Garage Loudspeaker Lab

Mark DeArman

Mark DeArman

Junior
Aug 15, 2011
389
0
16
Puyallup, Washington, United States
www.cascadeacoustic.com
A few years ago I started the project of turning the long abandoned 1970s McCauley Sound factory back into a functional loudspeaker workshop. I think I posted some pictures of the place at the time. The roof leaked everywhere. It was stacked floor to ceiling with enclosure prototypes, lawn furniture, broken lawn mowers, old aluminum casting equipment, ... what else, terrible Russian OEM loudspeakers.. Well it was a lot of crap.

I can finally say today that it is really functional. All the junk is gone. The roof has been replaced. All the cables have been sorted out. All of those little plastic drawers filled with miscellaneous screws have been sorted.

I thought I would post some final follow up pictures. I posted some, light at the end of the tunnel pictures about six months ago I think. But now all the shelves are up, all the turn of the century computers are gone. Yeah! It's still dirty, but it works.

Oh yea, and we've rented the old forge where we used to sandcast all of our own aluminum baskets to a local cabinet company so we have a full woodshop and paint booth of the premises again too.

Anyway, it is neat to have a space to work on projects outside of the normal production workflow.
 

Attachments

  • lab1.jpg
    lab1.jpg
    1.6 MB · Views: 1
  • lab2.jpg
    lab2.jpg
    1.6 MB · Views: 1
  • offce3.jpg
    offce3.jpg
    1.2 MB · Views: 1
  • office1.jpg
    office1.jpg
    1.3 MB · Views: 1
  • office2.jpg
    office2.jpg
    1.2 MB · Views: 1
  • workshop1.jpg
    workshop1.jpg
    1.6 MB · Views: 1
  • workshop2.jpg
    workshop2.jpg
    1.5 MB · Views: 1
Last edited:
Re: Garage Loudspeaker Lab

Lee Douglas said:
Love the "Hey, Idiots..." note on the wall.
Click to expand...
And don't try to measure impedance in a noisy room-it doesn't work (correctly) either.

Things are not always as easy as people would like them to be.

At least to do CORRECTLY. Sure, you will get a "number", but is it correct?????????
 
Re: Garage Loudspeaker Lab

Ivan Beaver said:
And don't try to measure impedance in a noisy room-it doesn't work (correctly) either.

Things are not always as easy as people would like them to be.

At least to do CORRECTLY. Sure, you will get a "number", but is it correct?????????
Click to expand...

Perfection is the enemy of good enough. As long as you can identify your error, the data is at least usable to a known extent.
 
Re: Garage Loudspeaker Lab

Well the system I made has a lot of integration time which is good against random noise, but as you push the box around the room, and it is sitting inside of or out of a room mode, it could throw off the measurement of Fs by a freq Hz, when everything is lined up perfect. It also throws off the estimate of how damped the box tuning is significantly, which is really irritating when setting the porting.\

Plus, like Ivan said, correlated noise will just cause all sorts of crap which you don't want. Lucky the whole room has open insulation covered with cotton on two sides. Works pretty well to keep the garage noise down.
 
Re: Garage Loudspeaker Lab

Mark DeArman said:
Well the system I made has a lot of integration time which is good against random noise, but as you push the box around the room, and it is sitting inside of or out of a room mode, it could throw off the measurement of Fs by a freq Hz, when everything is lined up perfect. It also throws off the estimate of how damped the box tuning is significantly, which is really irritating when setting the porting.\

Plus, like Ivan said, correlated noise will just cause all sorts of crap which you don't want. Lucky the whole room has open insulation covered with cotton on two sides. Works pretty well to keep the garage noise down.
Click to expand...

It doesn't need to be "dead quiet", but reasonably free from louder noises for impedance measurements.

I learned that many years ago when working on a system and doing impedance measurements in a church while the organist was playing.

My measurements kept going all over the place depending where it was in the sweep and what notes she was playing.
 
Re: Garage Loudspeaker Lab

Ivan Beaver said:
I learned that many years ago when working on a system and doing impedance measurements in a church while the organist was playing.
My measurements kept going all over the place depending where it was in the sweep and what notes she was playing.
Click to expand...

Haha. I bet, especially if you're using a fast sweep. The ones I always run into which ruin things are air nozzles, die grinders and nail guns. Yikes.

As a aside on swept measurements, I'll try and phrase it correctly. If you have a sweep and random noise, the long sweep will pick up more noise. If you have low random noise and lots of correlated noise (like impulses and tones), the short sweep will smear them out into the measurement ruining more of the measured frequency range.
 
Last edited:
Re: Garage Loudspeaker Lab

Mark DeArman said:
Haha. I bet, especially if you're using a fast sweep. The ones I always run into which ruin things are air nozzles, die grinders and nail guns. Yikes.

As a aside on swept measurements, I'll try and phrase it correctly. If you have a sweep and random noise, the long sweep will pick up more noise. If you have low random noise and lots of correlated noise (like impulses and tones), the short sweep will smear them out into the measurement ruining more of the measured frequency range.
Click to expand...
But a windowed sweep will allow a narrower window with a longer sweep-meaning that the noise has to be very "precise" in timing to mess with the measurement.

I know of cases that people have done long sweeps (like 2 minutes or more) DURING a performance-with the sweep real low in level (as compared to the regular music signal) and still get pretty good data.

It all depends on the tools available.
 
Re: Garage Loudspeaker Lab

Ivan Beaver said:
But a windowed sweep will allow a narrower window with a longer sweep-meaning that the noise has to be very "precise" in timing to mess with the measurement.

I know of cases that people have done long sweeps (like 2 minutes or more) DURING a performance-with the sweep real low in level (as compared to the regular music signal) and still get pretty good data.

It all depends on the tools available.
Click to expand...

Well it is all a balancing act depending on the interference. Music tends to be highly uncorrelated with a gliding tone and the rejection is normally very good. Certain parts like snare hits will correlate well with the sweep, but the long window length implied by the long sweep will tend to localize the snare hit very tightly in frequency. Anyway, we could talk about this for weeks if you want. I always love to hear more empirical/anecdotal evidence of how well gliding tones work compared to MLS style measurements.
 
Re: Garage Loudspeaker Lab

Ivan Beaver said:
But a windowed sweep will allow a narrower window with a longer sweep-meaning that the noise has to be very "precise" in timing to mess with the measurement.
Click to expand...

Regardless of the post-processing window you apply, you are still effectively convolving the recorded waveform with a window the size of the sweep first, to recover the impulse response. Depending on the tapering, instantaneous frequency (ie. linear, exp/log chirp), and other things it will approximately be sinc function whose main-lobe width is related to the sweep length. Ie. Longer sweep, narrower main-lobe and better frequency localization, shorter sweep, wider main-lobe and more smearing of any correlated noise.
 
Last edited:
Re: Garage Loudspeaker Lab

To continue, all I was trying to say in the first place, is that if you have a colored noise environment, (A/C units, computer fans, and etc.), it seems to me that the longer sweep length seems to pickup and redistribute more of that broadband noise into the measurement.

This is an anecdotal statement, I have not actually setup and tested this, but it seems to be true from what I've seen. If what we are measuring is LTI you could average multiple FFT's to remove the noise, like is done in traditional spectral analysis.

I have tested a exponential sweep mixed with music many times. Music, which we say can be approximated by pink noise in a global sense, in the interval of the sweep, is never going to have a spectrum close to colored noise.

P.S. I should also say, that I don't do that many live-sound measurements, most everything I work on is under ideal conditions in the lab, so you guys might know more about it than I do. Tools like Sys-Tune which develop a probabilistic model of the system based on any stimulus and enough time, I would think would work very well in a live environment.
 
Last edited:
Re: Garage Loudspeaker Lab

Mark DeArman said:
To continue, all I was trying to say in the first place, is that if you have a colored noise environment, (A/C units, computer fans, and etc.), it seems to me that the longer sweep length seems to pickup and redistribute more of that broadband noise into the measurement.

This is an anecdotal statement, I have not actually setup and tested this, but it seems to be true from what I've seen. If what we are measuring is LTI you could average multiple FFT's to remove the noise, like is done in traditional spectral analysis.

I have tested a exponential sweep mixed with music many times. Music, which we say can be approximated by pink noise in a global sense, in the interval of the sweep, is never going to have a spectrum close to colored noise.

P.S. I should also say, that I don't do that many live-sound measurements, most everything I work on is under ideal conditions in the lab, so you guys might know more about it than I do. Tools like Sys-Tune which develop a probabilistic model of the system based on any stimulus and enough time, I would think would work very well in a live environment.
Click to expand...
I was talking about TEF, not dual FFTs.

TEF has the greatest noise immunity by far of any measurement system (but I am not aware of all of them).

Several years ago I was in a Church doing some measurements.

We had just made a measurement and saved it. Just after that there was an industrial type shop vac (think LARGE) sucking out some water out of a floor pocket, within a few feet of the measurement mic and MUCH louder than the swept sine we were using.

So we decided to measure again with the noise on-just to see what the difference was. The two traces laid on top of each other. So the vacuum was not affecting the measurement at all.

That is the wonderful thing about the TEF-you don't have to have a quiet environment to measure in.

But that is ALSO the bad thing about the TEF. Because of its ability to "window out" other influences on the sound from the speaker-it does not always give you a good "correlation" to what you are hearing.

I do not use TEF for sound system alignments for that very reason. I prefer dual FFT-because it is more like what I am hearing and MUCH faster

But for loudspeaker measurements of performance, it is great and does things other systems can't.

It depends on "What am I here to do". and then choose the proper tool.

But as amazing as it is-I feel it will die within my lifetime. Truly sad :(
 
Re: Garage Loudspeaker Lab

Ivan Beaver said:
I was talking about TEF, not dual FFTs.
Click to expand...

Yeah TEF functions analogous to Farina's Matched Filtering and Heyser's corrected TDS. It is not the same way that SMAART and Sys-tune function, although the results they produce should be similar, for most systems.

And yes I do agree it (the TEF/Corrected Heyser method) offers many advantages. In fact my own measurement system is based on a similar principle. I also agree with it's disadvantages for live insitu setups. But, anyway, everything I said applies to all systems and is just a general result of the Fourier transform time-bandwidth product.

Okay, here I go again editing. I believe the TEF actually uses Heyser's down convert method so, it's not exactly like the matched filter I described in the post above, although they have been proven to be mathematically equivalent. It has a LP filter which is applied to the baseband signal after it is downconverted/dechirped. You never see this window in action but it's parameters are chosen based on your sweep length/rate. This is the window I speak of, not the one which you apply after the measurement is done. Anyway.

P.S. If you want to talk in terms of spectrum analyzers, you would call this filter/window the resolution bandwidth filter.
 
Last edited:
Re: Garage Loudspeaker Lab

And of course the idea behind all of the real measurement systems is to try to give the user an accurate representation of what the loudspeaker/system is actually doing.

Of course there are lots of "little things" that will affect what the computer screen shows (and doesn't show). It all comes down to the user understanding the various parameters that are entered into the settings.

You can get all kinds of "different results" with different settings. And there are all kinds of different ways to "average" a response.

Sometimes this is very good (makes it easier to see) and very bad (hiding peaks and dips).

And agreed there are advantages and disadvantages to each system.
 
Top Bottom