I think one of the most important issues with beginners who get a decent tool like LspCAD is understanding the MLS length vs gate length. I was very fortunate that fellow DIYer Ed West spent the day after DIY2001 to explain it to me. The new JustMLS tutorial is a lot better about how to use gates but it does not explain why they selected the gates they did. I also noticed some of my MLS on my site are not nearly as clean as the ones I get now. So lets talk JustMLS.
There are two lengths to worry about. The first is the
"MLS Length" that is chosen on the set up screen and determines the
length of the psht sound: 32768, 16384, 8192 and 4096.
But the mic is not recording for the entire psht time. The
time the mic is recording is set on "Measure!" screen in the gate
"Length" field:
So if you set the MLS length for a long 32768 psht and the gate length to 10ms, you are not using most of the psht. So set the MLS down to 4096 or 8192. Remember you must calibrate JustMLS for each time separately.
The longer the mic is recording, the lower the frequency it will record. But the longer the mic is recording, the more reflections it will pick up.
The farther away from the driver the mic is, the more reflections you will pick up. The closer the mic is to the driver, the less accurate reading on what the whole speaker is doing.
I have been learning with my current project that the reflections will do a lot more harm to your MLS than not recording the low end, which is mostly below the XO work anyway. That is why the JustMLS tutorial suggests 7ms and 10ms. Also, set the chart window to 200-20000, the blue numbers at the bottom left of the MLS.
The charts below show the relationship between the gate
(length), the number of wave cycles and the length of the recorded sound in
wave distance. The first chart shows a 10ms gate which records 11.3 feet of
sound. That allows for only 1.6 cycles of a 160Hz wave. Less than one wave of
80Hz is available. Not much, but plenty for XO work in the 600Hz and up ranges.
This setting will keep the reflections down, but not make them go away. If you
are recording 11.3 feet of waves with a ceiling and floor 4 feet from the
driver, you have picked up the first two hard surface reflections.
This "Joe Speaker" MLS shows the unsmoothed reading at 10ms. I know
my room is not that great for reflections. The severe drop off at 200Hz has
nothing to do with the speaker, but is a lack of complete (or enough) cycles
for the computer to record.
Now a 50ms gate increases the recorded wave to 56.5 feet.
Just one 20Hz cycle, butta you betcha you gotta getta somma reflections. So you
still don't get to clearly record the bottom octaves but added a bunch of
reflections.
Now we move to the 175ms mark. You can get that 20Hz reading
albeit only 3 complete cycles. But look at the room refections. Wow!
This is the same "Joe Speaker" at 175ms. Not just larger swings, but
a very different reading. Looking like this, I would not use any of the lower
section anyway. You could potentially smooth out the humps using the octave
smoothing, but you will likely smooth something important along the way.
So you will need a gate length of 50ms to get one cycle of 20Hz. One cycle will not give a very true reading. So you will have to go up from there. But at one 20Hz cycle, you are at 56 feet of recording and enough for all the frequencies to reflect across the room at least 4 times in six directions. If you want, say, four 20Hz cycles, you have all the frequencies bouncing about at least 28 times. Ouch!
So set the gate length to 7ms or 10 ms. Measure out of doors. I know I may have gotten something wrong so feel free to politely correct. I hope this helps.
Copyright Peter Jay Smith 2005 Return to helarc.com