Near-field mic position?

[Elbert]

Just finished my WM-61 based mic and ready to head off in to test-land with a fresh copy of HolmImpulse on my laptop!

For the woofer, a SEAS W17CY001, I plan to do nearfield measurements for the low-end and splice this to gated far field (40cm) to get the overall response.

After some experimentation I noticed that the LF response varied according to how far away from the centre I placed the mic.

Can anybody suggest what nearfield mic-response yields the most correct LF response?

PS as the W17CY001 has a phase-plug, it is of couse not possible to put the mic dead centre.

Thanks in advance!

[head_unit]

By definition (of I believe Mark Gander, in his original Audio Engineering Society paper where he invented this), the nearfield is where the measurement DOESN'T change when you move the mic.

You need to be almost touching the cone. If you move closer to the cone and the response gets a bit more SPL, you were not in the near field yet. Similarly if you move across the face of the cone.

However, this only applies at "low" frequencies, "low" maybe varying with the driver but say below 100 Hz tops.

Make the near- and far- measurements in the same physical layout, or you may have a very hard time splicing the measurements together. Vance Dickason's Loudspeaker Cookbook should have a section about this.

[Elbert]

Thanks for the reply!

Now, I did put the mic as close as possible to the cone, say 2-3mm.

What I found initially was that as I moved from the vicinity of the voice coil to the cone edge, the LF response seemed to droop off.

Even when measuring next to the voice-coil the roll-off didn't match the box simulations I did.. Hmmmm..
Then I switched from gated to raw response, and then the response was suddenly really close to the simulation results!

The far field (40cm) result was gated at around 600 Hz, and when I spliced these two measurements, the resemblance to the factory SEAS plot was darn close!!

Looks like I might be on to something here...

[Elbert]

Hmm.. after reading through the "Loudspeaker desing cook-book" some more, I read that Vance states that nearfiled measurements are only good up to about 100 Hz due to "change in radiation resistance".

Any comments on this?

[Ron E]

Originally Posted by Elbert

Hmm.. after reading through the "Loudspeaker desing cook-book" some more, I read that Vance states that nearfiled measurements are only good up to about 100 Hz due to "change in radiation resistance".

Any comments on this?

Do a search for Don Keele and nearfield measurements and you will find a paper that explains the method and limitations. Nearfield is only good at frequencies where the driver is essentially omnidirectional. When the driver starts to beam, you get rolloff and notches. Nearfield also does not measure diffraction, so you get a false sense of response shape unless your speaker is in an infinite baffle.

[head_unit]

As far as your measurements not matching your simulations, what program are you using? Most software does a simulation of the speaker as a basic RLC circuit, but that is crude and only good for comparing one type of box to another.

You need something really sophisticated like LEAP which will make a more sophisticated model-which also requires more exact measurements of parameters and impedance. With LEAP I could see good low frequency correlation of my measurements vs. my simulations.

[ucla88]

Practically speaking, 3-5mm is fine, midway between the center and edge for a phase plug driver.

There really should not be a dramatic difference between the sims and the nearfield. But, it's still a little tricky to get the LF measurement with very good accuracy. If you want a decent nearfield to 10hz, you need to get 200 or milliseconds for your impulse interval. It's a bit hard, sometimes in a smaller room to get this nicely, even using the nearfield technique. The sims don't need to be from LEAP etc, although that certainly will work. LF box modeling is not very complex-actually the models are probably easier and more correct than the measurement.

If you're getting 5-10dB or more variations in you low end curves just by moving 5mm side to side on the cone, then something is wrong with your setup/technique.

I agree with getting D.B. Keele's original paper from the 70's on this if you're really interested. You can get useful information well above 100 Hz usually, and you can expect/predict where the nulls will occur as well.

[Elbert]

Thanks for the replies!

Well, the experiments with nearfield I've performed so far actually seems to match the predicted response surprisingly well, so I haven't any reason as per se to question the results other than from the statement made by Vance Dickason in his book.

Thanks for the advice on that white paper, that's exactly the sort of thing I find usefull, understanding limits and possibilities.

As the speakers I'm working on are relatively small 2-way closed boxes, getting as low as 10Hz or even 30-40Hz isn't really that important.

Anyway, I've decided to do a continuous thread on the whole measurement/ speaker project, and this will undoubtedly be a positive contribution to that!

Thanks again!