various LR filters (2nd, 4th and 8th order) and optimizing them and see what results I get both on- and off-axis.
The first round of filter designing gave very satisfactory results, and I'd like to go for a new round to see if I can improve the result a little.
Since I'll be listening at tweeter height at approximately 2.8 m from the speaker position and since I'll be using a NAD C658 with DIRAC (full licence) as a pre-amp, my mission is somewhat simpler (I hope).
What I'm basically trying to do is to find the right tweeter (and/or woofer/mid) delay to implement various LR filters. I'm thinking of using the following steps (Using REW to Determine Time Delays Between Drivers - Technical/Modifications - The Klipsch Audio Community) but I'm trying to find a starting delay for the tweeter (there is approximately 2.5 cm between the coils of the tweeter/mid and tweeter/woofer).
In the past months of using my filters (build with the help of this forum), I found out that I corrected way to much small dips and bumps and after removing them the system sounds better.
Therefore (and since I have some extra spare time the coming weeks) I decided to do a new fresh round of filter designing given that I now a little bit better what my preferences are.
My plan is to adjust the speaker isolation (burrito technique) since I maybe now overdid the isolation on the inner sides of the speakers. Then I'd like to remeasure the frequency responses of the drivers and design various LR filters (2nd, 4th and 8th order) and optimizing them and see what results I get both on- and off-axis.
The first round of filter designing gave very satisfactory results, and I'd like to go for a new round to see if I can improve the result a little.
The ultimate goal would be to not need DIRAC at all. The more you use it, the more work you have to do on your speakers.
OK, so your first guess is that the tweeter coil is at the baffle and your woofer is 2.5cm back. In some respects this can get you moving.
Do you use REW in dual channel mode?
Hi, a few thoughts...
First, if you stick with LR filters, there is no time difference to account for from the xover, when using the same xover frequency on both sides.
12 & 36 dB/oct will need the low-passed side inverted.
24 & 48 dB/oct stay polarity positive both sides.
Do that, and the measured time difference between drivers will be solely due to distances between the acoustic centers and measurement/listening position.
Which is really cool, because the timing offset stays the same as you try the different LR orders.
(Fully complemetary xovers totally rock imo,...particularly if you do driver-by-driver optimizations prior to implementing the xovers...)
The measured time difference between the acoustic centers should be reasonably close to the coil alignment difference guesstimate...
think of the coil alignment as a 'smell test' number.... if measured is too far away from it, something is wrong (as in more than a few cm for the speaker you describe)
First, if you stick with LR filters, there is no time difference to account for from the xover, when using the same xover frequency on both sides.
12 & 36 dB/oct will need the low-passed side inverted.
24 & 48 dB/oct stay polarity positive both sides.
Do that, and the measured time difference between drivers will be solely due to distances between the acoustic centers and measurement/listening position.
Which is really cool, because the timing offset stays the same as you try the different LR orders.
(Fully complemetary xovers totally rock imo,...particularly if you do driver-by-driver optimizations prior to implementing the xovers...)
The measured time difference between the acoustic centers should be reasonably close to the coil alignment difference guesstimate...
think of the coil alignment as a 'smell test' number.... if measured is too far away from it, something is wrong (as in more than a few cm for the speaker you describe)
I've measured the coil differences at 2.5 cm (since there is a pointy thing glued on the coil center as can be seen on the photo there is a bit of guessing involved) which should be approximately 75 us.
I don't use REW in dual channel mode (I measure each speaker individually just to check if I'm interpreting dual channel mode correctly). I can't use timing programs on REW or with other software since I use a USB-dac w optical out to the FA-123 and measure with an USB microphone.
Another trick I learned earlier was to design the filters (let's say LR 4th order) and then invert the tweeter (and afterwards the mid-woofer) to see if a dip arises at the XO frequency. At the greatest dip the timing is perfectly (or near that) alligned because of the 180 phase difference.
As you can read, I'm still learning and searching towards the best method.
However, on other note, I just redid the isolation of the speakers towards a burrito method on the woofers and the effect seems positive in terms of muffled 'reverberations' (I don't know the appropriate term). However, a placebo effect is also likely since I've spend and hour or two on performing the task.
I don't use REW in dual channel mode (I measure each speaker individually just to check if I'm interpreting dual channel mode correctly). I can't use timing programs on REW or with other software since I use a USB-dac w optical out to the FA-123 and measure with an USB microphone.
Another trick I learned earlier was to design the filters (let's say LR 4th order) and then invert the tweeter (and afterwards the mid-woofer) to see if a dip arises at the XO frequency. At the greatest dip the timing is perfectly (or near that) alligned because of the 180 phase difference.
As you can read, I'm still learning and searching towards the best method.
However, on other note, I just redid the isolation of the speakers towards a burrito method on the woofers and the effect seems positive in terms of muffled 'reverberations' (I don't know the appropriate term). However, a placebo effect is also likely since I've spend and hour or two on performing the task.
Yep, the 'invert and look for deepest null, while varying delay' is the time honored old-school way of aligning subs to mains in the live sound world.
But measuring is so much faster, and more accurate, huh?
If you're not using loopback timing with REW, you can use acoustic timing reference. That should work for timing I think, no matter what equipment you are using in your processing and measuring chains.
Biggest thing is, measure timing without the LR crossovers in place.
Then the difference in acoustic reference timings will be the distance between acoustic centers.
But measuring is so much faster, and more accurate, huh?
If you're not using loopback timing with REW, you can use acoustic timing reference. That should work for timing I think, no matter what equipment you are using in your processing and measuring chains.
Biggest thing is, measure timing without the LR crossovers in place.
Then the difference in acoustic reference timings will be the distance between acoustic centers.
Nice, the method you propose alligns with my plan (driver optimization, time allignment and then xover design.
However, what do you mean by fully complentary xovers?
But, if you guys want to think along with me, and given that I have no data on my drivers, what crossover would you suggest?
Overlay:
Bass (SPL/phase):
20200706-Bass-SPLPhase — ImgBB
Mid (SPL/phase):
20200706-Mid-SPLPhase — ImgBB
Tweet (SPL/phase):
20200706-Tweet-SPLphase — ImgBB
This images provides the freq responses of the drivers after trying to flatten them out over a broader range and separte with SPL and phase. I still have to make new measurements, but at least you can see the limits of the driver ranges. These were indoor measurements I think (given the resonances). No filters were applied here.
I capped the tweeter by sweeping from 500 Hz and upwards.
Overlay:
An externally hosted image should be here but it was not working when we last tested it.
Bass (SPL/phase):
20200706-Bass-SPLPhase — ImgBB
Mid (SPL/phase):
20200706-Mid-SPLPhase — ImgBB
Tweet (SPL/phase):
20200706-Tweet-SPLphase — ImgBB
This images provides the freq responses of the drivers after trying to flatten them out over a broader range and separte with SPL and phase. I still have to make new measurements, but at least you can see the limits of the driver ranges. These were indoor measurements I think (given the resonances). No filters were applied here.
I capped the tweeter by sweeping from 500 Hz and upwards.
Here ya go....Linkwitz-Riley Crossovers: A Primer
Scroll down to where you find this under Linkwitz-Riley Alignment
"Two things characterize a Linkwitz-Riley alignment:
1. In-phase outputs (0° between outputs) at all frequencies (not just at the crossover frequency as popularly believed by some).
2. Constant voltage (the outputs sum to unity at all frequencies)."
That's what fully complementary does.
Takes xover timing out of the puzzle.....although many do their best to put in back in lol
O sorry, another question comes to mind.
When measuring acoustic timing for the individual drivers.
It seems reasonable to measure at the level of each driver (i.e. 1 meter).
However, measurement at fixed position (tweeter height, 1 meter distance) is also an option. When using no filter, what seems the best option to do?
When measuring acoustic timing for the individual drivers.
It seems reasonable to measure at the level of each driver (i.e. 1 meter).
However, measurement at fixed position (tweeter height, 1 meter distance) is also an option. When using no filter, what seems the best option to do?
You know, sometimes it just pays to try stuff....and google questions, and try again...
I don't know how hypex software works, but nearly every decent xover software has drop down menus for the LR orders.
I mean, it's the most standard stuff there is....
You should be able to go straight to LR 8th without stacking lesser orders........
if not, most certainly screw hypex lol....
When measuring timing, do it at listening distance.
Most folks probably center the mic vertically somewhere between HF and Mid.
I don't know how hypex software works, but nearly every decent xover software has drop down menus for the LR orders.
I mean, it's the most standard stuff there is....
You should be able to go straight to LR 8th without stacking lesser orders........
if not, most certainly screw hypex lol....
When measuring timing, do it at listening distance.
Most folks probably center the mic vertically somewhere between HF and Mid.
The acoustic offset may not match the physical offset. People use the location of the voicecoil as a rule of thumb but it depends on how rigid the cone is and so forth as to where the actual acoustic offset is. Also, different measurement software have different ways of normalising the phase response (nulling the delay from the measurement automatically), so you end up with a frequency/phase response measurement with an unknown offset.
The only way to accurately design a crossover is to measure one driver, then the other driver and then both together (without any crossover). By summing the individual measurements together in software, you should get the latter real life measurement of both drivers playing together. If it doesn't match then you can adjust the delay/offset of either singular driver measurement in software until it sums to perfectly match the real life measurement. Then you export the individual responses with relevant delay/offsets included and design the crossover.
The only way to accurately design a crossover is to measure one driver, then the other driver and then both together (without any crossover). By summing the individual measurements together in software, you should get the latter real life measurement of both drivers playing together. If it doesn't match then you can adjust the delay/offset of either singular driver measurement in software until it sums to perfectly match the real life measurement. Then you export the individual responses with relevant delay/offsets included and design the crossover.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.