The Joudas project

[rpb]

You don't have to sweep that loud if you don't want to. Try taking gated measurements. That will remove room reflections that affect the responses. I've only played with open back once, and the measurements were goofy. Speaker sounded nice though. Do you have xo parts on hand? If so, connect a couple WAG filters, and play a little. Measure off axis to the sides too. Forget "polars" for now. Get comfortable with measuring, and interpreting the results. At some point, convert your measurements to an frd file if you want to do a sim.

[trevordj]

You don't have to sweep that loud if you don't want to. Try taking gated measurements. That will remove room reflections that affect the responses. I've only played with open back once, and the measurements were goofy. Speaker sounded nice though. Do you have xo parts on hand? If so, connect a couple WAG filters, and play a little. Measure off axis to the sides too. Forget "polars" for now. Get comfortable with measuring, and interpreting the results. At some point, convert your measurements to an frd file if you want to do a sim.

Thanks, I will try messing with some gated measurements and some off axis measurements today. I have only a few crossover parts to mess with but I will take this test baffle through the entire process and order some parts just for the experience. I figure that I will need some parts on hand anyway at some point so anything I order will be useful at some point. -Trevor

[trevordj]

Ok, I was partially successful at best with my measurements yesterday. I followed the Vituixcad measurement guide closely taking gated far field measurements at 1m for both the woofer and tweeter. My reference time is determined by a loopback into my soundcard. The gate is 2ms to 5ms.



I did 10 degree horizontal measurements from 0 degrees to 90 degrees at 1m for both the woofer and tweeter in the test baffle.

Tweeter:



Woofer:



Woofer nearfield:



Woofer blended response using FRD response blender:

[trevordj]

In terms of a crossover for this test baffle, this is about the best could do. The impedance dips dangerously low but this combination of woofer and tweeter is challenging to work with:



A notch helps a bit with the transfer function and the impedance but it still looks ugly.

[LOUT]

In terms of a crossover for this test baffle, this is about the best could do. The impedance dips dangerously low but this combination of woofer and tweeter is challenging to work with

Instead of the series filter-to-ground at the end of the woofer's LowPass, how do things look if you go back to the simpler 4part version, but add ~1ohm resistance in series to the large capacitor to tame that woofer's peak instead? As-is, that resistance would need to handle a lot of power/heat though, so I think it's probably better to fix via other adjustments altogether.

EDIT: I might be misreading the schematic, but isn't that second/small cap in the LP supposed to be much larger?...like 30-40uf instead of 4uf to be effective?
It also looks like the first inductor is really small for the large capacitor after it which is probably causing the low impedance dip, I'd expect a 2mH coil or maybe larger...I'm not familiar with these woofers though.

[rpb]

Don't take this the wrong way, but I think your measurement setup needs some work. Or it may be the result of an open baffle. Did you position the speaker 4' from any wall, or floor? Can you change the gate on measurements you've already taken? Can you zoom in on the impulse, and identify the first reflections?

Here's a measurement. See the first reflection in the impulse? The gate needs to keep that reflection out.

[rpb]

There should not be wild swings in the response. Any room reflection will mess up the measurement.

What woofer is that? Will it be run lower than 300hz, or so?

[trevordj]

Instead of the series filter-to-ground at the end of the woofer's LowPass, how do things look if you go back to the simpler 4part version, but add ~1ohm resistance in series to the large capacitor to tame that woofer's peak instead? As-is, that resistance would need to handle a lot of power/heat though, so I think it's probably better to fix via other adjustments altogether.

EDIT: I might be misreading the schematic, but isn't that second/small cap in the LP supposed to be much larger?...like 30-40uf instead of 4uf to be effective?
It also looks like the first inductor is really small for the large capacitor after it which is probably causing the low impedance dip, I'd expect a 2mH coil or maybe larger...I'm not familiar with these woofers though.

I will admit I mostly start with a standard filter and then I noodle with it until it looks good. Here is what it looks like with what you suggested:



I appreciate the feedback. This remains very new to me so I am fumbling my way through it.

[trevordj]

Don't take this the wrong way, but I think your measurement setup needs some work. Or it may be the result of an open baffle. Did you position the speaker 4' from any wall, or floor? Can you change the gate on measurements you've already taken? Can you zoom in on the impulse, and identify the first reflections?

Here's a measurement. See the first reflection in the impulse? The gate needs to keep that reflection out.

Thanks for your input, no offense taken at all. I am posting this and fumbling through at the risk of looking foolish but with the benefit of learning. The speaker was 32" from the ground and 5' from the walls. Here is the impulse response. It is certainly less obvious to me where the first relection would be given the noise floor appears higher.





If I change my gate to very narrow than it looks more expected but I don't get very much usable data.





It looks like I need to redo the measurements, get the speakers further off the ground, and lower the noise floor.


The woofer is the Vifa NE180W-08. I am not planning on using it for anything, just using it to test. I am currently building a box for it through to see if that simplifies things. I appreciate the input.

[rpb]

See if this works. Highlight a small portion of the impulse, and hit enter. In HOLM, that's how you zoom in., and a right click returns it.

[trevordj]

See if this works. Highlight a small portion of the impulse, and hit enter. In HOLM, that's how you zoom in., and a right click returns it.

[trevordj]

Despite the dumbassery I am having a good time. I really appreciate all the input. I threw together a quick sealed box for the woofer so I can see to what degree the open baffle is contributing. I know I am going off the rails a bit compared to the project at hand but it is all tangentially related. I don’t want to mess with the final project until I can get this process down.

I need to practice my miters anyway so I did four sides with locking miter joints and the baffle and back went on with just miters. -Trevor

[Chris Roemer]

The longer the gate the lower the measurements are accurate to - EXcept for getting reflections included in.
NORMally, (if you've got 8ft ceilings) if you can get the mic 1m (40") from your tweeter (axis), w/4ft to the floor, and ceiling, AND nearby walls/"stuff", then you can gate @ 3.33ms, which (IIRC) should be good down to 300Hz. On a 3-way (maybe crossing near 300Hz) that doesn't really help you.
For lower FR data (unless you could get into a "gym" where you could double all those distances (mic 8' off floor) - and THEN get data to 150Hz, which still isn't much) most will take a "nearfield" measurement (1/4" off the dustcap) where being THAT close swamps any (quieter) reflection data. Problem is that then you lose baffle-step and diffraction effects, which you THEN need to simulate, for accuracy.

[trevordj]

The longer the gate the lower the measurements are accurate to - EXcept for getting reflections included in.
NORMally, (if you've got 8ft ceilings) if you can get the mic 1m (40") from your tweeter (axis), w/4ft to the floor, and ceiling, AND nearby walls/"stuff", then you can gate @ 3.33ms, which (IIRC) should be good down to 300Hz. On a 3-way (maybe crossing near 300Hz) that doesn't really help you.
For lower FR data (unless you could get into a "gym" where you could double all those distances (mic 8' off floor) - and THEN get data to 150Hz, which still isn't much) most will take a "nearfield" measurement (1/4" off the dustcap) where being THAT close swamps any (quieter) reflection data. Problem is that then you lose baffle-step and diffraction effects, which you THEN need to simulate, for accuracy.

Thanks Chris, so do you basically follow the late Jeff Bagby's approach of combining nearfield and farfield measurements as he described here:

http://audio.claub.net/software/FRD_Blender/White%20Paper%20-%20Accurate%20In-Room%20Frequency%20Response%20to%2010Hz.pdf

It sounds like I definitely need to get the speaker off the floor and further away from the walls for my farfield measurements. Should I continue to do these measurements in room or is it more beneficial to just put the speaker on a ladder, say, in my garage or outside and get longer gated measurements that way? Should I continue getting nearfield and farfield measurments and combining them with response blender or should I just be doing nearfield measurement and then correcting for baffle step and forget about farfield? I think I read that nearfield measurements are good to about 800hz but maybe I am not remembering that correctly.

This is good stuff, thank you!

[rpb]

Thanks Chris, so do you basically follow the late Jeff Bagby's approach of combining nearfield and farfield measurements as he described here:

http://audio.claub.net/software/FRD_Blender/White%20Paper%20-%20Accurate%20In-Room%20Frequency%20Response%20to%2010Hz.pdf

It sounds like I definitely need to get the speaker off the floor and further away from the walls for my farfield measurements. Should I continue to do these measurements in room or is it more beneficial to just put the speaker on a ladder, say, in my garage or outside and get longer gated measurements that way? Should I continue getting nearfield and farfield measurments and combining them with response blender or should I just be doing nearfield measurement and then correcting for baffle step and forget about farfield? I think I read that nearfield measurements are good to about 800hz but maybe I am not remembering that correctly.

This is good stuff, thank you!

Another way to do a 2-way is to measure from about 3' with the speakers woofer about 48" from the floor. That will give you the accuracy needed for a 2-way, and that's what I did when I measured the one in post #54. Some may suggest that the accuracy below 400hz is questionable. My 2-way xo isn't doing anything below 400hz. (Not much anyway.) I'm just looking at 300hz and up. If you just want to live in sim world, then maybe you need to take a bunch of extra steps. If you just want to make a xo using a sim as a ballpark, and tweak after measuring, then all that other work is not needed.