Testing Drivers for Design

I use a specific procedure for measuring drivers that may differ somewhat from those used by others. It works for me. This single setup method does a good job of handling diffraction and baffle step issues for the individual speaker design. It also means that measurements from one design are not easily used with an alternative box or a different combination of drivers.

I design the boxes and install all the drivers, ports and stuffing. Lately I've been using a test baffle that includes rabetted driver cutouts. Since I primarily use biscuits, it's easy to duct tape the test baffle to the box. I wire each driver separately.

I do my testing outdoors. For "bookshelf" speakers, I mount them on a stand about 4.5 feet high. The mic is set up roughly 1 meter, (40") away. Usually, I place the mic on the same horizontal plane as the tweeter with the speaker rotated a few degrees off axis. Towers are measured on the ground. I bought a cheap laser level to help with the alignment process.

For calibrating the sound card, I take the speaker output directly to the line in, set the levels, and calibrate at various sample rates and lengths. When you get a good calibration file ("Justmls.cal") make a backup. I've trashed mine a couple of times.

I've found setting the sampling to 44,100 with 8192 as the MLS sample length seems to be the best compromise. Typical window lengths of 7 to 30 ms seem to give me the best results. I use the tweeter to establish the reference volume level for all tests. With 2 channel, you need to set the levels first.

Enter the distance between the mike and the baffle in the "Offset" box, then press "Test". Next check the Time Domain window, make sure the spikes lines-up at left. (Ingemar covers this in the users guide). You may need to tweak the offset distance a bit. If you use this single point setup, you may need to adjust the mike distance for each driver. Make sure choose the (Ch 1-2) value for your measures. Export your measurements frequently. If you change scaling or smoothing, these appear to be reflected in the measurement files.

With the initial calibration done, I now measure the tweeter. Using the same mic setup and volume levels, I then measure each of the other drivers. Typically I push the test button several times until I'm convinced that the measurements aren't significantly changing. (Using these Lengths the sub 200 Hz area is not accurately measured) I usually save a set unsmoothed, and a set with 1/6 or 1/12 smoothing for design. I also use Snagit from Techsmith to capture some screen shots for reference.

Once each driver is measured individually, I take a final measurement with all drivers in parallel. This curve will used as an initial "target" value in LspCad. By tweaking the XYZ locations of individual drivers to match the the "target" curve, I can make sure the acoustical centers of the drivers are properly modelled. If you use the same sound levels for everything, LspCad will handle approximate tweeter padding.

After, measuring (and saving) all of FR data. I then measure the impedance of each driver IN THE BOX (using an 8 ohm mills resistor) This captures the effects of port/box loading.

For finished speakers, I typically do a ground plane measurement on the concrete of the driveway with the mic at about two meters.

My Gear and Setup

Most of my initial frustrations with JustMLS were traced to a funky jack on the line level input of my sound card. I drilled out a "blank" on my computer case (where all the cards are plugged in) and mounted two rca jacks in it. I soldered a CD-to-soundcard cable to the jacks, and plugged it in to the auxillary pins on my Soundblaster 16 sound card. Now I use rca patch cords for all of connections. I've found this more reliable than soundcard jack.

I started out using a Wallin jig, Panasonic DIY mikes, and Eric Wallin's battery powered mic preamp. They worked OK but took a lot of futzing to get everything right. When I accidently touched speaker leads, I fried the jig which in turn fried my shop receiver. So I decided to step up to a low hassle solution. Wallin Jig 2 And Wallin Mike Preamp I think I had about $60 into this setup, but it seems like it always took me an hour to get things working right.

I now use the Behringer ECM-8000 measurement microphone (about $50). I find it clean and durable. Throw in a mike stand, 20 foot cable for about $30.

The mike needs a phantom power source. I looked around for cheap amps but the best solution seemed to be the UB-802 mixing console. ($60) This allows a bit more flexiblity and an adjustable gain control. I've also used the console for mastering audio tracks from vinyl and narrating videos.

Update July 2009,

The old Win 98 box gave up the magic smoke so I'm now using my old Thinkpad with an Edirol UA-1EX USB sound card.

I'm leaving the old box on the page as some of you might find it useful.

I also built two sets of probes with the resistors built into the ends. With alligator clips on one end and an RCA jack on the other. I used 20 ft lengths of two conductor, shielded, 20 ga cable I found at a surplus store. I soldered an alligator clip to a mills 8 ohm resistor for doing impedance. You'll also need one patch cord with an RCA to 1/4" mic plug to get from the mixer to the computer. I use a 25 watt plate amp from Apex Jr for powering the speakers. I can setup in about 10 minutes.

You won't need much of a computer. My shop box is an old AMD 300, 64 Mb of RAM running Win 98 SE.