Nearfield/Farfield Merge and Baffe Step

[gmikol]

Okay, so I'm in the process of taking my first set of real measurements, and I'm puzzling over something that I haven't seen addressed. Feel free to tell me I'm way off base on this one.

The farfield gated measurement has a limited low-frequency accuracy, something like 250 or 300Hz in typical home measurement setup.

The nearfield response has an upper limit that's defined, approxmately, by the baffle width, perhaps 500 or 600Hz in many cases.

Unfortunately, for the baffle I'm modeling (approx. 7" x 14") the overlap region is right in the middle of the 4pi to to 2pi transition region (the "baffle step"). The gated farfield response takes the baffle step into account, but the nearfield response does not. It is, in effect, an infinite baffle over the limited frequency range, up to 500 or 600Hz or so.

No where I've read suggests that a baffle step / baffle diffraction simulation should be applied to the nearfield prior to merging. Intuitively, it seems to me like there needs to be some sort of compensation applied to the nearfield response prior to merging with the farfield response.

Can someone explain to me why this is not the case? I find impossible to believe that I'm right about this and everyone else is wrong, so I need to know where my error in thinking is.

Thanks--

Greg

[jkrutke]

No where I've read suggests that a baffle step / baffle diffraction simulation should be applied to the nearfield prior to merging. Intuitively, it seems to me like there needs to be some sort of compensation applied to the nearfield response prior to merging with the farfield response.

Your intuition is correct. If you want to merge a nearfield with a farfield, The nearfield should have a diffraction/baffle step sim applied. I do this with all merged response curves. If you don't you end up with curves that look like what stereophile does: a big hump in the bass.

The frequency that you use the nearfield up to is limited by driver diameter. Generally, you need to merge far below the typical repeating null that shows up as parts of the cone away from the mic go out of phase with the cone right near the mic.

[gmikol]

Thanks Zaph--

But I want to make sure I'm on the same page as you...My nearfield measurement was conducted with the driver mounted in the box. I know you do a lot of testing on an IB.

From Claudio Negro's site (which I've been using as a reference):

at this low frequencies the driver diaphragm acts like a rigid piston and the near-field measure is directly proportional to the far field one and is not affected by the environment in which the driver is.

He says that the NF response is "proportional" to the FF response, which would suggest that a baffle step compensation need not be added.

I'm looking forward to anyone else who might like to chime in on this, as well.

--Greg

[Curt C]

He says that the NF response is "proportional" to the FF response, which would suggest that a baffle step compensation need not be added.

I'm looking forward to anyone else who might like to chime in on this, as well.

--Greg

Nearfield measurement by definition cannot measure the diffraction effects (including baffle step) of the baffle. They will be 'proportional' to IEC baffle measurements. As John says, splice a simulated low frequency response which does take into acount the diffraction and enclosure below where your far field measurements become ambiguous. It is quite problematic to make good low frequency measurements, so using the modeled responses are in many cases the best solution.

C

[dlneubec]

Love the Avatar Curt. :T

You've got some mad scientist competition though. :W

[Jim Holtz]

Love the Avatar Curt. :T

You've got some mad scientist competition though. :W

I love it too. Does that fit Curt or what???? :T

Jim

[fjhuerta]

I always did the nearfield - farfield thing and then applied baffle step and diffraction... oops. :S

Would this work with smaller designs?

Anyway, since most of my speakers aren't really huge, what I discovered is that by raising my speakers about 2 meters from the floor and then measuring on an open space is that the first reflection occurs at around 150 Hz. So I can easily ignore it and forget about splicing

[gmikol]

Well, since it pretty much rains non-stop here from November to May, it's hard to measure outside.

I've been wondering if you could extend the frequency resolution by putting some absorbing material on the floor and hanging some from the ceiling above the speaker and mic to absorb the initial floor/ceiling reflection so the only issue is the back-wall and side-wall reflections, which could be as much as 10ms, if I choose my location carefully.

Will this work? What kind of material would be best? I've got a bunch comforters/quilts. Would that work? Or should I spend a couple of bucks on some OC703 or OC705 panels?

[Curt C]

Well, since it pretty much rains non-stop here from November to May, it's hard to measure outside.

I've been wondering if you could extend the frequency resolution by putting some absorbing material on the floor and hanging some from the ceiling above the speaker and mic to absorb the initial floor/ceiling reflection so the only issue is the back-wall and side-wall reflections, which could be as much as 10ms, if I choose my location carefully.

Will this work? What kind of material would be best? I've got a bunch comforters/quilts. Would that work? Or should I spend a couple of bucks on some OC703 or OC705 panels?

Good idea! Another possibility is ground plane measurements, which uses the floor boundary to advantage.

I highly recommend Joe D'Appolito's book 'Testing Loudspeakers' Its a great resource for speaker measurement techniques. -Dr. Joe uses a stack of foam for his in-room measurements IIRC, but fiberglass or wool batts, or yes, even quilts and comforters would work well also.

C

[augerpro]

How tall is your ceiling? This will determine the low frequency cutoff. I generally merge near and far one octave above this frequency. If your software is not able to model diffraction so it can be added to the nearfield measurement you can use one the FRD tools called SPL Combiner or something like that.

[fjhuerta]

Well, since it pretty much rains non-stop here from November to May, it's hard to measure outside.

I'm measuring indoors... home has quite high ceilings (around 5 meters) and the living room / playroom / dining room are all in a huge open space

[JonP]

I'm measuring indoors... home has quite high ceilings (around 5 meters) and the living room / playroom / dining room are all in a huge open space

There ya go... you could put them up 2.5m on a pedistal, and get the effect of being outside... if the side walls, reflective furniture, etc, could be that far away.

Looking at your impulse response, if you can zoom into it amplitude wise, you will see the reflection and know how much time (=distance) you have free. This of course, is the way to set your windowing.

You're pretty lucky to have such a huge room... now you need to build a safe and easy to use platform to raise stuff up on.

As for insulation, I'm sure it will work, but up to a point. I've read about various materiels, and they all have in common a reducing amount of attenuation as the frequencies get lower. So, as you go down lower at some point you'll be getting a noticiable return. But, it probably would help a lot in making the room seem larger than it is.

[gmikol]

Yeah...It looks like I've got a couple more things I've got to play around with. I've also been trying to merge my nearfield and port responses and I'm getting wacky results. I know I wasn't supposed to adjust mic gain between those 2 steps, but maybe I did anyway...Or maybe the port on these super-cheap bookshelf speakers that I'm playing around with really isn't doing anything.

I'm gonna have to let this thread languish, though...I'm about to get slammed by work and then I go on vacation for a week.

fjhuerta is lucky to have someplace with 5m ceilings and a big open space to test in...I bet that's one really nice house. :T

--Greg