Yeah, the problem with Omnimic is that it is using peak detection of the impulse to determine the start of the FFT window. Once you start measuring around the rear of the speaker, you will have room reflections that are higher amplitude than the signal from the speaker facing away from the mic so it starts having some trouble with that.
So I did some quick and dirty measurements this morning. I only did 0 deg, 30deg, and 60deg as a test, and by the time alignment calculator made some adjustments of +6mm at 30 degrees, and -5mm at 60 degrees.
The comparison on power response of including the delay or not is pretty much negligible for this basic test. @4thtry I suppose you could do the same test with your data set, load into VituixCAD unmodified vs with all the individual calculated delays included, and see how much the power and DI graph is changed, if at all.
I did another test and simply modified the off-axis delays by more extreme values, and certainly the phase relations here do affect the overall power response, but a couple mm one way or the other through the off-axis measurements isn't going to throw things off by a real meaningful amount.
My thoughts here is that the individual delay of the off-axis angles is not quite as important here, as being able to create the 3D baloon of data data around each driver, measure each driver on it's own axis fully, and doing so with a 2 channel configuration simplifies the entire process maintaining correct phase relation at all angles without any effort at all. The only effort required for the 2 channel measurement is to keep a constant distance from mic to baffle for all measurements.
You may find that creating the baloon of data measured only at tweeter axis with min phase response, and delay calculated at on-axis angle only provides a "better than nothing" result, as long as you have understanding of the shortcomings of that information.
Thanks, dcibel! I really appreciate you taking the time to test this for me. I have really learned alot in the past few days by reading (and re-reading) your detailed posts.
So, if I am understanding you correctly, basically what it comes down to is that the small to medium time alignment errors that I am measuring in the forward facing hemisphere will have little to no impact on VituixCAD's power response or DI curves. By time correcting OmniMic's off-axis FRD's, I can accurately model on-axis as well as any individual forward facing off-axis angle, but I cannot accurately model power response, DI, or listening window information. To do that properly, I need a 2 channel microphone setup that can accurately measure the rear facing hemisphere as well.
On the positive side, this seems much better than what I was previously doing in XSim. In XSim, I could only model on-axis and perhaps two or three polar angles (like 45H and + or - 10V). Doing more than this was much too difficult, as I would have to quickly flip driver icons back & forth on the schematic screen as I tweaked the xover. Not a very good system at all. In VituixCAD, I can at least watch both on and off axis curves update instantly as I spin or optimize xover part values. That aspect alone is a real improvement for me. Can't wait to get my 2 channel system set up and running! Thanks again!
I installed a couple internal braces to dampen the side panels:
I routed 4 small donuts from the curly maple woofer cut outs. I didn't cut all the way through with the router. Instead, I left about 1/8" of material and finished the donut cut outs with a small hand saw.
I then silicone glued two of the donuts to the internal ends of my bass reflex port tubes:
After the glue set for a few days, I rounded over the donuts to produce curly maple flares. I gave the curly maple flares several coats of Arm-R-Seal and then silicone glued the ports in place using the other two donuts as mounting flanges.
@Nicholas_23 said:
I wonder if satin or gloss helps with the grain popping unsanded . You have voids in the grain, then fill it with a satin/gloss and get a 3D holographic view when finished and dried. Bounces the light around a bit if you will. Really sharp! Im curious when and how you do the woofers.
@Nicholas_23 -- I finally got around to re-foaming and re-sizing the cone to replace the woofer that I had botched up a while back. So, here are a few photos of the process.
First off, I cut the cone down by about 1 inch with a hobby knife:
Rear of woofer showing the backside of the downsized cone:
Test fit of new foam surround and 1/4" masonite ring adapter. I used the PE #260-955 Advent refoaming kit.
First, I glued the new foam surround to the 1/4" masonite ring, being very careful not to de-form the circular shape of the new surround.
Then, I glued the inner edge of the new foam surround to the cone, allowing the 1/4" masonite ring to float free or loose (very important):
After allowing the glue to dry several hours, I lifted the 1/4" masonite ring up by about 1/2 inch or so and held it in place with a standard #2 pencil on each side:
I then ran a bead of black 100" silicone cement under the ring and along the sides. When I pulled out the pencils, the masonite ring pressed into the silicone cement. The new foam surround self centered the masonite ring as the silicone cured.
To finish up, I removed the temporary alignment shims and glued a new dusk cap in place. Done.
The finished woofer sounds OK, with an Fs of about 22Hz and Qts of .34. The higher frequencies in the 500-1000Hz region are not a perfect match, but overall efficiency and alignment of the two woofers is roughly the same. One take away is that you need to be very careful to use the same amount of glue on the cone and dust cap as this can affect the high frequency performance quite a bit.
I built a pair of matching hardwood stands. They are 9 inches high and use the same mahogany + maple pattern as the main cabinets. To improve the visuals, I angled the stands 17 degrees to create a "launching pad" type profile.
Next up: Dual channel spinorama and VituixCAD modeling. Will post detailed screenshots as I go along. I have been reviewing the various spinorama pdf's and Erin's Audio Corner reviews in an effort to develop a better understanding of the various curves.
@Nicholas_23 said:
Will you be placing the Xovers in the new stands you built?
I was thinking about that. But too late. I have already cut recessed holes for the input cups. Lots of empty space, though. I could easily bolt a plate amp under each one for bi-amping.
If you built a front panel for the lower compartment, you could put a small sand bag or two in each one for ballast plus damping. This may or may not benefit the bass response, but it certainly has potential...
@Billet said:
If you built a front panel for the lower compartment, you could put a small sand bag or two in each one for ballast plus damping. This may or may not benefit the bass response, but it certainly has potential...
Speaking of damping, I bought a Flexgard 1/4" thick rubber utility mat on sale today. $29 for a 36x60" roll. Very dense rubber, has the look and feel of the rubber used on automobile tires. I glued a small test piece of the rubber to a small block of mahogany to see how it sticks to wood using 100% silicone sealer. Sometimes silicone does not stick well to foam rubber material, but this is very dense rubber and may work better. What would be the best type of glue to use to glue rubber to wood? I am thinking of cutting small 6x8" or 8x10" pieces of this rubber and gluing it to the interior enclosure walls at various spots to dampen resonances. As I tap along the enclosure surfaces, many areas go Boooonnnngggg with a slow decay.
The enclosure will bong less with drivers and stuffing, any stuffing at all, it doesn't need to be directly fastened to the panels, because the majority of that long decay is just the waves bouncing around in the open unrestricted space. Close it up with a pillow stuffed in one, leave the other cabinet empty and compare, the difference is not subtle.
@4thtry said:
Next up: Dual channel spinorama and VituixCAD modeling. Will post detailed screenshots as I go along. I have been reviewing the various spinorama pdf's and Erin's Audio Corner reviews in an effort to develop a better understanding of the various curves.
I can provide a little insight on the "spinorama" for design. Simply put, the set of responses here are each progressively including more and more of the off-axis data as you move down the list from on-axis, listening window, early reflections, in-room, sound power. DI is a comparison of either on-axis or listening window to the sound power response depending on the option selected (listening window by default). The combination of all these plots provides rather detailed insight into the combined overall effect of driver size, separation, cabinet shape, and crossover without having to interrogate each off-axis response individually.
I personally don't pay much attention to the ER and ERDI for crossover optimization, can really turn those guys off to reduce the clutter.
I set a my optimization targets by listening window and in-room response. Set listening window target on the SPL chart at -0.2dB/oct from 100Hz to 10kHz. Set the in-room target on the power & DI chart at -0.8dB/oct from 100Hz to 10kHz, or a similar slope that should draw a nice trend line through the in-room plot. Then optimize the overall response by listening window and in-room response at 50/50 weight.
Thanks for the target line suggestions. I will be doing this shortly. It took me a while to get everything set up, but I finally completed the full set of dual channel spinorama measurements on this speaker. Total of 39 measurements; 13 for tweeter, 13 for mid, and 13 for woofer (0,15,30,45,60,75,90,105,120,135,150,165,&180). Microphone distance was 39" on axis to the driver for all measurements. Microphone distance to the floor was 72.5" for the set of 13 tweeter measurements; 66.9" to the floor for the set of 13 midrange measurements, and 54.5" to the floor for the set of 13 woofer measurements. The vaulted ceiling distances were even higher. So I should be able to get a reflection free gate close to 5ms or so.
I created a zipped up folder with all FF measurements, woofer NF measurements, in box zma files, and dimensional drawing in case anyone is interested in looking at the raw IR files. It is about 75 megs, so I cannot post it here. Using VituixCAD's "Convert IR to FR" tool, I also created a zipped up folder of the same files in FRD format. It is only about 600K, but the FRD's have my microphone calibration file and gating decisions baked in.
I loaded the raw Soundeasy IR files into VituixCAD's "Convert IR to FR" and then re-saved them as single column mlssa.txt formatted IR files. This eliminated the redundant data column and reduced the zip file size by 40%.
EDIT: Large attachment removed to conserve server space.
Great, first glance they all look like perfectly good measurements. Use of a dome midrange is putting the impulse peak at the same time location, so they are "time aligned". Midrange response is a bit unfortunate but I think you still have some useful bandwidth there to make it work.
I've generated my first model using the spinorama data set. Here are a few screen shots of the process:
Initially, I loaded the raw IR files into the Converter tool and set the gate for all files to 4.8ms. Then I used the "feed speaker" check box to automatically load the resulting frd files into the various driver frequency response tabs.
Comments
Yeah, the problem with Omnimic is that it is using peak detection of the impulse to determine the start of the FFT window. Once you start measuring around the rear of the speaker, you will have room reflections that are higher amplitude than the signal from the speaker facing away from the mic so it starts having some trouble with that.
So I did some quick and dirty measurements this morning. I only did 0 deg, 30deg, and 60deg as a test, and by the time alignment calculator made some adjustments of +6mm at 30 degrees, and -5mm at 60 degrees.
The comparison on power response of including the delay or not is pretty much negligible for this basic test. @4thtry I suppose you could do the same test with your data set, load into VituixCAD unmodified vs with all the individual calculated delays included, and see how much the power and DI graph is changed, if at all.
I did another test and simply modified the off-axis delays by more extreme values, and certainly the phase relations here do affect the overall power response, but a couple mm one way or the other through the off-axis measurements isn't going to throw things off by a real meaningful amount.
My thoughts here is that the individual delay of the off-axis angles is not quite as important here, as being able to create the 3D baloon of data data around each driver, measure each driver on it's own axis fully, and doing so with a 2 channel configuration simplifies the entire process maintaining correct phase relation at all angles without any effort at all. The only effort required for the 2 channel measurement is to keep a constant distance from mic to baffle for all measurements.
You may find that creating the baloon of data measured only at tweeter axis with min phase response, and delay calculated at on-axis angle only provides a "better than nothing" result, as long as you have understanding of the shortcomings of that information.
Thanks, dcibel! I really appreciate you taking the time to test this for me. I have really learned alot in the past few days by reading (and re-reading) your detailed posts.
So, if I am understanding you correctly, basically what it comes down to is that the small to medium time alignment errors that I am measuring in the forward facing hemisphere will have little to no impact on VituixCAD's power response or DI curves. By time correcting OmniMic's off-axis FRD's, I can accurately model on-axis as well as any individual forward facing off-axis angle, but I cannot accurately model power response, DI, or listening window information. To do that properly, I need a 2 channel microphone setup that can accurately measure the rear facing hemisphere as well.
On the positive side, this seems much better than what I was previously doing in XSim. In XSim, I could only model on-axis and perhaps two or three polar angles (like 45H and + or - 10V). Doing more than this was much too difficult, as I would have to quickly flip driver icons back & forth on the schematic screen as I tweaked the xover. Not a very good system at all. In VituixCAD, I can at least watch both on and off axis curves update instantly as I spin or optimize xover part values. That aspect alone is a real improvement for me. Can't wait to get my 2 channel system set up and running! Thanks again!
I installed a couple internal braces to dampen the side panels:
I routed 4 small donuts from the curly maple woofer cut outs. I didn't cut all the way through with the router. Instead, I left about 1/8" of material and finished the donut cut outs with a small hand saw.
I then silicone glued two of the donuts to the internal ends of my bass reflex port tubes:
After the glue set for a few days, I rounded over the donuts to produce curly maple flares. I gave the curly maple flares several coats of Arm-R-Seal and then silicone glued the ports in place using the other two donuts as mounting flanges.
You know you've got an issue when your inside parts, are as pretty as the outside parts.
I sealed up all interior surfaces with Arm-R-Seal as well. Trying to slow down the rate at which the wood picks up or releases moisture.
@Nicholas_23 -- I finally got around to re-foaming and re-sizing the cone to replace the woofer that I had botched up a while back. So, here are a few photos of the process.
First off, I cut the cone down by about 1 inch with a hobby knife:
Rear of woofer showing the backside of the downsized cone:
Test fit of new foam surround and 1/4" masonite ring adapter. I used the PE #260-955 Advent refoaming kit.
First, I glued the new foam surround to the 1/4" masonite ring, being very careful not to de-form the circular shape of the new surround.
Then, I glued the inner edge of the new foam surround to the cone, allowing the 1/4" masonite ring to float free or loose (very important):
After allowing the glue to dry several hours, I lifted the 1/4" masonite ring up by about 1/2 inch or so and held it in place with a standard #2 pencil on each side:
I then ran a bead of black 100" silicone cement under the ring and along the sides. When I pulled out the pencils, the masonite ring pressed into the silicone cement. The new foam surround self centered the masonite ring as the silicone cured.
To finish up, I removed the temporary alignment shims and glued a new dusk cap in place. Done.
The finished woofer sounds OK, with an Fs of about 22Hz and Qts of .34. The higher frequencies in the 500-1000Hz region are not a perfect match, but overall efficiency and alignment of the two woofers is roughly the same. One take away is that you need to be very careful to use the same amount of glue on the cone and dust cap as this can affect the high frequency performance quite a bit.
I built a pair of matching hardwood stands. They are 9 inches high and use the same mahogany + maple pattern as the main cabinets. To improve the visuals, I angled the stands 17 degrees to create a "launching pad" type profile.
Those look really nice Bill! Can't wait to hear them.
Yes - very cool! You now have the perfect spot for some mono blocks.
Kinda has a lava lamp feel now👍🏻 Looking good dude!
A true artist! Wowzer
Thanks, guys. I plan to demo these at InDIYana in May. I should have the crossovers done by then. If you can make it, stop by and give them a listen.
Next up: Dual channel spinorama and VituixCAD modeling. Will post detailed screenshots as I go along. I have been reviewing the various spinorama pdf's and Erin's Audio Corner reviews in an effort to develop a better understanding of the various curves.
Amazing work and a version of Pythagoras in hell . . .
@4thtry , makes sense now that Ive seen the pics , boy I bet that took a steady hand !
Will you be placing the Xovers in the new stands you built?
I was thinking about that. But too late. I have already cut recessed holes for the input cups. Lots of empty space, though. I could easily bolt a plate amp under each one for bi-amping.
If you built a front panel for the lower compartment, you could put a small sand bag or two in each one for ballast plus damping. This may or may not benefit the bass response, but it certainly has potential...
Speaking of damping, I bought a Flexgard 1/4" thick rubber utility mat on sale today. $29 for a 36x60" roll. Very dense rubber, has the look and feel of the rubber used on automobile tires. I glued a small test piece of the rubber to a small block of mahogany to see how it sticks to wood using 100% silicone sealer. Sometimes silicone does not stick well to foam rubber material, but this is very dense rubber and may work better. What would be the best type of glue to use to glue rubber to wood? I am thinking of cutting small 6x8" or 8x10" pieces of this rubber and gluing it to the interior enclosure walls at various spots to dampen resonances. As I tap along the enclosure surfaces, many areas go Boooonnnngggg with a slow decay.
Contact cement.
Contact cement is good idea I think.
The enclosure will bong less with drivers and stuffing, any stuffing at all, it doesn't need to be directly fastened to the panels, because the majority of that long decay is just the waves bouncing around in the open unrestricted space. Close it up with a pillow stuffed in one, leave the other cabinet empty and compare, the difference is not subtle.
Contact cement for the win! Put some stuffing in the box and re-tapped. You are correct, makes a huge difference in the bong.
I can provide a little insight on the "spinorama" for design. Simply put, the set of responses here are each progressively including more and more of the off-axis data as you move down the list from on-axis, listening window, early reflections, in-room, sound power. DI is a comparison of either on-axis or listening window to the sound power response depending on the option selected (listening window by default). The combination of all these plots provides rather detailed insight into the combined overall effect of driver size, separation, cabinet shape, and crossover without having to interrogate each off-axis response individually.
I personally don't pay much attention to the ER and ERDI for crossover optimization, can really turn those guys off to reduce the clutter.
I set a my optimization targets by listening window and in-room response. Set listening window target on the SPL chart at -0.2dB/oct from 100Hz to 10kHz. Set the in-room target on the power & DI chart at -0.8dB/oct from 100Hz to 10kHz, or a similar slope that should draw a nice trend line through the in-room plot. Then optimize the overall response by listening window and in-room response at 50/50 weight.
Thanks for the target line suggestions. I will be doing this shortly. It took me a while to get everything set up, but I finally completed the full set of dual channel spinorama measurements on this speaker. Total of 39 measurements; 13 for tweeter, 13 for mid, and 13 for woofer (0,15,30,45,60,75,90,105,120,135,150,165,&180). Microphone distance was 39" on axis to the driver for all measurements. Microphone distance to the floor was 72.5" for the set of 13 tweeter measurements; 66.9" to the floor for the set of 13 midrange measurements, and 54.5" to the floor for the set of 13 woofer measurements. The vaulted ceiling distances were even higher. So I should be able to get a reflection free gate close to 5ms or so.
I created a zipped up folder with all FF measurements, woofer NF measurements, in box zma files, and dimensional drawing in case anyone is interested in looking at the raw IR files. It is about 75 megs, so I cannot post it here. Using VituixCAD's "Convert IR to FR" tool, I also created a zipped up folder of the same files in FRD format. It is only about 600K, but the FRD's have my microphone calibration file and gating decisions baked in.
Can you plunk them on google drive or one of the many other popular cloud storage services?
Checking into setting up an account.
I loaded the raw Soundeasy IR files into VituixCAD's "Convert IR to FR" and then re-saved them as single column mlssa.txt formatted IR files. This eliminated the redundant data column and reduced the zip file size by 40%.
EDIT: Large attachment removed to conserve server space.
Great, first glance they all look like perfectly good measurements. Use of a dome midrange is putting the impulse peak at the same time location, so they are "time aligned". Midrange response is a bit unfortunate but I think you still have some useful bandwidth there to make it work.
I've generated my first model using the spinorama data set. Here are a few screen shots of the process:
Initially, I loaded the raw IR files into the Converter tool and set the gate for all files to 4.8ms. Then I used the "feed speaker" check box to automatically load the resulting frd files into the various driver frequency response tabs.