Looking for some help with JustMLS

[JonW]

Hi Folks,
I’ve never posted here before, but have read much and learned a lot. Thanks.

I’m trying to design my first speaker and have hit my first snag at the measurement stage. So far I’ve made some ugly test cabinets and put in the drivers (Scanspeak 18W/8531 woofer and 6600 tweeter). I bought LspCAD and am trying to get measurements with JustMLS. Things are not looking right, so maybe someone out there can provide me with insight.

I tried to follow the tutorial in the JustMLS manual, linked here:
http://www.ijdata.com/LspCAD_download.html
I’m using an M-Audio Mobile Pre sound card and a Behringer ECM8000 microphone. The setup now looks like this:


I’m using 2 channel mode. According to the tutorial, he starts with the woofer, 8000 sample rate, 32768 MLS length. The mic at about 5 mm from the dustcap. The default is offset=0, length = 1024 ms, so I used those. Measuring gives a plot like this:

Ugly and I’m guessing that does not look correct. I tried changing around the “length” and “offset” values in the “measure” window and things pretty much stayed looking the same.

Next up, according to the tutorial, is the port measurement. I put the mic right at the port opening. Adjusted the levels up. Measured and got this:

That can’t be right, I’d think.

Then it’s move to the tweeter. In the tutorial he has the mic at 50 cm and changes the sample rate to 48000, so I did those. The program puts the length to 170.67 ms, offset=0. I put the offset to 50 cm and then adjusted levels. I also changed the length to 7 ms, according to the tutorial. Here’s what happened:

Changing around the “offset” and “length” settings in the “measure” window didn’t change the overall look of the measured plot.

So I’m doing something wrong and/or leaving something out. Any insights are appreciated. Thanks.

-Jon

[Henckel]

Your measurements dont look right. in order to do fault finding i would start with 44100 samples and 1 channel mode - once this is working -you can move on to 2 channel and other sample rates.
On a pure guess i would expect that microphone and reference channel is mixed and apperently the soundcard dont rightfully support 8000 samples

Finally it might be worthwile to look at the time domain tab

[JonW]

Thanks Henckel. In fact, I did start with single channel mode. However, I never got that looking correct. I do not recall what sample rate I focused on there.

It is quite possible I have mixed up something. I cannot claim to know what I am doing. If I recall correctly, here is what I did:

Microphone goes into the Mic input of channel 1/left.
The RCA cable of the probe (with a 1/4” adapter) goes to input of the channel 2/right.
The RCA cable of the output to the amp (with a 1/4” adapter) is plugged into channel 1/left output. It then goes to my preamp processor (volume control) then to the amp.

For calibration, I unplugged the mic and ran calibration for the right channel. To calibrate the left channel, I took the probe cable out and put it into channel 1/left input. Then recalibrated. I suspect this is all incorrect. But even so, the manual said calibration in 2 channel mode is not crucial. Yet my plots are wildy off.

Some more info: I used 2 channel mode and sometimes, when looking at just the channel 2 signal, it was far from flat- a steep rise from maybe 20 Hz up to 10000 Hz. So of course Ch 1-2 looked awkward. But even just looking at channel 1 still did not look even close to correct. Later (not sure what was different), I got the calibration to look flat and Ch 2 was flat. It was flat for all the plots I showed above. And by “flat” I mean close to flat, but not 100% linear as if there’s not signal at all.

Regarding the time data… I checked it and it looked very close to what was in the tutorial for the tweeter measurements. This pic is taken from the tutorial and is pretty much what mine looked like. The peaks were much smaller and narrower when measuring the woofer:

[5th element]

I use the Wallin Jig for my measurements with JustMLS, the only problem here is that the jig is not wired exactly how MLS wants it, this results in not being able to calibrate anything. Is this a problem? Not really as I know its all flat, but if I DO try and calibrate, the results are often VERY strange and sometimes look the way yours do.

I get measurements like you did on your tweeter, when my microphone isn't plugged in properly, there's a slight bad connection somewhere I cant find, you have to wiggle the cable around a bit, real hi tech stuff.

The results you are getting suggest something is fundamentally wrong with the setup. If you DONT calibrate anything and just go taking measurements JustMLS will assume everything is flat and go on from there. You could try this, if it solves half the headache, by approaching more reasonable results then you can conclude that the calibration phase is, in fact, in error.

On a pure guess i would expect that microphone and reference channel is mixed and apperently the soundcard dont rightfully support 8000 samples

What Henckel says here is interesting and does make reasonable sense. One other thing to consider is the sound card itself, do you have any special effects turned on? No matter how benign. Maybe you dont have any coming through the speakers, but its possible the mic input has something there you don't know about? Or perhaps you have some sort of noise cancellation thing on the microphone input. I know one of my sound cards had this and it would most likely ruin the results, it could be turned off too though.
I use a M Audio Revolution 7,1 and thankfully its free from many 'creative' style signal processing, I have no idea if what you are using is the same.

Oh and about altering the sample frequency, don't bother, just stick with 44.1.

Once you get sorted on actually making measurements, there is a good thread here

Treat twin woofers as one woofer?

outlining how to make useful measurements.

Matt

[JonW]

Hi Matt,

Thanks for the help.

OK, I’ll try jiggling the wires.

And I like your idea of trying to measure without any calibration. I would think that any non-linearity from the sound cad would be minor relative to the measurements. So running uncalibrated should show me the general trend from a driver. Then calibration will just make it a little nicer.

No special sound effects in the sound card. I use it 100% for the microphone and nothing else. Heck, I don’t even know if it has an effects in it. It’s an M Audio MobilePre.

I’ll just stick with the 44.1 sample frequency.

Yes, that’s a good thread. I had already read through it, but I will again, once I can actually take measurements.

[bwaslo]

Jon,

You might want to look at the Windows mixer for your soundcard, and make sure that the Play function is playing only Wav audio and not the line or mic input combined into it. Mixers like to mix things and you want to play out only what the software generates when testing.

You might take a peek at the Record mixer for the card, too, make sure input levels are set correctly and that the correct input is selected.

[JonW]

Thanks. I'll look into all that. As well as the other suggestions. I've been too busy at work these past days to get to try anything different. Maybe tomorrow night I can try to get things working better. I hope...

[JonW]

Hi Guys,

Well, here’s an update: Nothing is working yet. It might be a problem with the sound card. Chasw98 helped me out more over the phone. Thanks, man. If I test channel 1 of the sound card, going out to the in, everything looks fine (pretty flat). But if I test channel 2, it is very far from flat. There is a 45 db rise from 20 Hz up to 20,000 Hz. That can’t be right. I called M-Audio and… nothing yet. They want me to email them some info about my PC (it’s at home right now) and then we’ll see what they say.

Some other things I tried, based upon suggestions:
-Rechecked all connections, jiggled wires, etc
-Tried measurements in both 1 and 2 channel modes
-Ran everything uncalibrated
-Kept the sample rate at 44,100 Hz
-Reinstalled the M-Audio software
-Tried every latency setting (80-2048) multiple times
-Kept the signal levels of JustMLS in the high green, just below yellow
-Kept the knobs of the M-Audio down low, also tried them with the green light on, but not the reds
-Switched cable positions to all possible input and output combinations
-Mic to line in rather than mic in
-Tried measuring a regular speaker (Modula MT)
-Used the good channel 1 to show that neither the preamp-processor or amp were adding in anything funny

So I’m still stuck. I’m hoping M-Audio comes through with a fix. Thanks for all the help so far.