Measurement Lab
I first purchased the English version of LspLAB 2.0 in June of 2000 from Johny Grenander in Sweden. This full-featured measurement package is an outstanding value and has worked quite well for me.
A few pics that may be of interest to you...
My acoustical measurement studio. Also know as the living room.
LspLAB produces the pulse; SoundBlaster Live! soundcard sends it to an old
Kenwood 35W integrated amp; then though about 20 ft of 12 ga speaker cable over
to the test baffle; out the driver over to the home built mic with Panasonic
capsule; to the LspLAB measurement box (which supplies phantom power to the mic)
then back into the soundcard . This setup allows me to place a
crossover on a prototype board in front of the monitor and run a pair of cables
to the speaker (hi and low). I click the mouse, take a measurement, swap a
component, repeat. In a short time I can get something that
measures pretty good. Then then the trick is to make it sound
good.
The 3/4" thick MDF baffle measures 7' x 4'. A 1' x 1' insert is cut from 1/4" MDF for each driver. All tweeters are flush mounted on this insert and the insert is then flush mounted on the baffle. All woofers are surface mounted on the insert then flush mounted on the baffle. The driver center is located 27 inches from the top and 23 inches from the right side from the microphone's point of view.
I only added the above pic in to give you a better view of the test baffle and to prove to the world that I make the world's worst digitized pictures. This was taken with a disposable film camera then scanned on a vintage HP scanner. I can typically get a cleaner impulse if I place a few blankets on the floor to absorb most of the floor reflection. It also helps to place some foam around the tripod head for the same reason.
A quick tour of my LspLAB settings
Below you will find numerous screen
captures that I've taken from LspLAB. My intention is to share with others
some of my settings, insights and notes for LspLAB. Likewise, if you see
something amiss in any of the pics below, please bring it to my
attention. It's all about sharing knowledge.
Here's what my soundcard compensation looks like. SoundBlaster Live! OEM.
Although LspLAB will work with many sound cards, Sound Blaster Live! was
the card that Johny used for the development of LspLAB. Works great for
me. Good price too. I don't bother
with probe compensation. It often gives me weird results. I don't
think it's absolutely necessary. Your mileage may vary.
The Hanning is a good all-around window type. Manual window start works
best for me - it's quite automatic actually. Speed of sound in
Albuquerque, New Mexico, USA is about 340 m/s. I never bother with trying
to get accurate dB readings on my SPL graphs. I arbitrarily add 60 dB to
get the sensitivity in the ball park.
If you follow the instructions above, your time marker will stop just before the
first reflection. Nice.
Although slow, I think that the Gated Time Window Sweep (left) gives the most
accurate frequency response measurements. I use the Sweep (right) with no
time window for in room response measurements.
MLS Length is a trade off between speed and accuracy. 65535 seems to work
good for me on my machine (800Mhz Pentium).
The Spike method is nice for measuring tweeters. I don't use it much.
Sometimes my waterfall plots look about right. Sometimes they look
weird. I wish I could specify the length of the time window. Above
it's set to 2.5484ms. I could set it to 1 ms or 5 ms and the chart would
still plot at 2.5484 ms.
I prefer the sweep method to the MLS method for impedance measurements. 128 freq/oct is a bit
excessive but that's the kind of guy I am (64 works just fine). Here I've just ran a
calibration sweep with a 10 ohm resistor. My calibration factor/status is
.9922 (very good). If you've got a calibration factor of .8 something
probably isn't right with your setup. If you've got a calibration factor of
.3 you've got a problem. Likely a blown IC in your box (most likely the LM324) or possibly
a dead battery.
Here I'm setting the levels the for impedance. I prefer to have the level
set a little below 0.
Here's an impulse of a tweeter. I followed the instructions for setting
my time window and as you can see the time marker stops just before my fist big
room reflection. I should note that extending the time marker out to 10ms
would not change the freq response plot very much. Going out to 12.5 would
start to introduce some nasty comb filtering.
Here's an impulse of an 8" woofer.
Here is an impedance sweep from 10-20,000 Hz of a 10 Ohm resistor using my
unmodified measurement box. If you are a LspLAB user you might appreciate
this. In the past my impedance curves, while perfectly useful, had a fair
amount of hash above 5k as seen above.
Here is an impedance sweep from 10-20,000 Hz of a 10 Ohm resistor with my modded
box. I have Sheldon Stokes to thank for helping me with the
mod. As a result, the noise is now about an order of magnitude less than what is was
before. The secret? I replaced the stock LM324 with a LM324
purchased from... get this... Radio Shack. I used part number 276-1711
made by STI (or ST) in Korea (or Malaysia). Apparently not all LM324's are
created equal. Don't attempt this mod unless you are very handy with a
soldering iron. Don't come crying to me or Johny Grenander is you fry your
box. If you do go ahead with the mod, you really should install a 14 pin
socket for the LM324. The socket sure makes things easier if you need to
replace the IC further down the road.
I'm no LspLAB expert and I'm certainly no loudspeaker measurement guru. However, if you have any questions, or topics about LspLAB that you would like to see on this page, please let me know.