So the rising response on the tweeter proved to be difficult to get a low crossover point - I wanted to hit 1.8-2K. It is problematic in that area. Not a criticism of the tweeter as much as it is a limitation of the mostly flat baffle and lack of wave guide. What I should do is rebuild the cabinets and use a wave guide or at least go to active and eq the top octaves flat. No big deal, I just have some moderate directivity issues.
Here is on-axis and reverse null of my current crossover. I am still deciding if it is what I want it to be, so I am resting my ears for the next few days and follow up switching between some decent cans and these speakers.
..and a distortion plot. 3rd and 4th above ~250Hz is largely below my noise floor at fairly loud volumes. The distortion rises below that, probably a combination of the room and the small woofer. Above that though, it is essentially non-measurable in my space. I had to bridge the APA150 to get this measurement, the little guys are pretty power hungry! The 2nd order dominant could very well be the amp, I don't know. I don't have the energy to set up a different amp to check things.
The crossover is currently eight components. Damped second order on the woofer with a small cap across the coil. Second order on the tweeter with a split L-Pad.
So far, I am pretty happy with things. I picked up an Omnimic and am using this project to learn it. A little clunkier than ARTA, but a few different features. I still prefer the distortion measurement in ARTA.
Anyways, will periodically revisit over the next week or so before I test in stereo and definitely before I commit to soldering.
Good thing about that Motus is that you can run it as high as you want without worry of distortion issues, only power response will be affected as you can see in your own testing, no issue around the crossover frequency
So a few changes here and there, got the crossover lowered a bit to ~2700hz with a better reverse null FWIW. Spent a ton of time listening to them yesterday, and some more this morning. I will likely run with this version.
There is some mild phase issues just above the XO point, but it is minor and I am really not going to chase it down - I don't believe it is worth it.
Distortion is still pretty nice:
I may end up going active at some point to be able to take advantage of this tweeters ability to cross low for a ribbon, I think with the right time delay and contouring I can get a 2nd order LR at 2K.
Also installed a WiFi card on the measurement rig, so no more running upstairs to get on the internet. It also has BT now. Fancy.
Just for S's and G's, here is the measurement in Omnimic.
Omnimic plays noticeably quieter than ARTA. I cannot seem to figure that out, like the test tracks are set too low in level or something. Weird. I have to really crank the volume knob on the amp to equal the output in ARTA - has anyone else run into that?
Another nitpick is the snapshot function outputs a pretty small file - is there a setting for that somewhere? I just did a screenshot of the above.
Here is what I am talking about. Both images used the built in PNG export, the ARTA image is of much higher quality. I opened them both in picture viewer, maximized to my screen, and printed said screen and saved out using Paint. I only really give a shit because it is just easier to identify stuff in the ARTA screen shots.
So far I am almost happy with the Omnimic - it is fairly decent software. If I could figure out why the playback levels are so low and if I can output images with a meaningful resolution I would be pretty well content. I like a lot of the features baked in, for sure.
My images from Omnimic are not small, and usually are .bmp, so they are not small files either.
Lately I've been saving with a smapper to .jpg so they aren't so large in filesize and easier to post without conversion.
@Tom_S said:
Looks great JR. I'm guessing they sound as good as the squiggles look!
They are pretty nice, to be honest. I might take the tweeter down a notch, though. It has very good off-axis response which I think might be contributing to a bit of perceived top octave energy. The sibilance is spot on where it is, though, so I'm not sure if I will change anything.
Omnimic outputs a standard file format for "snapshots" which are fairly low resolution. On the plus side, it does ensure that all snapshots are the same size so you can flip back and forth and overlay for comparison without worrying that the graph is a different shape.
If you want bigger images just take a screenshot. "Greenshot" is a great tool for this, or just hit Win+Print Screen and have fun with whatever image editor you like.
There's a lot to line about to like about Omnimic software, and a lot to dislike as well. I have several complaints:
-pseudo noise and sine sweep are played at different levels. Annoying.
-pseudo noise has limited high frequency response, that could have been compensated for in software.
-lacking resolution of snapshots
-distortion plot should create average of several sweeps, this would significantly lower the noise floor for increased accuracy.
-"play from soundcard" has no options to play only left or right speaker, or bass removed tracks. Band-aid solution here is to simply ditch that lame feature, rip the CD to Flac files and play back using other software, making the whole experience a bit more clunky.
On the plus side, it has great response averaging and processing features, nice polar plot generator, and nice EQ features. And "absolute" SPL.
Two more minor tweaks resulting in a slightly lower crossover point and addressed the phase issues just above the crossover point. While still not perfect in the 4.5k range, it is much reduced. On this measurement axis, I lost a wee bit of reverse null. I swapped a 2 ohm parallel resistor in where it was a 2.5 ohm. Not a significantly measurable difference, but over several octaves it is noticeable and I prefer where it is at now. The distortion remains virtually unchanged, as both drivers across the crossover region are just too clean for me to measure in this space.
Some current squigglies.
In case anyone is curious about current crossover:
I do like Omnimic for the off-axis measurements, more or less.
Not too terrible for a non-waveguided essentially flat baffle (3/4" roundovers). The off-axis of the ribbon is pretty damn good, and as a system the directivity in the 3-5k region is offset by the dip in the on-axis. It sounds very balanced and smooth, and the off-axis indicates a wide sweet spot.
Here is one of those newfangled plots using the above measurements:
I use a phone app and a $8 lazy susan I bought at McMaster-Carr. One of my winter projects is to do some more work to that setup, with pins drilled at precise angles, or some other type of gauge. The phone app is likely not super accurate, but oh well.
I did ~ 0,2.5,5,7.5,10,15,30, and 45 degrees.
Ill spend more time on this crossover tomorrow, see if a good nights rest sharpens my hearing. Sometimes we have fatigue and are unaware of it.
When you load all those off-axis curves, under the AddedCurves menu, select "normalize" and pick the on-axis measurement file again. Hit OK on the "1kHz" dialog. This will remove all the response wiggles and show you only the difference between measurements for a better picture of the polar response.
Yeah. I like OM Polar "heat map" for off axis. For me, it's much easier to see problems rather than the squiggles. I used this for several years before I discovered that you usually want to normalize.
Agree with dcibel - OM distortion runs are barely usable. The program really needs an update, but I suspect PE paid Bill a one time fee, and there's no incentive to update it.
But Chahly - Stahkist don't want speakers that look good, Stahkist wants speakers that sound good!
Comments
So the rising response on the tweeter proved to be difficult to get a low crossover point - I wanted to hit 1.8-2K. It is problematic in that area. Not a criticism of the tweeter as much as it is a limitation of the mostly flat baffle and lack of wave guide. What I should do is rebuild the cabinets and use a wave guide or at least go to active and eq the top octaves flat. No big deal, I just have some moderate directivity issues.
Here is on-axis and reverse null of my current crossover. I am still deciding if it is what I want it to be, so I am resting my ears for the next few days and follow up switching between some decent cans and these speakers.
..and a distortion plot. 3rd and 4th above ~250Hz is largely below my noise floor at fairly loud volumes. The distortion rises below that, probably a combination of the room and the small woofer. Above that though, it is essentially non-measurable in my space. I had to bridge the APA150 to get this measurement, the little guys are pretty power hungry! The 2nd order dominant could very well be the amp, I don't know. I don't have the energy to set up a different amp to check things.
The crossover is currently eight components. Damped second order on the woofer with a small cap across the coil. Second order on the tweeter with a split L-Pad.
So far, I am pretty happy with things. I picked up an Omnimic and am using this project to learn it. A little clunkier than ARTA, but a few different features. I still prefer the distortion measurement in ARTA.
Anyways, will periodically revisit over the next week or so before I test in stereo and definitely before I commit to soldering.
Good thing about that Motus is that you can run it as high as you want without worry of distortion issues, only power response will be affected as you can see in your own testing, no issue around the crossover frequency
Yeah, it is really solid to over 3k.
So a few changes here and there, got the crossover lowered a bit to ~2700hz with a better reverse null FWIW. Spent a ton of time listening to them yesterday, and some more this morning. I will likely run with this version.
There is some mild phase issues just above the XO point, but it is minor and I am really not going to chase it down - I don't believe it is worth it.
Distortion is still pretty nice:
I may end up going active at some point to be able to take advantage of this tweeters ability to cross low for a ribbon, I think with the right time delay and contouring I can get a 2nd order LR at 2K.
Also installed a WiFi card on the measurement rig, so no more running upstairs to get on the internet. It also has BT now. Fancy.
Just for S's and G's, here is the measurement in Omnimic.
Omnimic plays noticeably quieter than ARTA. I cannot seem to figure that out, like the test tracks are set too low in level or something. Weird. I have to really crank the volume knob on the amp to equal the output in ARTA - has anyone else run into that?
Another nitpick is the snapshot function outputs a pretty small file - is there a setting for that somewhere? I just did a screenshot of the above.
Here is what I am talking about. Both images used the built in PNG export, the ARTA image is of much higher quality. I opened them both in picture viewer, maximized to my screen, and printed said screen and saved out using Paint. I only really give a shit because it is just easier to identify stuff in the ARTA screen shots.
So far I am almost happy with the Omnimic - it is fairly decent software. If I could figure out why the playback levels are so low and if I can output images with a meaningful resolution I would be pretty well content. I like a lot of the features baked in, for sure.
Looks great JR. I'm guessing they sound as good as the squiggles look!
My images from Omnimic are not small, and usually are .bmp, so they are not small files either.
Lately I've been saving with a smapper to .jpg so they aren't so large in filesize and easier to post without conversion.
InDIYana Event Website
They are pretty nice, to be honest. I might take the tweeter down a notch, though. It has very good off-axis response which I think might be contributing to a bit of perceived top octave energy. The sibilance is spot on where it is, though, so I'm not sure if I will change anything.
Omnimic outputs a standard file format for "snapshots" which are fairly low resolution. On the plus side, it does ensure that all snapshots are the same size so you can flip back and forth and overlay for comparison without worrying that the graph is a different shape.
If you want bigger images just take a screenshot. "Greenshot" is a great tool for this, or just hit Win+Print Screen and have fun with whatever image editor you like.
There's a lot to line about to like about Omnimic software, and a lot to dislike as well. I have several complaints:
-pseudo noise and sine sweep are played at different levels. Annoying.
-pseudo noise has limited high frequency response, that could have been compensated for in software.
-lacking resolution of snapshots
-distortion plot should create average of several sweeps, this would significantly lower the noise floor for increased accuracy.
-"play from soundcard" has no options to play only left or right speaker, or bass removed tracks. Band-aid solution here is to simply ditch that lame feature, rip the CD to Flac files and play back using other software, making the whole experience a bit more clunky.
On the plus side, it has great response averaging and processing features, nice polar plot generator, and nice EQ features. And "absolute" SPL.
I use MWsnap.
InDIYana Event Website
Two more minor tweaks resulting in a slightly lower crossover point and addressed the phase issues just above the crossover point. While still not perfect in the 4.5k range, it is much reduced. On this measurement axis, I lost a wee bit of reverse null. I swapped a 2 ohm parallel resistor in where it was a 2.5 ohm. Not a significantly measurable difference, but over several octaves it is noticeable and I prefer where it is at now. The distortion remains virtually unchanged, as both drivers across the crossover region are just too clean for me to measure in this space.
Some current squigglies.
In case anyone is curious about current crossover:
I do like Omnimic for the off-axis measurements, more or less.
Not too terrible for a non-waveguided essentially flat baffle (3/4" roundovers). The off-axis of the ribbon is pretty damn good, and as a system the directivity in the 3-5k region is offset by the dip in the on-axis. It sounds very balanced and smooth, and the off-axis indicates a wide sweet spot.
Here is one of those newfangled plots using the above measurements:
I use a phone app and a $8 lazy susan I bought at McMaster-Carr. One of my winter projects is to do some more work to that setup, with pins drilled at precise angles, or some other type of gauge. The phone app is likely not super accurate, but oh well.
I did ~ 0,2.5,5,7.5,10,15,30, and 45 degrees.
Ill spend more time on this crossover tomorrow, see if a good nights rest sharpens my hearing. Sometimes we have fatigue and are unaware of it.
Good night, my brothers.
When you load all those off-axis curves, under the AddedCurves menu, select "normalize" and pick the on-axis measurement file again. Hit OK on the "1kHz" dialog. This will remove all the response wiggles and show you only the difference between measurements for a better picture of the polar response.
Good to know! I'll do them over tomorrow.
Yeah. I like OM Polar "heat map" for off axis. For me, it's much easier to see problems rather than the squiggles. I used this for several years before I discovered that you usually want to normalize.
Agree with dcibel - OM distortion runs are barely usable. The program really needs an update, but I suspect PE paid Bill a one time fee, and there's no incentive to update it.
For the most part, how we interpret data is largely habit. I'm still learning how the heat map works for audio.
Another fun OM feature to play with is the weighted average. It's useful for making a "listening window" response of sorts.
First stereo session.
Imaging is almost binaural at this distance and toe-in.
Really nice!
Disregard.
How are you liking that Motus driver now that you've lived with it for a while?
It's better than most.
I've got a pair on the way.
You will not be disappointed. I sure hope they go back into production again - I would love a pair of the 8".
Call Erich at diysoundgroup up. He has the remaining stock.
InDIYana Event Website
Oh, I've been listening with wire nutted crossovers for a long time. Finally soldered them up today.
I like simple crossovers.
Just did GRT-8-W with 2mfd, parallel notch and a resistor. Crossed a bit higher as it's a 3-way speaker.