XO Design for MDF MCM MTM

ani_101

XO design for the MCM MTM.

I tried out a preliminary XO and am getting third order for tweeter and fourth order for woofer. Need a quick reivew on the XO and if possible any optimizations of suggestion for improving. The XO file is attached in this link, please feel free to play with it: 
https://drive.google.com/open?id=0B2bbY-YB6ikpWk13VVhxc1FEQVE

The XO point is 2k, the tweeter distortion measurement is here: https://diy.midwestaudio.club/uploads/editor/h8/lf5zgonqd6q7.png
Some discussion about the tweeter can be found in this link: https://diy.midwestaudio.club/discussion/341/broad-depression-in-tweeter-response-7-8khz

XO Screenshots:

jr@mac

Any particular reason you are not using any compensation for the baffle step, Ani?

ani_101

The measurement is taken free air, away from all wall (min 4 feet), so it that is needed then how do i add it? Just tilt the response down, i wasn't sure where and what to add due to the squiggles in the response all over, it is +/- 3-4 db all throughout.

ani_101

ok, added a bit of BSC, and have two versions. Please let me know which looks better from a graph point of view, or if i need to add more. What do you look for in a BSC and are there any rule of thumb?

V5:

ani_101

V6: Slightly different BSC. Is this better or we need something else....

jr@mac

I like the 2nd one better. 

PWRRYD

I agree with JR, the second one is certainly better, but in my opinion it will still be too forward and hard to listen to long term.  I know we as DIYers want to be as meticulous as possible and see (correct) every dip, blip, and peak;  i.e. super high resolutions on our measurements.  But sometimes a little good old smoothing on the measurements (like 24th octave) helps show the big picture.

jr@mac

Ani, quick tip is to look at your woofer transfer function curve. It should be about 6db down above 800-1000 or so. Transfer function isnt the absolute authority, but can help get you there. 

jr@mac

Then, as Craig says some smoothing and a gate would not hurt, either. 

ani_101

Can we smoothen in PCD? Or smoothen the measurement after XO? I'll try getting it down to 6 db... Might have to zoom in a bit.

When you say forward, does it mean the high are still playing higher and should be taken down a bit? I have a bit more downward tilt right to left on the charts?

ani_101

PS I do have a gate around 200hz...

ani_101

One more try.

V6 XO, more aggressive BSC, and the tweeter response is also tilted down. Now the issue is if I have all those tiny inductors on hand....

kennyk

Have you tried any of these xo's? Have you tried a 2nd order? I ask because I like 2nd order in a 2-way.
Hell, I just like 2nd order. 

ani_101

I did try second order, but couldn't keep the woofer breakup down enough. With the requirement to cross higher and squash the breakup I need a steep roll off on the woofer.

Are you asking for second order on woofer or tweeter or both?

ani_101

No I haven't tried these XO yet, but am putting it together today depending what parts I have on hand.

jr@mac

Ani, one old trick to reducing parts count while hammering a woofer breakup into submission is to use a circuit such as this:


Expect to do a little experimenting with that capacitor paralleled around the main woofer coil. You will likely find that a high parts count is not necessary for the woofer circuit. 

jr@mac

I know, my electrical CAD engineering chops are second to none. 

PWRRYD

Actually much clearer than Wolf's multi colored schematics

ani_101

Thanks JR, i'll try that and am working on a 2nd order on the woofer and 3rd on the tweeter.

6thplanet

What's up with phase? I see they track each other nice, but the rotation doesn't happen where the xover is happening? I've never seen that?

ani_101

6thplanet said:

What's up with phase? I see they track each other nice, but the rotation doesn't happen where the xover is happening? I've never seen that?

I don't think I understand that. What i am looking for and what should be happening?

jhollander

To get the woofer to 3K I like a small inductor for the knee then a larger inductor notch for the BSC.  I'd also do the tank on the break up, although this is not really possible to show in PCD.

The filter would look something like this, sorry I can't do legible sketches...

I also noticed the PCD measurement distance was set at 2 meters?

isaeagle4031

ani_101 said:

6thplanet said:

What's up with phase? I see they track each other nice, but the rotation doesn't happen where the xover is happening? I've never seen that?

I don't think I understand that. What i am looking for and what should be happening?

Most times the phase rotation (where the phase appears as a vertical line or lines) occurs at the crossover point. It's not unheard of not to be at thst point however. 

It can be an indicator of a LR alignment with the reverse null. Someone please correct me if my thinking is not on par.

jhollander

The raw driver phase appears to flip a few times which can be an indicator of non-minimum phase files.

PWRRYD

Yes you are correct Chuck.  The point of phase rotation on the graph isn't important.  The actual phase tracking of the drivers through the xo area is.  Reverse null is a good indicator with LR4 slopes, but is never as defined or deep and narrow with shallower slopes like LR2.  And my measured reverse nulls never look as pretty as my simulated ones.  Playing in "Sim Land" I find that the difference between 0.13 mH and 0.14 mH on a tweeter shunt can have a huge effect on the simulated reverse null.  But in real life (sorry Chris Romer) the difference is mute.

ani_101

jhollander said:

The raw driver phase appears to flip a few times which can be an indicator of non-minimum phase files.

It is actual ZMA taken with DATS for tweeter and the pair of woofers in parallel and then min phase extracted in response modeler before loading into PCD.

But I do think the mic distance is wrong, it is not 2 meters, i will measure the distance and correct it and try out JR's and your suggestion on the XO.

ani_101

ok, changed the PDC measurement distance to 1.32meters, which is the actual measured distance...

this is my second order o nthe woofer, breakup still stere, about 25db down, would have liked it further squished down, but will try adding a small cap across the L2 inductor and measure and change and measure and see if it makes a difference on the woofer response.

John, I am not able to follow your instructions in PCD. Could you kindly elaborate what i can simulate in PCD and which component I need to experiment with?

V8:

jhollander

Ani, if you use the notch in PCD you can not do the tank to my knowledge.  Try these values in the picture.

I'm sure it's a typo but you are extracting minimum phase from the FRD files. The minimum phase files and your x and y offsets and your measurement distance are entered in PCD to find your z offset.

isaeagle4031

Reduce the 2uf to .33 and the resistor input 99999999. Pcd will see it as infinite.

ani_101

OK, this is the final XO i got, that will (hopefully) make it tomorrow to SD along with the multi-colored wires. Thanks for the different pointers and tips. I combined some of the tips and changed values and measured and selected the one which looked good enough then did some listening again while changing, adding, removing components, and this one seemed to be the most promising.

Woofer XO:


Tweeter XO:


Measurement:

JasonP

I really thought these sounded great at DDIY, fwiw. They needed just a touch more of baffle step IMO, but where very close to perfectly balance. I apologize if I misheard you Ani and played my demo instead of JRs, but getting to listen to my own demo made it very clear to me you nailed the crossover integration.

I looked for a driver measurement and found this: https://www.hifisound.de/Do-it-yourself-Products/Speaker-Drivers/Low-Midrange/Gradient-Select-W-130-AL-8-Shielded-Aluminium.html It seems to have a downward slope already, hence the need for less baffle-step in the crossover.