Minimum distance for farfield measurement

[Pablo]

Since the path free of reflections seems to be longer if you choose a mic distance of less than 1 m, I´d like to ask how close could I place the mic, staying in a farfield enviroment, in order to get a longer window.

The speaker to be measured is a very small two-way with a 1" tweeter and a 4.5" woofer.

I figure it may be related to the diameter of the bigger driver or the baffle width, but don´t know exactly how to estimate it.

Any tip?

Thanks all.

Pablo.

[fjhuerta]

I believe that 1 meter is the shortest distance so that two drivers, in a two way, will sum up correctly.

But I could be wrong here. I do know, though, that distances bigger than 1 meter are recommended for three ways. I've used them, and they work fine.

[Pablo]

Thanks for the answer, Javier.

I believe that 1 meter is the shortest distance so that two drivers, in a two way, will sum up correctly.

That´s exactly the point I was asking for.

I´d like to improve the resolution around the point where far and nearfield measurements are spliced, but I don´t know if it´s better to improve the farfield lower limit (increasing the window by placing the mic closer) or by splicing both higher (trusting then the nearfield higher freq).

Sorry if I´m not expressing clearly

Pablo.

[Dave Bullet]

1 metre for a 2 way is a good rule of thumb.

D'appolito or one of those guys recommended a mic distance of at least 3 (or 5) times the centre to centre driver distance.

Assuming you have a 1" tweeter with a 4" flange and 4.5" driver, with say a 5.5" overall diamater, then your c:c would be approx 3.25".

Going with the 5 times as the conservative, then a mic distance of 16.25" should be enough for a small 2 way.

For the 3 way I am working on, I calculate at least 1.5 metres is needed based on the above 5 times c:c distance (between tweeter and woofer with a midrange in the way).

A 10 degree phase difference at 1000Hz is far more significant with that shorter wavelength than a 10 degree difference at 100Hz.

The other factor is of course baffle effects. Even though the distance above will adequately sum the drivers, baffle diffraction may still be influencing the signature at that distance.

What I recommend you do is start with 1 metre and keep decreasing the mic distance until the signature obviously changes. This will allow you to increase your gating to get the lowest possible frequency measurement before reflections affect the response.

Cheers,
David.

[Pablo]

Thanks David, I´ll try that.

The c-c distance 5 times seems reasonable. I´ve searching around the web and found Struck & Temme´s "Simulated free field measurements" 1992 article:

http://www.listeninc.com/files/pdf/AESFreeField.pdf

They suggest:

For the response to be measured in the Far Field of the source, d should be at least 3 times the largest linear dimension of the source, M

d > 3M

From the graph, M seems to be the longer internal distance (¿diagonal?) of the enclosure. No mention to sumation of several sources, so who really knows.

By the way, I got puzzled after taking a look to the comparison between their free-field and truly anechoic measurements.

They don´t seem to apply correction for the diffraction in the nearfield (something that most of us do) before splicing to the farfield and surprisingly both results show a close correlation (+/- 1dB higher in the low end, less than diffraction simulators would compensate for).



Maybe we are overestimating the diffraction correction in our nearfield measurements?

Pablo.

[Dennis H]

They don´t seem to apply correction for the diffraction in the nearfield (something that most of us do) before splicing to the farfield and surprisingly both results show a close correlation (+/- 1dB higher in the low end, less than diffraction simulators would compensate for).

If you are splicing at a frequency below the baffle step (driver acting as a true omni source), it shouldn't make any difference whether you apply BSC to the nearfield or not. It's useful to apply BSC if you want to splice a bit higher where the driver is still beaming. That article you linked discusses it on page 13 but BSC sims are outside the scope of the paper so they just say it isn't a big problem as long as you splice low enough.

[Pablo]

You are right Dennis. Rereading the paper, they choose c/pi*M, ie 10950/M, as the upper frequency limit for nearfield. So this should work fine for small boxes.

Pablo.