Measurement approach, 2-way closed box

[Elbert]

The purpose behind this thread is to post as I go along making measurements on a project, so as to:

A: Present approach and method in the hope that it will be useful for others in a similar situation

B: Allow for input and correction through a systematic process from those more knowledgeable to ensure the process is optimal, both for my own and fellow DIY'ers benefit!

The Project:

Closed box with SEAS W17CY-001 6,5" mid-woofer and SEAS 27TBCD/GB-DXT waveguide Tweeter.

Equipment:

ASUS Eee with HolmImpulse and a DIY Panasonic WM-61a Linkwitz mod. Mic.

The loudspeaker will eventually be active, and the purpose of the measurements are to:

A:
Acheive optimum driver integration in X-over design

B:
Optimise actual room response through response shaping circuits, e.g. baffle step, etc..

C: identify physical design-flaws like box resonances, sub-optimal box damping, diffraction, etc..

Here we go!

[Elbert]

The first step will be to obtain a starting reference point.

To this end, I will use the SEAS data-sheet frequency plots, and a box simulation made with WinISD.

The Box-volume is 12L, giving a calculated system Q of 0,85 and a Fsc of 75Hz.

The simulation does not take in to account any box-stuffing, so the actual result must be subject to verification through measurement.

The Q of 0,85 gives a slight LF rise, but barely above 0,5 dB, which should be acceptable. This might also be reduced by box stuffing, which could increase the "acoustic volume".

[Elbert]

The frequency plot on the factory data-sheets are quite detailed, as seems to be customary for SEAS.

The interesting thing will be to see how the drivers behave once they are assembled in to the enclosure. That is what I intend to use as a baseline for any further work.

[Elbert]

The first challenge is therefore to achieve an "anechoic" measurement that will give the response of the drivers plus influence of the boxes.

The gating functionality of the measurement software should do part of the job covering low mid-range and upwards..
Hopefully, the low frequencies can be handled by nearfield measurement.

So, the first challenge will be to see if a nearfield measurement can give a usable LF response, and if this can be combined with a gated farfield measurement to "replicate" an anechoic measurement.

IN the next post I'll post some nearfield measurements and see if that can be of use.

[gedlee]

I hope that you will be doing polars of the drivers as well as the completed system. Otherwise just looking at axial data is not going to tell us very much.

Holm is ideal at this task, and I can help with data presentation if you want, but I need the polar data to do that.

[Elbert]

Good point!

Well, by polar I assume you mean off-axis plots?

I have obtained a crude but functional turn-table to that end ( The sort you use for Porcelain painting, from my late Grandmother).

But I Thought it was a good idea to sort out the low frequency nearfield measurements first, and then expand with gated farfield measurements both on- and Off-axis.

For the sake of making the thread as "educational" as possible, I'll try to do things one step at a time, getting it right before moving on!

Very happy to have your comments Dr. Geddes, considering the knowledgeable contribution you have made on numerous other topics, I'm sure your advice can be of great help to guide this thread in a direction that may hopefully make it in to a useful tutorial for those of us doing this for the first time!

[gedlee]

I'd worry about the LF last. Once you get everything else right the LFs will be easy.

[Elbert]

Dr. Geddes,

Despite your advise, I did the low frequency near-field measurements first.

1:
The set up was easy, I just put the mic as close as I could to the woofer cone.

2:
I was curious if it would actually work, it seamed to good (easy) to be true.

I placed the mic in 3 different positions:

1: As close as possible to centre (next to the phase-plug) (blue)
2: Midway between the phase-plug and the cone-edge (red)
3: Next to the cone edge (green)

The first plot shows the response from the three different positions with the gate function active.

Now, this doesn't quite resemble the original woofer data-sheet plot or the calculated box plot..

At this point I thought that nearfield measurement would not work after all, for no good reason I could identify..

[Elbert]

Then it struck me that switching the gating off could improve matters. After all, the purpose of near field measurements is to eliminate room and radiation effects through measuring the immediate pressure variations at the extreme proximity thus eliminating the need for gating in the first place.

The next plot represents the same near field measurements without gating.

Obviously, the response goes a bit haywire above 700 Hz, but that was expected, and according to theory that states that frequencies with wavelength equaling the cone radius (or above) can not be measured nearfield.

The variation in the different measurement positions doesn't seem to be that huge either..

Now the low frequency response seems to correlatew rather well with the calculated response plot!

[Elbert]

But which of the three measurements positions are best in terms of true representation?

The white paper by D. B. Keele Jr. explains that the ideal is a on axis position, so I'll use that from now on.

The next plot shows the gated and raw measurement superimposed.