New OB project underway - need lots of help understanding woofer-midrange FR graphs..

[fjhuerta]

Hello all,

After months of planning, I finally cut some MDF and took some measurements.

My project is something really new for me. It's an OB / vented woofer hybrid, with a hybrid crossover: passive on the mids and highs, and active between those two and the woofers.

The OB is an MTM with two RS-150s 8's and an unbaffled Neo3 PDR. The vented woofer module has 2 RS-150s,



Those are a couple of SketchUp modules I drew before cutting wood, in order to have an idea about their size, etc.

After cutting wood and doing stuff, today I began measuring them.



There are a couple of things:

1) The panels are slightly larger than the final versions (by 2 cms)
2) The panels are slightly wider and with different offsets side by side (by 4 cms)

Everything else is as it should be - the woofers and tweeters are recessed, etc.

I only measured the upper OB module.

Here are the measurements I got from 0, 15, 30, 45, 60 degrees from the woofers.



I have an issue here. At 1,600 Hz.

When I try to design a crossover at 1.9 - 2.0 KHz, the optimal points for my drivers, the 1,600 Hz (dipole peak? Is that it?) shows its ugly face as a bump in the overall frequency response. Ack.

It's nasty.

My questions are: is it avoidable? How can I avoid it? Going to a wider baffle? Narrower? Must I play around with the crossover points?

Any help will definitely be greatly appreciated!

[Steve Henry]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

I like your design & assume you would rather not change the baffle width. What several OB designers have done is cross over the woofer - mid at different frequencies, maybe low pass 1400 Hz & the mid high pass at 1800 Hz & let the dipole peak (assuming that is what it is) fill in the gap. John Busch does this with his Manzanita design.

[fjhuerta]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Hello Steve,

Yeah, I'd rather not change the shape. According to ABCDipole, I can't go wider than 20 cms. I re-measured the baffle and it's closer to 28 cms, so I'm hoping that when I do the final cuts the dipole peak will be somewhat less prominent.

I did some sims with your ideas, and the dipole peak does go away. I also went from a 4th order L-R to a 6th order L-R and dropped the frequency to 1.6 KHz and it looks even better. Impedance is OK, too.

Something that made me wonder too is that the peak is also obvious on the tweeter response plots, so I'm now sort of positive it's a dipole peak due to the baffle width, and it's going to change when I cut the baffle.

[JasonP]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Right, adjusting the crossover point is what I did with the Wavefront, and what I am planning to do with the Boson. If you still make more measurements, I'd be interested to see how the woofers measure from the rear, as the basket Reflection/Compression/Lensing (RCL!?) has a serious effect on radiated energy. I ended up having to cancel my Wavefront design just because of that. I had the speaker sounding good, until you really let loose and turned it up and then "HONK HONK HONK" from a rear radiating peak :/ I plan on studying this more before I commit to the XO design on the Boson. When I look at the Orion, I can only imagine Linkwitz got away with it because he crossed below that point for his Seas mid (1400 hz).

[JasonP]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Yeah, I'd rather not change the shape. According to ABCDipole, I can't go wider than 20 cms. I re-measured the baffle and it's closer to 28 cms, so I'm hoping that when I do the final cuts the dipole peak will be somewhat less prominent.

To my knowledge, changing the width will only move the baffle peak up or down, not mute it at all. So its kind of a matter of getting it where you want it

[Paul Carmody]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

To my knowledge, changing the width will only move the baffle peak up or down, not mute it at all. So its kind of a matter of getting it where you want it

Bingo. The only way to get rid of the dipole peak is to put a box around the speaker

I ended up using a notch filter to tame the dipole peak on the Sunflower Redux.

[fjhuerta]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Yeah... the thing is, the dipole peak is tamed.

On-axis, that is. I can tame it easily.

Once I start moving off axis, things get ugly. At 15º everything looks fine. 30º, I have some extra output at 1600 Hz coming from the tweeter AND the woofer. At 45º everything looks horrible.

I mean - I think this is logical. The further I get off axis, the more rear output I'm going to be listening to. I suppose that what's happening is - rear and front waves are cancelling higher up in frequency the more I get off axis.

The thing is, I can't remember seeing any FR graphs of dipoles where this is evident. SL's PHOENIX off axis plots do not show this behaviour, IIRC.

Is it normal? Should I just ignore it and consider that on-axis FR takes precedence over off-axis?

On the other hand, I've seen the graphs for Jamo's OB speaker, and they definitely shaped the FR. Did they do it to get rid of the dipole peak off-axis?

[JasonP]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Ok I spent a lot of time thinking on this and here is what I'm guessing. Perhaps someone with more acoustic know-how can disprove of shed better light on whats happening.

When you compare your baffle to the phoenix, you'll notice that you have a lot more empty baffle space around the woofer. Also you say above "I have some extra output at 1600 Hz coming from the tweeter AND the woofer" I think this tells us right here that since its affecting both tweeter and woofer its a function of the baffle itself, likely a gain from the surface given you are above the range that the baffle will function as a dipole cosine. If I get the wavelength equal to your board width, I get about 1100 and 1.5x that is 1600 or so. Coincidence? perhaps.

Well that is my thinking after chewing on it mentally for a bit.

Oh, and the drivers he is using in the Phoenix start to become directional at that point, so perhaps that is hiding the peak that would otherwise be there?

[JasonP]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Actually you could confirm if its the rear wave easily enough, even if you don't want to measure the rear output. I'd put a sheet of fiberglass batting over the rear of the drivers and measure again, see what changes :P

[fdieck]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

If I am not mistaken the RS 150 has an efficiency of 88 dB. Two in parallel gives 94 dB which is about 4 dB greater than the tweeter. Some of this may account for the peak below the crossover. I would use the second RS 150 in a 2.5 way configuration to extend the low end response only (see http://www.zaphaudio.com/ZDT3.5.html for an example.) and use as narrow a baffle that gets me an acceptable low frequency roll off to the woofers.

This also might be helpful for a baffle width suggestion since it uses about the same size drivers.

http://www.adireaudio.com/Files/DDRDipoleDesign.pdf

A very useful link.

http://www.dipolplus.de/frameset.htm?/thema11.php

[jhollander]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

I’m still a bit of a novice but my thought would be your x-o isn’t phase aligned. If you modeled in PCD filp the tweeter polarity back and forth and see if you get a bump. You want deep nulls such that the wrap around freq drives the response down.

[cjd]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

If the peak is in a bad spot, move it. Wider or narrower baffle, and remember wings count. Also, a baffle with everything centered and parallel edges focuses all the issue, so you can also look to spread the problem over a wider range.

Model it so its easier to get a hang of what does what, then come back and play.

Given your past work I suspect you'll manage crossover magic with aplomb. I do think some of the threads of JonMarsh's (isis in particular) cover a lot of ground on dipole... JohnK has also shared a ton of info here and there.

Reading without experiencing is different... reread and it'll make so much more sense.

C

[rpb]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

I'm wondering about the effect of the position of the top woofer relative to the top edge of the baffle. It looks to be about the same as the distance to the sides of the baffle, which could be an issue. Try clamping on a scrap of wood to the top edge to extend the baffle a few inches, and see if anything changes. You could also try cutting some off the top, but that would be harder to undo if it didn't help. You could design the x-over with a slight dip on axis, and eliminate part of the peak off axis. I doubt that it really matters all that much. If you were to measure at about 45 degrees vertically, you will have nulls due to the woofers cancelling each other, and from the tweeter and woofer cancelling. Here's some measurements on an MTM I built. These are verticaly off axis. Probably about 45 and 60 degrees.

http://i1221.photobucket.com/albums/dd467/rpb3/vertical.jpg

[fjhuerta]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Thank you for all your thoughtful responses.

When you compare your baffle to the phoenix, you'll notice that you have a lot more empty baffle space around the woofer. Also you say above "I have some extra output at 1600 Hz coming from the tweeter AND the woofer" I think this tells us right here that since its affecting both tweeter and woofer its a function of the baffle itself, likely a gain from the surface given you are above the range that the baffle will function as a dipole cosine. If I get the wavelength equal to your board width, I get about 1100 and 1.5x that is 1600 or so. Coincidence? perhaps.

Yep, it's the baffle. I just double checked the tweeter's FR and it doesn't drop at all.. not down to 500 Hz or so. That's pretty impressive for a Neo3 (basically impossible). So I guess I'm actually looking at the baffle gain. It'll change, though, when I cut the baffle to size, so I'm not that worried about it anymore - I believe that a smaller baffle will actually push the dipole peak up in frequency, which will help with driver integration.

If I am not mistaken the RS 150 has an efficiency of 88 dB. Two in parallel gives 94 dB which is about 4 dB greater than the tweeter. Some of this may account for the peak below the crossover. I would use the second RS 150 in a 2.5 way configuration to extend the low end response only (see http://www.zaphaudio.com/ZDT3.5.html for an example.) and use as narrow a baffle that gets me an acceptable low frequency roll off to the woofers.

Quite right. 2 RS-150's have more output than a single Neo3 - even worse when it's unbaffled. The interesting thing is, when I eliminate the dipole peak and shape a bit the response using a 4th order electrical x-over I get pretty nice looking curves which integrate pretty nicely with the Neo3. It has a slowly rising response in the bass, which I have learned I like a lot.

I’m still a bit of a novice but my thought would be your x-o isn’t phase aligned. If you modeled in PCD filp the tweeter polarity back and forth and see if you get a bump. You want deep nulls such that the wrap around freq drives the response down.

The issue isn't a x-over related one.. it's more of a baffle width issue. When I modeled the prototype x-over I checked phase alignment first, and impedance second ;)

If the peak is in a bad spot, move it. Wider or narrower baffle, and remember wings count. Also, a baffle with everything centered and parallel edges focuses all the issue, so you can also look to spread the problem over a wider range.

Model it so its easier to get a hang of what does what, then come back and play.

Given your past work I suspect you'll manage crossover magic with aplomb.

Thanks C! :D I suspect things will look better with the final baffle. The crossover seems simple, yet uses a lot of parts at this point. But I need to take measurements with the finished speaker.

I'm wondering about the effect of the position of the top woofer relative to the top edge of the baffle. It looks to be about the same as the distance to the sides of the baffle, which could be an issue. Try clamping on a scrap of wood to the top edge to extend the baffle a few inches, and see if anything changes. You could also try cutting some off the top, but that would be harder to undo if it didn't help. You could design the x-over with a slight dip on axis, and eliminate part of the peak off axis. I doubt that it really matters all that much. If you were to measure at about 45 degrees vertically, you will have nulls due to the woofers cancelling each other, and from the tweeter and woofer cancelling. Here's some measurements on an MTM I built. These are verticaly off axis. Probably about 45 and 60 degrees.

The final baffle will be different... the woofers are almost flush with the sides of the speaker, and there'll be 15 cms. between the top driver and the edges of the baffle. So yeah, I hope the peak will change - for the better. And I'll definitely take into consideration the dip on axis. I find it beneficial on plenty of recordings.

[fjhuerta]

Re: New OB project underway - need lots of help understanding woofer-midrange FR grap

Just wanted to quickly update this thread.

The baffles / boxes are done. The woofers and tweeters are mounted. I painted the mounting plates of the Neos in gloss black. I modeled the crossover. Things are looking VERY good! Cutting the baffles to size made all the difference in the world!

Just need to choose the X-over freq between the mid array and the woofer box. I've modeled everything from 200 to 120 Hz. It all looks beautiful. I thought I needed something higher, but the Jamo Open Baffles use a single 5 1/4" mid and they x-over to the 15" woofers at 180 Hz. So two 5 1/4" should probably go an octave lower.

Pics and everything will be coming soon...