Jump to content
Ricci

Measurements, Close Mic in Room versus outdoor GP, etc

Recommended Posts

Here are a few measurements that I have taken of various drivers in my Othorn cabinets. I knew previously from modeling most of these drivers in the cab that most were not going to match up to produce performance that would be considered good or "useful". However I also wanted to see the actual results from loading various drivers, some of which are very high inductance into a tapped horn and being able to compare that result against what I had assumed they would be in the simulations. The measurements are near-field with the M30 microphone element about 2" from the front plane of the mouth and centered on the mouth of the Othorn cab. The cab was sitting on the concrete floor of my garage with the center of the mouth about 28" off of the floor. I've also done enough near-field bass measurements to know that the near-field response ends up tilted slightly overall compared with a true ground-plane measurements. I've attached a few measurements of the Othorn in a GP setting as well to show these differences.

 

Here is a graph showing the 1m GP response of the Othorn with the B&C 21SW152-4 driver and with the cabinet close mic'd in the top 2 measurements. The bottom 2 measurements are the same but with the driver switched out to the TC Sounds Pro5100. These are the 2 drivers that I had tested in this cabinet both outdoor GP and close mic'd. As you can see there are some artifacts in the close mic measurements and also a few variations in the general shape of the response curve as well when compared with a GP measurement. Note that there is no smoothing on these graphs.

post-5-0-36400900-1428420405_thumb.jpg

 

Keeping in mind the differences between close mic measurements and GP as shown above for the following measurements. These are all taken close mic as outlined above with no filtering on the signal. No smoothing.

 

 

 

B&C 21SW152-4 Close Mic

post-5-0-82174600-1428421172_thumb.jpg

 

 

TC Pro5100 Close Mic

post-5-0-90538200-1428421176_thumb.jpg

 

 

Faital 18HP1060

post-5-0-75377600-1428420403_thumb.jpg

 

 

JBL 2242HPL

post-5-0-75311700-1428420404_thumb.jpg

 

 

BMS 18N862

post-5-0-17837500-1428420404_thumb.jpg

 

 

TC LMS Ultra

post-5-0-06219500-1428420405_thumb.jpg

 

 

B&C 21IPAL

post-5-0-45011900-1428420404_thumb.jpg

 

 

Funk TSAD18

post-5-0-66079600-1428420406_thumb.jpg

 

 

Mach5 UXL18

post-5-0-92891600-1428420406_thumb.jpg

 

 

Sundown ZV3 18D1

post-5-0-18227600-1428420407_thumb.jpg

 

 

FI Caraudio Q18D2 with inductance ring (This one shows just how important it is to accurately simulate the driver and cab you are using. This driver is completely unsuited to this horn cabinet and the result is very peaky and non linear response with high amounts of ringing and distortion over narrow bandwidths. Many of the other drivers are unsuited for this cab as well, I just picked this one because in this case it is the worst match of the lot.  Using just the simple manufacturers parameters this driver actually looks decent in this cab. We can see that this is simply not the case outside of very simplified simulations.)

post-5-0-14407000-1428420406_thumb.jpg

  • Like 6

Share this post


Link to post
Share on other sites

Great to see these measurements. Because I think we all should learn how to simulate high inductance drivers so we can get an accurate simulation. Since the discussion with another forum member and yourself talked bout inductance I am wondering what you use to simulate this? Or do you just measure the driver first and then simulate with that information?

 

Obviously I cant measure ever driver before hand so it would be nice to know how to simulate from the OEM specs high inductance drivers. Then (since I am a mod on HTS) I can help spread the word about how to simulate properly and measure properly. The inductance changes the response by a LOT in simulations and reality and if not done makes for a nasty reality.

 

Thanks for posting these.

Share this post


Link to post
Share on other sites

Out of curiosity, what are the dimensions of the mouth of the horn?  It's interesting that your close mic data tracks the GP data pretty well in the data you provide.  I wonder how things would change if the sub being tested were in a sealed configuration with a large woofer instead?  How about a ported configuration?  Obviously, if the ports are far enough away, things are going to change a lot.

Share this post


Link to post
Share on other sites

Othorn has a final segment of S5 = 2083

I'm not familiar with horn design parameters.  Is that a quantity in square centimeters?  Is the opening ~46 cm square?  If so, that's bigger than I thought.

Share this post


Link to post
Share on other sites

Square CM's. Have you seen what the Othorn looks like?

 

21TH30isohatch.jpg

 

I dont have anything to show you the scale but remember the driver used is the BC 21" driver. So it is a large opening but not sure what size you were thinking.

Share this post


Link to post
Share on other sites

Or do you just measure the driver first and then simulate with that information?

 

Obviously I cant measure ever driver before hand so it would be nice to know how to simulate from the OEM specs high inductance drivers.

 

I measure the driver first...There really is no way around it. The only drivers you can ignore this on are the professional bass drivers because most of them from the large manufacturers put a lot of effort into lowering inductance and they generally model out quite closely.

 

There is no way to use the OEM information unfortunately. Some of them don't even provide an LE spec. You would need a 3 part complex LE spec at the least and virtually no driver manufacturers provide that. The best thing you can get is an impedance measurement with the conditions of the impedance measurement known. Box used etc...If you can fool around in a program and make the impedance curves match closely you'll be very close to having it. Unfortunately the drivers that need this data are the ones that don't publish it. Most won't have an impedance measurement available.

Share this post


Link to post
Share on other sites

Out of curiosity, what are the dimensions of the mouth of the horn?  It's interesting that your close mic data tracks the GP data pretty well in the data you provide.  I wonder how things would change if the sub being tested were in a sealed configuration with a large woofer instead?  How about a ported configuration?  Obviously, if the ports are far enough away, things are going to change a lot.

 

2083 square cm. All output radiates from this opening.

 

I have some measurements of close mic'd sealed systems versus measured at 1 or 2m ground plane (half-space) as well. The basic result is that the low end is tilted upwards a bit and the top end is tilted down a bit in the close mic. Imagine setting a hinge point on a measurement at roughly 60Hz and rotating the measurement on the graph clockwise a few degrees. That's the basic result.

Share this post


Link to post
Share on other sites

Thanks. I have my 15Hst just sitting here and I may have to do an impedance test if I can then. While I will be going sealed in my alignment I am very curious to know what the inductance will add to any other alignments. Can WT3 do this measurement?

 

I think I did an impedance test that way before on a TH cabinet.

Share this post


Link to post
Share on other sites

2083 square cm. All output radiates from this opening.

 

I have some measurements of close mic'd sealed systems versus measured at 1 or 2m ground plane (half-space) as well. The basic result is that the low end is tilted upwards a bit and the top end is tilted down a bit in the close mic. Imagine setting a hinge point on a measurement at roughly 60Hz and rotating the measurement on the graph clockwise a few degrees. That's the basic result.

Interesting.  Thanks for sharing your observations.

Share this post


Link to post
Share on other sites

Really nice work Josh. 

 

If you haven't, you might want to cross post to some of the boards that get a little more traffic, as this information will help save an awful lot of folks from failed projects.  I ended up here because one of your measurements was posted at AVS and I specifically searched on it, otherwise I would not have known about it. 

Share this post


Link to post
Share on other sites

I like the JBL 2242HPL just because I like JBL not that I have same diver, but interesting done outside.

 

How did you do the test any pictures of outdoors.

 

Was this done at daytime as lost of noise in background with traffic and airplanes.

 

Night time is silent still sound outdoors.

 

Did you set gain on mic preamp low and speaker amp level high to minimize background sounds being picked up that could would otherwise add to the chat.  

Share this post


Link to post
Share on other sites

2083 square cm. All output radiates from this opening.

 

I have some measurements of close mic'd sealed systems versus measured at 1 or 2m ground plane (half-space) as well. The basic result is that the low end is tilted upwards a bit and the top end is tilted down a bit in the close mic. Imagine setting a hinge point on a measurement at roughly 60Hz and rotating the measurement on the graph clockwise a few degrees. That's the basic result.

 

I've observed similar results (low emphasis and high-end roll-off) in the most recent (still over two years ago...) close-mic measurements of my tapped horn.

 

I'm not at a point yet where I can really spend any time on speaker and sub R&D, but that day is coming...eventually.

Share this post


Link to post
Share on other sites

Andy your post is a little confusing man. In short yes there are high levels of background noise outdoors from trucks, cars, trains, planes, etc. Also wind can be a problem. However since the measurements taken by me involve the upper end of the speakers performance envelope it is quite loud and able to get well up out of the noise floor. Also a lot of repeat measurements are taken and then averaged which also helps out considerably. Once you do this type of thing long enough it is very easy to tell what is noise contamination and what is not. This is also the reason that the distortion measurements represented are only at quite strong output levels where it gets high enough out of the noise floor to be useful.

 

 

Really nice work Josh. 

 

If you haven't, you might want to cross post to some of the boards that get a little more traffic, as this information will help save an awful lot of folks from failed projects.  I ended up here because one of your measurements was posted at AVS and I specifically searched on it, otherwise I would not have known about it. 

 

Thanks John but I don't have time to post over at the other places much anymore. Super busy lately with some other things that are not audio related so I've been slacking. Hell I haven't even posted here in a while. Feel free to use the info however you want though.

Share this post


Link to post
Share on other sites

Andy your post is a little confusing man. In short yes there are high levels of background noise outdoors from trucks, cars, trains, planes, etc. Also wind can be a problem. However since the measurements taken by me involve the upper end of the speakers performance envelope it is quite loud and able to get well up out of the noise floor. Also a lot of repeat measurements are taken and then averaged which also helps out considerably. Once you do this type of thing long enough it is very easy to tell what is noise contamination and what is not. This is also the reason that the distortion measurements represented are only at quite strong output levels where it gets high enough out of the noise floor to be useful.

 

 

I don't think 60dB is well above the noise floor at 10 Hz. At least not enough to be commenting on the disparity between close mic and GP at that frequency.

 

And, wouldn't "...a lot of repeat measurements are taken and then averaged..." be "smoothing" the measurement?

 

When I compared my close mic measurements of Tumult to Ilkka's GP of same, it was his measurement results that were "tilted", showing a 10dB/octave roll off. I also never saw any difference in smoothing vs no smoothing.

 

HGv3saE.jpg

Of course, I've never measured horns and I'm sure that has a lot to do with the discussion but close mic is far more practical than dragging the stuff outside and smoothing the traces makes for much easier location of the critical data when tweaking sub designs, in my experience.

Share this post


Link to post
Share on other sites

I don't think 60dB is well above the noise floor at 10 Hz. At least not enough to be commenting on the disparity between close mic and GP at that frequency.

 

And, wouldn't "...a lot of repeat measurements are taken and then averaged..." be "smoothing" the measurement?

Smoothing?  Yes, but it is completely different from smoothing across frequency as in "1/3 octave smoothing" and what not.

 

Instead, it is smoothing out the effects of noise.  Assuming the noise can be considered to be essentially uncorrelated in time (independent and identically distributed in a statistical sense), then the central limit theorem applies and the signal-to-noise ratio (SNR) of the averaged measurements tends to increase with the square root of the number of measurements, for a large number of measurements.  Or another way of putting it is that each time you double the number of measurements, you can expect SNR to increase by about 3 dB.  You can also increase the duration of the sweep to get more SNR, but because real-life bass noises tend to be correlated for short periods of time (e.g., the time it takes for a truck to pass), this approach has its limits.  Longer sweeps are very useful to overcoming any noise generated within the measurement equipment itself.

 

This is measurement theory 101 stuff.

Share this post


Link to post
Share on other sites

Smoothing?  Yes, but it is completely different from smoothing across frequency as in "1/3 octave smoothing" and what not.

 

Instead, it is smoothing out the effects of noise.  Assuming the noise can be considered to be essentially uncorrelated in time (independent and identically distributed in a statistical sense), then the central limit theorem applies and the signal-to-noise ratio (SNR) of the averaged measurements tends to increase with the square root of the number of measurements, for a large number of measurements.  Or another way of putting it is that each time you double the number of measurements, you can expect SNR to increase by about 3 dB.  You can also increase the duration of the sweep to get more SNR, but because real-life bass noises tend to be correlated for short periods of time (e.g., the time it takes for a truck to pass), this approach has its limits.  Longer sweeps are very useful to overcoming any noise generated within the measurement equipment itself.

 

This is measurement theory 101 stuff.

 

Yeah, IOW, it's smoothing.

Share this post


Link to post
Share on other sites

Yeah, IOW, it's smoothing.

I don't think you get it.

 

If for some reason the native response of the sub is peaky, i.e., due to high Q resonances, the multiple measurements with averaging will not smooth that part of the response out because it is present to the same extent in all the measurements.  In that sense, it's not smoothing at all.  On the other hand, if it is background noise that is contributing peaks and dips in a single measurement, then those peaks and dips will be diminished in the averaged response.  Likewise, the peaks and dips from noise in the other measurements will also be diminished in the average.

 

So it's probably better to call it "de-noising" rather than "smoothing".  The multiple measurements with averaging provides a more correct picture of the response without the background noise.

Share this post


Link to post
Share on other sites

 

So it's probably better to call it "de-noising" rather than "smoothing".  The multiple measurements with averaging provides a more correct picture of the response without the background noise.

 

"de-noising the background noise" is not measurement theory 101 stuff. :D

 

Think I'll stick to calling it smoothing.

Share this post


Link to post
Share on other sites

Call it whatever but as SME said this IS basic measuring 101. Averaging of multiple measurements has nothing to do with smoothing. It has everything to do with increasing the signal to noise ratio.

 

The tilt in the response shape is not just a horn only phenomena it applies to virtually every bass system I've measured close mic versus outdoors. It really isn't enough to worry about since it is very minor but it is there. I'll see if I can post some data on that later.

 

  • Like 1

Share this post


Link to post
Share on other sites

I don't think you get it.

 

if it is background noise that is contributing peaks and dips in a single measurement, then those peaks and dips will be diminished in the averaged response.  Likewise, the peaks and dips from noise in the other measurements will also be diminished in the average.

 

So it's probably better to call it "de-noising" rather than "smoothing".  The multiple measurements with averaging provides a more correct picture of the response without the background noise.

 

 

Call it whatever but as SME said this IS basic measuring 101. Averaging of multiple measurements has nothing to do with smoothing. It has everything to do with increasing the signal to noise ratio.

 

 

I'm saying that the peaks in a "peaky" response is noise, or "call it whatever". ;)

 

Which of the two traces gives a more "correct picture of the response", smoothed or no smoothing?

 

kuNauzH.png

 

The first and only smoothed trace is the more accurate result to the no smooth/multiple sweep/averaged method, regardless of the number of sweeps done. That's Bosso 101 for finding the Qtc and F3 of the sub being measured. THAT, I get. ;)

Share this post


Link to post
Share on other sites

If the peaks or dips in a response measurement are repeatable and are not due to extraneous noise like is typical outdoors then they are real and part of the response at the mic position. In that case the unsmoothed measurement is the more detailed and high resolution one.

 

There's nothing wrong with a bit of smoothing. It's another tool in the kit. It is a different tool from averaging though.

Share this post


Link to post
Share on other sites

I agree. ^^^

 

Averaging has it's purpose, I just haven't found it to be useful in determining the native response of a subwoofer.

 

Ilkka's measurements were all 1/24 octave smoothed as it was his maximum resolution. Below 100 Hz, I believe 1/24 octave smoothing is extremely accurate.

Share this post


Link to post
Share on other sites

No arguments there. I usually use 1/12th when I do need to clean something up in the bass range.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×