SevenSixTwo - InDIYana 2018 Coax Design

[KEtheredge87]

Why would you use near field for the coax horn? I would think the files to use would be:
Far field coax horn
Far field mid, blended with near field below 300Hz
Far field woofer, blended with near field below 300Hz

Also, for level I'd measure all at the same level, and try to find a balance somewhere between the woofer being too quiet and the tweeter being too loud.
PS... not too experienced with measurements myself either, so apply grain of salt and wait for some more experienced folks to chime in

Hey wogg, Now that you mention it... you're right. That doesn't make sense.

I had to go back and re-read Jeff Bagby's white paper (yet again) to straighten myself out here. Near-field is to remove standing waves and room reflections from the measurement. In the process, that creates an upper limit to the frequency range that we can use from the near field measurement. In this case, my upper frequency would be somewhere around 800 - 850 Hz, so very far below the range that horn is supposed to handle. I suspect your list is a correct way to approach the challenge.

Also, good idea on finding a middle ground for measurement dB level. Having different input volume levels in an attempt to match output dB level felt wrong last night.

[rpb]

Alright guys... midday update. I'm finally set up for some frequency response measurements in my basement! I followed Kevin K. 's advice and made myself a bigger base for the speaker stand, which helped tremendously. I used a scrap of 3/4" Baltic Birch from my Jedi Mind Trick subwoofer project that had a bunch of jigsaw cuts and odd profile shapes already cut through it. I also took the other Dayton Audio speaker stand I had and clamped it to the big-foot version to get this speaker higher off the floor. The Coax axis (say that 5 times fast...) is a bit more than half--way between the floor and the ceiling, so I feel like that's as good as I can ask for there. I also got my shipment of Rockboard60 2" thick mineral wool slabs in yesterday, so I set 6 of those around the area to do (hopefully) something positive to the impulse response in my space. The basement is still kinda empty, so there's plenty of slap echo and sound reflection if left alone.

This leads me to the next questions in my learning journey...
A) How the heck do I record frequency responses for a three way XO design??
B) What general volume levels should I be setting when my coax has HF / LF sensitivities of 103 dB / 92 dB and my woofer has 84.8 dB? I ask because It feels like setting one common volume for all measurements would make the coax HF too loud for near field and overload the OmniMic, while the woofer's LF would be too quiet for a clean far-field measurement.

All the examples that I've seen thus far assume a two-way speaker design. Since there are no three way measurement examples I have seen, my assumption continues that the methods must scale with the number of individual drivers in use.

This is the method I have in my head, and hopefully some kind soul either confirm this or guide me to the light

01) Set up 12 uF blocking capacitor in series with the coaxial HF horn driver to roll off around 1680 Hz (Cap value = 1,000,000 / (6.2 * Cutoff freq of 1680hz * Znom of 8ohm)
02) Set up for far-field gated measurements. In this case, 20" away from a 6.5" coaxial driver (microphone in line with the coax axis)
03) Measure far-field gated midrange measurements
04) Measure far-field gated woofer measurements
05) Set up for near field measurements (about 1/4" away from the coax and woofer dust caps, respectively)
06) Measure near field coax HF horn response
07) Measure near field coax midrange response
08) Measure near field woofer response
09) Attempt to calculate Z-offset for the coax driver with Omnimic... get confused... give-up on the OmniMic guided approach and just use PCD / X-sim
10) Measure in-box impedance sweeps with DATS V2 for each driver.
11) Use response blender and baffle modeler on the midrange and woofer measurements to combine nearfield and farfield for each driver, respectively
12) Extract minimum phase for each new file
13) Finally input blended FRD files and MinPhase ZMA files to PCD for XO modeling.

Am I on the right track, or do I have something wrong? And yes, the pictures of near field and far field measurements shows that I tried some stuff last night, realized I was tired and did it wrong (gated my near field of all things ) and figured it was time to ask for help!

Thanks for any help!

[ATTACH=CONFIG]n1365780[/ATTACH] [ATTACH=CONFIG]n1365783[/ATTACH] [ATTACH=CONFIG]n1365782[/ATTACH] [ATTACH=CONFIG]n1365781[/ATTACH]

I can help you make a x-over, but I do things a little differently than some of the other guys. I'll try not to clutter your thread, and post some ideas in this post.

Since it is so much more sensitive, measure your mid with the level turned down 20dB. Make your frd, but add back the 20dB with software.

Make a sim and measure the mid with the simmed x-over. Use the same amp setting as you used for the woofer when you measured it.

Start with very low spl sweeps, and use ear plugs in case something unexpected happens.

I usually measure from about 1m to 1.5m.

[KEtheredge87]

I can help you make a x-over, but I do things a little differently than some of the other guys. I'll try not to clutter your thread, and post some ideas in this post....

Hey rpb, thanks so much for the offer! Since this particular project is competing in the InDIYana 2018 event this year, I feel like it's only sporting that I give it my best shot at designing the XO without asking for direct help on the design itself. I felt that confirming the right measurement method wasn't getting unfair help, as I imagine the resources are out there on this forum or others if I spent enough time researching.

Having said that, I do intend to post my progress along the way, and I always appreciate feedback. At this point I have some XO parts on hand from my unfinished SuperBees 2-way project, and I ran an extensive bunch of simulations in PCD with traced FRD and ZMA files a few months back before making the driver selection. As for the XO points (and if I recall correctly) I found that the Dayton Esoteric 7" woofer modeled to be very well behaved up even beyond 900 Hz, so I think I last left the XO design crossing between low and mid somewhere between 600 and 900 Hz. The Celestion coax actually has a published suggested XO design from the manufacturer that recommends an XO around 2000 Hz. They're calling for 3rd order on the HF and 2nd order on the LF parts. I think I will use that as a starting point and see what improvements or changes my setup may require.

I'm off to try and make some better measurements of the setup tonight and see what I can get accomplished! Thanks again for your willingness to help a fellow speaker nut!

[KEtheredge87]

I'm trying to stay generic on how to measure, and not planning on any specific values, or rolloff rate for your design. It's just an approach. I'll delete or continue at your option.

All good sir. Sounds like you got my intent. Wanted to say thanks without sounding like saying "no please go away" 😉 -- Sent from my SM-N900V using Tapatalk

[isaeagle4031]

For taking sweeps on a 3way, this is my method. Mic on tweeter axis, about 1m. 90db sweep. Record tweeter, mid, woofer. Do not move mic or change the volume. Next connect tweet and mid, run sweep and record. Do the same with tweet and woofer. Mic position and volume stay consistent. Be sure to also capture phase response.

In PCD, enter the listening (mic) distance. Add in mid (as woofer) and tweet. Import overlay of combined response. Toggle z axis on the mid until the responses overlay as close as possible. Record that number. This is your acoustic offset for the mid.

Do the same with the woofer and tweet. You will also enter the y axis (vertical). Again record the numbers.

Take zma files for each driver on box. While the tweeter will not change, the enclosure will of course change the values for the mid and tweet. This will also show you the fb, if ported.

You now have the files necessary to design your crossover.

[jhollander]

01) Set up 12 uF blocking capacitor in series with the coaxial HF horn driver to roll off around 1680 Hz (Cap value = 1,000,000 / (6.2 * Cutoff freq of 1680hz * Znom of 8ohm)
02) Set up for far-field gated measurements. In this case, 20" away from a 6.5" coaxial driver (microphone in line with the coax axis)
03) Measure far-field gated midrange measurements
04) Measure far-field gated woofer measurements
05) Set up for near field measurements (about 1/4" away from the coax and woofer dust caps, respectively)
06) Measure near field coax HF horn response
07) Measure near field coax midrange response
08) Measure near field woofer response
09) Attempt to calculate Z-offset for the coax driver with Omnimic... get confused... give-up on the OmniMic guided approach and just use PCD / X-sim
10) Measure in-box impedance sweeps with DATS V2 for each driver.
11) Use response blender and baffle modeler on the midrange and woofer measurements to combine nearfield and farfield for each driver, respectively
12) Extract minimum phase for each new file
13) Finally input blended FRD files and MinPhase ZMA files to PCD for XO modeling.

Am I on the right track, or do I have something wrong? And yes, the pictures of near field and far field measurements shows that I tried some stuff last night, realized I was tired and did it wrong (gated my near field of all things ) and figured it was time to ask for help!

Thanks for any help!

[ATTACH=CONFIG]n1365780[/ATTACH] [ATTACH=CONFIG]n1365783[/ATTACH] [ATTACH=CONFIG]n1365782[/ATTACH] [ATTACH=CONFIG]n1365781[/ATTACH]

1. lose the cap
2. further back would be better, but watch the impulse response for blips
2.5 measure the tweeter first and set the volume level with the tweeter
4.5 measure tweeter plus mid before moving mic and volume
4.6 measure tweeter plus woofer before moving mic and volume
4.7 (Coax only) rotate the box on its front baffle axis 20 degrees and repeat the far field tweeter and mid measurements before moving mic and volume. You can use these FRDs later to see what your x-o does off axis.
5 add turn down the volume
6 skip nearfield on the tweeter
12 before 9. no need to extract min phase on the tweeter plus mid, woofer files

[KEtheredge87]

Thanks John and Chuck! I appreciate the tips on 3-way measurement techniques. I took a new set of measurements tonight based on my conversations with wogg earlier today, but I definitely like the thought process you guys have laid out. I didn't consider that I'd need combinations of tweeter + mid AND tweeter + woofer (although that seems like a no-brainer now that you've mentioned it!).

The whole "capacitor on the compression driver" bit came from me reading the OmniMic manual. I assume they tell people who are actually reading the manual to do that so they won't accidentally destroy a ribbon tweeter. I saw compression drivers listed in that same category, so I figured better safe than sorry. However, I trust that John knows when to apply that method and when it's not warranted, and I'd prefer NOT to have to do extra OmniMic work to add in sections of the frequency band that I rolled off after the fact.

Tomorrow I'll take another set of measurements (third time's the charm right?). Before that happens though, I may need to remove some foam from the vented cabinet. I took an impedance sweep of the woofer tonight, and Fb has mysteriously decreased from my expected ~31Hz down to ~27Hz. According to BassBox, that means I have either managed to squeeze a 19 inch long port inside my 15" deep box, or I went very heavy on the damping fill. Since that last port tuning exercise, the cabinet was glued together and some caulk was added to every inside seam. Otherwise the difference between now and then is an extra strip of foam inside the cabinet sitting on top of the slot port. I added this to make sure the testing wires didn't rattle and scrape around on top of the vent during measurements. Maybe this pushed me over the edge? BassBox's nonsensical 19" slot port idea says my F3 would go from a predicted 38Hz up into the low-mid 40Hz, not to mention cause my power handling to suffer. Adjusting the foam is the only real option at this point. There's no cabinet surgery to be done that won't destroy everything. Guess we'll see where this ends up!

[tomzarbo]

Thanks John and Chuck! I appreciate the tips on 3-way measurement techniques. I took a new set of measurements tonight based on my conversations with wogg earlier today, but I definitely like the thought process you guys have laid out. I didn't consider that I'd need combinations of tweeter + mid AND tweeter + woofer (although that seems like a no-brainer now that you've mentioned it!).

The whole "capacitor on the compression driver" bit came from me reading the OmniMic manual. I assume they tell people who are actually reading the manual to do that so they won't accidentally destroy a ribbon tweeter. I saw compression drivers listed in that same category, so I figured better safe than sorry. However, I trust that John knows when to apply that method and when it's not warranted, and I'd prefer NOT to have to do extra OmniMic work to add in sections of the frequency band that I rolled off after the fact.

Tomorrow I'll take another set of measurements (third time's the charm right?). Before that happens though, I may need to remove some foam from the vented cabinet. I took an impedance sweep of the woofer tonight, and Fb has mysteriously decreased from my expected ~31Hz down to ~27Hz. According to BassBox, that means I have either managed to squeeze a 19 inch long port inside my 15" deep box, or I went very heavy on the damping fill. Since that last port tuning exercise, the cabinet was glued together and some caulk was added to every inside seam. Otherwise the difference between now and then is an extra strip of foam inside the cabinet sitting on top of the slot port. I added this to make sure the testing wires didn't rattle and scrape around on top of the vent during measurements. Maybe this pushed me over the edge? BassBox's nonsensical 19" slot port idea says my F3 would go from a predicted 38Hz up into the low-mid 40Hz, not to mention cause my power handling to suffer. Adjusting the foam is the only real option at this point. There's no cabinet surgery to be done that won't destroy everything. Guess we'll see where this ends up!

Keith,
I'm sure you probably know this, but there is a 'bass removed' file on the Omnimic disc that removes the bass from the short sweep. I think it's #7 and #14 L and R on the old disc.... if you're concerned about overloading the tweeter. The full-range sweep doesn't seem to be an issue that I've noticed on any tweeter I've used yet, but I've only used an AMT tweeter other than the normal domes and cones so far. It's such a quick sweep I guess the voltage doesn't have a chance to damage the super-fine tweeter voice coil windings. It still makes me cringe sometimes because the sweeps seem pretty loud sometimes and I worry. I think I used the bass removed file on the AMT if I recall. Like John said, tweeter measurements first is good so you're sure you're getting enough volume for decent measurements, but not pushing it too far; a midrange and woofer will obviously always have less of a hard time with that sweep and be able to tolerate that burst no problem.

I still use Bass Box here and there and it really does make some rather large changes on low end response depending on the amount of fill specified. I always wondered if those were real-world swings, or just '0's and '1's misbehaving!

Sorry if I've just restated the obvious.

TomZ

[KEtheredge87]

Hey Tom, thanks for adding more experience to the discussion! I'm getting the feeling that Dayton Audio is trying to cover their legal bases a bit more by recommending a blocking capacitor on times where it doesn't seem to be needed. Ribbons and AMTs are a great time to use one since the nature of the transducer itself is much more sensitive, but it seems like a compression horn (especially one designed after the professional audio market) should be able to handle a few quick sweeps. I did know the OmniMic disc has bass removed tracks, although the instruction manual reads like I need to add some additional signal back to that measurement as a post-processing step. If I don't have to protect the driver that much, I won't bother with this and I'll go full range sweep.

As for misbehaving ones and zeroes, I am always suspicious! I'll be taking a bit of lining out from the vented cabinet and see what that does to my impedance sweep. I can use some bread ties to hold the woofer's wires off the floor for testing purposes, especially since I didn't really want to put foam on that surface anyway. I hope this is just real world physics being in-line with my gut instinct (generally a good thing in the engineering profession ).

Thanks for following along!

[wogg]

For taking sweeps on a 3way, this is my method. Mic on tweeter axis, about 1m. 90db sweep. Record tweeter, mid, woofer. Do not move mic or change the volume. Next connect tweet and mid, run sweep and record. Do the same with tweet and woofer. Mic position and volume stay consistent. Be sure to also capture phase response.

In PCD, enter the listening (mic) distance. Add in mid (as woofer) and tweet. Import overlay of combined response. Toggle z axis on the mid until the responses overlay as close as possible. Record that number. This is your acoustic offset for the mid.

Do the same with the woofer and tweet. You will also enter the y axis (vertical). Again record the numbers.

Take zma files for each driver on box. While the tweeter will not change, the enclosure will of course change the values for the mid and tweet. This will also show you the fb, if ported.

You now have the files necessary to design your crossover.

See Keith, told ya someone would have better / more detailed instruction than me

This is great, didn't think about nailing the z offsets for the design and this is really simple.

Tomorrow I'll take another set of measurements (third time's the charm right?). Before that happens though, I may need to remove some foam from the vented cabinet. I took an impedance sweep of the woofer tonight, and Fb has mysteriously decreased from my expected ~31Hz down to ~27Hz. According to BassBox, that means I have either managed to squeeze a 19 inch long port inside my 15" deep box, or I went

very heavy

on the damping fill. Since that last port tuning exercise, the cabinet was glued together and some caulk was added to every inside seam. Otherwise the difference between now and then is an extra strip of foam inside the cabinet sitting on top of the slot port. I added this to make sure the testing wires didn't rattle and scrape around on top of the vent during measurements. Maybe this pushed me over the edge? BassBox's nonsensical 19" slot port idea says my F3 would go from a predicted 38Hz up into the low-mid 40Hz, not to mention cause my power handling to suffer. Adjusting the foam is the only real option at this point. There's no cabinet surgery to be done that won't destroy everything. Guess we'll see where this ends up!

Definitely lighten up the fill, or go with a little foam lining on the walls. Hopefully that helps. If it doesn't you'll effectively have a bit of an EBS alignment, which isn't a huge deal. It'll still sound good and possibly get you a hint of the deeper notes, but run out of steam under power quicker.

[KEtheredge87]

Well... Tonight the engineer in me is annoyed. I did some quick studies on both the sealed coaxial section and the vented woofer section with DATS. I measured different levels of poly fill stuffing from gross over-stuff (16oz in 0.4 cuft) to totally empty while monitoring Qts (Qtc for sealed enclosure) in DATS. BassBox model said I should get around 0.78 for Qtc, but I couldn't measure lower than 0.89 when grossly over-stuffed. Of course that sounded awful even with the impedance sweep. I settled on a natural sounding impedance sweep with around 6oz of fill for a Qtc of about 0.91. I'm sure this will be fine, but I get perplexed when the analysis doesn't match reality well enough. Similarly, I took some foam lining out of the vented cabinets to see if my Fb would rise back to 31hz. I got excited when removing the foam from the top of the port and the ceiling in that chamber measured an Fb of 32hz while I held the driver in place by hand. Unfortunately the moment I put the screws back in place for the real seal the Fb dropped right back down to 27.8 Hz. I'm having trouble modeling a vented cabinet with my port geometry in BassBox that ALSO has an Fb of 28Hz. BB insists my Fb should be 31Hz. I am probably making a mountain from a mole hill, but again, it bugs me when I can't get the simulation tools to match reality. Guess I'm putting too much faith in the underlying assumptions that are used in these programs. Since the only way to bump the Fb up at this point is to shorten the port, I'll be living with them as they are and seeing what I can get from XO designs. I was t setting out to make an EBS alignment for this cabinet, and the idea of decreased power handling doesn't thrill me since the esoteric driver has fairly high power handling specs. It feels kinda wrong not to let that sucker slam when it can, but I'd have to rebuild the cabinets from square one to change the port now. Guess that's what I get for trying a slot port! -- Sent from my SM-N900V using Tapatalk

[isaeagle4031]

If you have the room, stuff a couple pieces of 2x4 or similar to the walls (Just lay it in there for testing) thus reducing the volume. That will raise the fb as well.

[rpb]

If you have the room, stuff a couple pieces of 2x4 or similar to the walls (Just lay it in there for testing) thus reducing the volume. That will raise the fb as well.

Styrofoam would work too. Odd shapes and locations would have some advantages in breaking up internal reflections.

[jhollander]

You might try comparing to another box modeling program. I like unibox.

[KEtheredge87]

If you have the room, stuff a couple pieces of 2x4 or similar to the walls (Just lay it in there for testing) thus reducing the volume. That will raise the fb as well.

Thanks for the suggestions Chuck and rpb. I just finished fiddling around with the cabinet some more. I tried sticking some 2x4 and MDF pieces in there up to a ~20% reduction in overall chamber volume. Unfortunately the Fb doesn't seem to be budging much. My initial volume of 0.65 ft^3 was giving me Fb=28.3Hz, where the final volume I tested so far was 0.495 ft^3, with Fb = 28.9 Hz.

I even removed my 3-way test wire panel assuming I had some leaks there and installed the actual terminal cup for the cabinet just to be sure, and no changes. BassBox seems to think my Fb should have raised a full 4 Hz by this point, so I'm losing confidence in that program more and more. I will have to fiddle around with Unibox or WinISD in a bit to try and check the math.

Assuming my DATS hardware/software hasn't gone wonky, at least the box's Fb is stable regardless of the amount of crud I put in there.