[DannyBadorine]

I am making my own speaker that is using 2 M165's, one M130 and a BG tweeter.  I did some measurements today and I'm wondering if anybody can help me interpret this:
The mid range on axis response is showing some problems between 180Hz and 500Hz (I'm not worried about any other problems on there).  When I measure 15 degrees off axis those peaks and valleys flatten out to just about an acceptable reading.  I am measuring outside but there is a wall about 6ft. away and I'm thinking that might be the problem.  I do have the M130 in a sealed cabinet but it's tuned to 90Hz so I wouldn't think that this is causing the problem.  I'm attaching a graph showing the frequency test I did where you can see the on axis and 15 degree off axis response.  I also put up a graph of the entire speaker response with a crossover where you can see the problems between 180Hz and 500Hz (I'm not worried about any other issues across the spectrum at this point.)
Any ideas?  I'm open to all feedback on this. Thank you in advance.

[HAL]

Are you using a time gated frequency response measurement? If so, do you see any first reflections in the impulse waveform? 

[DannyBadorine]

Are you using a time gated frequency response measurement? If so, do you see any first reflections in the impulse waveform?

I am doing a frequency sweep from 20Hz-20KHz over 6 seconds.  I'm not sure how to use a time gated frequency response measurement but I will research it.  Unless there is an easy way to explain it to me.  I am using REW software.

[Hobbsmeerkat]

Got a picture of the speaker design you're measuring?
I have a feeling it might be an issue with edge diffraction. Which may explain things esp with off-axis being better, since the time arrival of those diffractions changes once you get off axis.

But I think pictures may help us get a better idea as well.

[DannyBadorine]

Got a picture of the speaker design you're measuring?
I have a feeling it might be an issue with edge diffraction. Which may explain things esp with off-axis being better, since the time arrival of those diffractions changes once you get off axis.

But I think pictures may help us get a better idea as well.

Yes, I'm thinking that it might be a baffle diffraction issue.  It's 8.5" wide and about 48" tall.  It's based off of the X-Statik.  Here's a picture of it in my pristine "anechoic chamber".  Yeah, it's my backyard.

[DannyBadorine]

 Here is another picture of it, but sideways......

[Hobbsmeerkat]

Okay yeah, thats edge diffraction. I would try rounding off the edges to either 1/2" or 3/8" and that should fix the issue. For a temporary fix, use a felt strip or a soft, thick fabric on the edges around the mid woofers and tweeter. Should also help smooth over the on-axis response.

[DannyBadorine]

Okay yeah, thats edge diffraction. I would try rounding off the edges to either 1/2" or 3/8" and that should fix the issue. For a temporary fix, use a felt strip or a soft, thick fabric on the edges around the mid woofers and tweeter. Should also help smooth over the on-axis response.

So edge diffraction happens in frequencies that low?  I thought it was higher frequencies. 
And thanks for your input, I will round the edges tomorrow and see how it helps.

[Hobbsmeerkat]

It's frequency is dependant on if the signal coming from the driver is beaming or radiating, and smaller woofers will begin radiating before larger ones.  From there the width of the baffle, and how sharp the edge is will determine its frequency and dramatically it shifts.
 
The peak is caused by the sound waves reflecting off the baffle, and the dip is caused by the sound waves wrapping around and bouncing off the side of the cabinet instead.

The sharper that edge and the wider the baffle is compared to the woofer determines the frequency at which it is caused, and how dramatic the peak & dip are.

Rounding off that edge softens the transition. For the size of your baffle, i would recommend the larger 1/2 round over.

The reason it smooths out at 15° off axis is because those reflections from the baffle are arriving at different times, and their phases are misaligned, so they neither sum nor cancel one another.

[DannyBadorine]

It's frequency is dependant on if the signal coming from the driver is beaming or radiating, and smaller woofers will begin radiating before larger ones.  From there the width of the baffle, and how sharp the edge is will determine its frequency and dramatically it shifts.
 
The peak is caused by the sound waves reflecting off the baffle, and the dip is caused by the sound waves wrapping around and bouncing off the side of the cabinet instead.

The sharper that edge and the wider the baffle is compared to the woofer determines the frequency at which it is caused, and how dramatic the peak & dip are.

Rounding off that edge softens the transition. For the size of your baffle, i would recommend the larger 1/2 round over.

The reason it smooths out at 15° off axis is because those reflections from the baffle are arriving at different times, and their phases are misaligned, so they neither sum nor cancel one another.

Awesome.  I understand this in principle and it's going to be good to see the change in practice.  I just thought that it happened more in the frequency range of the tweeter.  I will round the edges and measure again tomorrow.  Thanks for your help.

[DannyBadorine]

It's frequency is dependant on if the signal coming from the driver is beaming or radiating, and smaller woofers will begin radiating before larger ones.  From there the width of the baffle, and how sharp the edge is will determine its frequency and dramatically it shifts.
 
The peak is caused by the sound waves reflecting off the baffle, and the dip is caused by the sound waves wrapping around and bouncing off the side of the cabinet instead.

The sharper that edge and the wider the baffle is compared to the woofer determines the frequency at which it is caused, and how dramatic the peak & dip are.

Rounding off that edge softens the transition. For the size of your baffle, i would recommend the larger 1/2 round over.

The reason it smooths out at 15° off axis is because those reflections from the baffle are arriving at different times, and their phases are misaligned, so they neither sum nor cancel one another.

I used a 1/2" rounder and it improved the frequency response but the problems are still there.  I think they will be within +/-3db if I'm able to level out the 100Hz-500Hz range with the 500Hz-2k range.  Since those all fall within the mid-range woofer, would you suggest that this could be fixed with a baffle step circuit within the crossover?  It seems like 3db would do it.  Or would you try to compensate with the low pass part of the crossover of the woofer to bring the higher frequencies down to the low mid?
This graph is the frequency response of the whole speaker after a 1/2" rounder of the horizontal baffle edges.

[Hobbsmeerkat]

Hmm.. That definitely took the peak out of the response just over 400Hz, tho the dip is definitely still there, which is agree pretty sure is baffle step and would likely need to be corrected within the crossover.

Then you can use tho top range of the bass drivers to take up the slack below 200Hz

[Danny Richie]

Email me what you have on it for a crossover and I'll take a look at it.

[DannyBadorine]

Hmm.. That definitely took the peak out of the response just over 400Hz, tho the dip is definitely still there, which is agree pretty sure is baffle step and would likely need to be corrected within the crossover.

Then you can use tho top range of the bass drivers to take up the slack below 200Hz

I'm currently crossing the low end at 100Hz.  I am going to experiment with raising that crossover point.  I think that can also make the low pass of the mid-woofer steeper so that it takes more of the upper mid out and that will flatten the mid range.

[DannyBadorine]

Email me what you have on it for a crossover and I'll take a look at it.

I'm currently tri-amping it with a mini-DSP so I can design the passive crossover.  I have some ideas of what to try today.  When I get closer, I will let you know my results.  I'm also likely going to send you the final speaker so that you can also design a crossover for it. 

[Norman Tracy]

Two different Danny’s, oh this could get confusing. Oh well, thirty years ago I was part of an engineering team with five different Mike’s and Michaels’s. So I suppose we can keep our Danny’s straight.

It’s been great fun helping Danny Badorine with the 3D CAD and then CNC machining on the baffles for his project being discussed here. This fit perfectly into a project I had running to develop a wave guide for the Neo 3. Danny B and I agree that once results warrant the better of these wave guides should find their way to Danny R at GR World Headquarters. For now I am adding some pictures to aid this discussion, and because this is the internet and we all know without pictures it didn’t happen.

Version 1.0 featured a round 6.25” diameter wave guide the design I had cooking when Banny B’s project popped up on AC. It curves up the wage guide from a start in a circle and reliefs around the Neo 3’s output slots. This wave guide is on the baffle discussed above. Both versions differ from GR’s in that they are deeper using two layers of 0.75” MDF to get a wave guide 1.5” deep. Danny Badorine and I share an interest in increasing the depth in search of better time alignment yielding one hopes a simpler crossover.





Version 2.0 which Danny B just received yesterday features a rectangular 5.4” x 6.35” waveguide with straight flare dimensioned for equal distance from the Neo 3’s slots to wave guide end. I will not take offence if similarities to prior art by Raidho and Boerresen are noted. Also will not take offense if someone wants to sponsor a set of Børresen 01s or 05s to use as a benchmark for this project. Let me know if you’re feeling generous and I will get you a shipping address.








Finally as a bonus question, who can point out the CNC machining mistake on Version 2.0? Hint, an end mill took one for the team.

[Peter J]

Finally as a bonus question, who can point out the CNC machining mistake on Version 2.0? Hint, an end mill took one for the team.

And left part of itself in the waveguide there at about 11:00?

[Norman Tracy]

The panel of distinguished judges rules Peter J's answer is....correct!

The dark artifact at 11 o'clock is actually the point of a screw holding the second layer of MDF onto the baffle. The three other screws were not as deep so did not protrude into the cut path in their corners. The ball end end mill actually survived to finish the job, otherwise that baffle would not be in CA right now. Usually end mill vs steel screw is fatal for the end mill but because I hit just the point not the head or shank we could keep going. It would 'chirp' the three or four times contact was made during the roughing and finish passes but keep going. Tool path on the finishing stage was from the bottom up so everything seen above the screw point's level was cut post contact.

[DannyBadorine]

It's exciting to see that Norman posted some pictures here.  The testing was revealing today.  I have tested both of Norman's baffles and my original baffle that utilized the flush mount for the tweeter.  Here are the graphs.  Interestingly, the round baffle is giving the best on axis results.  I think we're going to end up with a version somewhere in the middle of the two that he has made.




I have also minimized the problems in the mid range that I originally posted about by adjusting the crossover points.  I've got most of the speaker to be close to +/-3db.  The low end is in phase and punchy.  I listened to music through the speaker and the instrument separation is really good.  I'm excited for the next step.

[mlundy57]

Don’t forget to check both the horizontal and vertical off axis response.