new 2 way

kenrhodes

Working on a new 2 way. Why does the tweeter (yellow) show more output than the system(blue)?

kenrhodes

Also a sanity check would be great. All measurements were taken at one meter on axis.

Thanks

dcibel

Click those phase buttons and the answer will be revealed.

Also doesn't look like you've applied any acoustic offset for either driver, S1 and S2 don't show a little m next to them.

Sanity check - doesn't appear that you've gated the measurement, maybe it is Omnimic blended response though which is fine. If you haven't gated the measurement, I wouldn't use that for crossover design.

kenrhodes

Blended responce, it was gated at 2.5 ms. 
Can you point me to a good read on acoustic offset?  These drivers are very close together. I hope that helps.

dcibel

kenrhodes said:

Blended responce, it was gated at 2.5 ms. 
Can you point me to a good read on acoustic offset?  These drivers are very close together. I hope that helps.

Jeff B has some fairly decent reads on the subject. Here's his paper on determining acoustic offset.

https://drive.google.com/file/d/0B-GpgK57lLMNMWYwajNHWjduaE0/view?usp=sharing

The paper describes the process using PCD, but you don't have to use PCD. IIRC, WinPCD has included some built-in features for this purpose. The biggest difference here between PCD and Xsim is that PCD allows you to enter x and y and z offset seperately, where Xsim wants only the relative difference in distance from speaker to mic, but that can be determined in the same way, just remember that this is not the Z value when doing the same in PCD. In fact you can follow the same steps and just use some simple Pythagorean theorem to determine the Xsim value.

Something that might be helpful as well is the paper on splicing near and far field measurements for a nice full frequency response:

https://drive.google.com/file/d/0B-GpgK57lLMNOTljaGdrNl83WW8/view?usp=sharing

If you go look through the help file in the latest version on Omnimic there is some instruction there on how to splice near and far field measurements right in the Omnimic software as well.

Billet

You could try either going first order on the woofer or third order on the tweeter and see if the two drivers are closer to being in phase.

dcibel

First step is to make sure the relative phase is accurate so the simulation matches reality.

Wolf

C3 appears to be shorted.

jr@mac

Wolf said:

C3 appears to be shorted.

Good catch.

kenrhodes

Billet said:

You could try either going first order on the woofer or third order on the tweeter and see if the two drivers are closer to being in phase.

The woofer break up is impressive. I couldnt get it culled with first order. The break up peeked at 12k and 10db above average.

kenrhodes

jr@mac said:

Wolf said:

C3 appears to be shorted.

Good catch.

  I don't see how its shorted to ground. Please show me what I did wrong, new to xsim.

jr@mac

Not shorted to ground, just shorted. If you look, your schemo shows a direct path across C3, a cap should always show open.

hifiside

They mean you have drawn a line through the cap-C3. Remove the line.

6thplanet

kenrhodes said:

Blended responce, it was gated at 2.5 ms. 
Can you point me to a good read on acoustic offset?  These drivers are very close together. I hope that helps.

Not sure if you meant it, but the way you answered not having your acoustic offset right with they are "close together" Your confusing that with center to center. Acoustic offset being the front to back distance of the origin of sound from the drivers.

rjj45

kenrhodes said:

Also a sanity check would be great. All measurements were taken at one meter on axis.

Thanks

Dude, if you want a "sanity" check, I'm the wrong guy to ask <grin>
Couple of my tips from the school of "why is this darn thing messed up?"

Jeff Bagby says that he does all of his measurements at 25 inches. I use that a lot, sometimes a bit farther.
If you do nearfield measurements, baffle diffraction is eliminated, so you can determine if screwy responses around 700-2K are due to your crossover / driver responses or a baffle effect.
For a typical USA room with 8 foot ceilings you can typically set your gate about 4.5 - 5ms, but that really only affects how cleanly you can measure 400-700 Hz, so it's not critical. Sometimes I will raise a TM up a foot or so to get the tweeter equidistant from the ceiling and floor and can bump up the gate a bit.
Use Jeff's method for determining acoustic offset, but don't obsess on it. Sometimes the curves just don't line up perfectly.
I LOVE Xsim. Wow.
I usually start with textbook 2nd order on the woofer and 3rd order on the tweeter, then play with the woofer stuff if I need to suppress breakup nodes. Jeff showed a really neat "trick" on a couple of designs where he used 2nd order on the woofer and then used a small inductor to ground underneath the cap to suppress breakup. I've used that a few times. 
Don't obsess with crossover phase matching at first. Get your overall FR pretty good and make sure that you aren't stressing the tweeter. You can and will use 1/12 octave and 1/24 octave OM measurements to fine tune the crossover, but it can be very useful to drop to 1/6 octave or even 1/3 octave measurements to look the the big picture and double check your octave to octave relative balance.
Finally, at the end, line up the phase at the crossover and try for a good reverse polarity notch. I usually find that Xsim and my Omnimic don't match up perfectly, but at this point, OM and some tweaking will get you home
You can do off axis responses at this point assuming your crossover a bizarrely complicate topology, they should be fine.
Then start intensive listening tests. Tweak as needed.
That's my process. Have fun!

dcibel

rjj45 said:

Jeff Bagby says that he does all of his measurements at 25 inches.

The rule of thumb would be around 4x the baffle width. 25 inches might be okay if all you ever measure are 6" midwoofers on a baffle that's not much wider than the driver. You want to be as close as possible in order to keep the gate as wide as possible, but still far enough away that your far field response captures the baffle diffraction. The bigger the speaker, the more difficult it becomes to get a good clean measurement indoors. 

jhollander

I remember measuring some 6 foot line arrays that were shaded. I had to measure at 2 meters to get a decent representation of the outside drivers, but I still gated at the first reflection.

kenrhodes

Wolf and JR thanks for seeing how dumb I was. Hifiside thanks for spelling it out for me.
6thplanet, they are both, the  offset and the CTC should be close. I will work on getting a better z dimension.

kenrhodes

took three measurements trying not to move anything and then changed the offsets. .75 inches:

kenrhodes

kenrhodes

This is as close to a perfect overlay that I could get.  this was at 25 inches summed responce.

kenrhodes

jhollander

Looks like you need a bit more offset.  Keep the tweeter at 0 then adjust the woofer to get the match.

kenrhodes

The tweeter was at .001inches just enough to get the "m". on the woofer side if I changed up or down at all the responce got worse. I fear if this is too far off I will need to remeasure.

kenrhodes

I have a slightly better responce out of this sim. I think its time to build. (correct me if I'm wrong)

kenrhodes

Thanks for all the help so far. If nothing I 'm having a great time learning.

jr@mac

What is your impedance doing above 20k?

rjj45

dcibel said:

rjj45 said:

Jeff Bagby says that he does all of his measurements at 25 inches.

The rule of thumb would be around 4x the baffle width. 25 inches might be okay if all you ever measure are 6" midwoofers on a baffle that's not much wider than the driver. You want to be as close as possible in order to keep the gate as wide as possible, but still far enough away that your far field response captures the baffle diffraction. The bigger the speaker, the more difficult it becomes to get a good clean measurement indoors. 

Thanks for the input. I'm pretty sure that you are correct. Jeff mostly designs 6 inch TMs, so that would work well for him. I remember reading some Emininence white papers that stated empirical and physics reasons to use a pretty long distance for measurements, and they based that on driver diameter I think.

rjj45

kenrhodes said:

I have a slightly better responce out of this sim. I think its time to build. (correct me if I'm wrong)

Looks like L1 is not actually connected to ground. Is that correct?

kenrhodes

rjj45 you are correct the inductor didn't improve the responce at all so it is not connected. 
JR iI will have to measre with a different program xsim is locked out at 20k.

jhollander

wrt the offset my comment is to get the high frequencies to fit the best and ignore the lower frequencies where the room can mess with the response.