Erin's Audio Corner (my new review YouTube channel/website)

[ErinH]

Here's some information for those of you who may want to conduct their own measurements with the ground plane method.

I've mentioned it numerous times but in case someone didn't catch it or doesn't already know: when you measure a speaker in the ground plane you need to tilt the speaker some angle in order to line up the speaker with the microphone axis. This is covered in various places online and discussed in D'Appolito's "Testing Loudspeakers" book. It's a pretty simple tangent equation but there are calculators online to make this easy for you. Here's one: HiFi Loudspeaker Design

You enter the mic distance from the speaker and the distance from the ground to the tweeter (or other reference axis; say, midpoint between tweeter and midrange). This site recommended using your phone in "selfie" mode to make sure the microphone shows up in the center of the image. I do this from time to time to sanity check the calculations. You can see my 4th image below displaying this method.

Now, what I've read before is it is OK if you are off a little bit. But I thought it would be helpful for me to show just how much the tilt angle matters so in this example I am using the same Buchardt S400 in my garage.

First I used a laser level to draw a line on my garage floor to make sure the microphone was on-axis with the speaker. I also used a distance finder to make sure the distance was at 2 meters.


**
Side note, if you are interested, these are the two items I use and am quite happy with them. About half the price of the hardware store versions. If you plan on doing GP measurements in your garage, these will make your life so much easier and improve accuracy of your aiming/setup. I recommend you buy them. If not these, something along these lines. The links below are made with my Amazon affiliate link so if you do purchase them I'll get 2%... hey, every little bit helps.
Laser Level
Laser Distance Finder (bonus: this thing has an angle detector which is perfect for determining speaker tilt angle)
**



Anyway, here's some photos:

Selfie Photo to make sure the microphone is at the center of the image; meaning it is lined up with the tweeter axis.

Now, keep in mind the Buchardt S400 has a 2 degree tilt on both the baffle and the rear. Literally, it looks like a parallelogram instead of a typical box enclosure.
1) The first test was ran with the speaker bottom flat on the floor; not tilted forward. (Red)
2) I then used the site above, entering the distance as 2 meters and the tweeter height as 4.5 inches (~ 10cm). I got a suggested tilt angle of 2.90°. I thought, hey, let's overcompensate here and also add the 2° tilt from the enclosure itself, so I came to 4.90° total. Therefore, the actual angle of the baffle perpendicular to the floor was now 2.90° (4.90°-2.00°). (Blue)
3) I then remeasured at 2.90° (no factory tilt accounted for) and the actual angle of baffle perpendicular to floor was 0.90° (2.90°-2.00°). (Green)




Here is the result.

You can see there is indeed a notable difference in the 2-4kHz region between the 3 measurements (and a smaller difference from 5-8kHz). I think this can be telling of the best axis to listen on but that's for a different time. For now, the point is: the tilt angle even within 5 degrees has approximately 3.0dB difference at about 3kHz. And the 0.9° vs 2.9° difference at 3kHz is approximately 1.5dB.

What does this mean? I think it means that care should be taken to ensure the intended axis of measure (the tweeter, between tweeter/mid, etc) is used as the reference plane and the speaker tilted as necessary to make this so.


Obviously, YMMV (your mileage may vary) depending on the speaker, distance, etc. But I would urge you to take care to make sure the tilt angle is correct before you continue with the measurement process.


- Erin

 

[ryanosaur]

Just wanted to let you know somebody is following along.


Cheers!

 

[shadyJ]

Just wanted to let you know somebody is following along.


Cheers!

Yes, absolutely. I wish I had figured out how to do full-range groundplane measurements. That would have saved me a lot of time of figuring out my testing rig construction, not to mention the expense. Seems like it would be trickier with anything large than bookshelf speakers, but I suppose you could do it with enough distance from microphone.

 

[Danzilla31]

Thank you for doing all this I've learned a lot as I've been lurking Lol

 

[ErinH]

I suppose you could do it with enough distance from microphone.

Yes.

 

[ErinH]

It was asked on DIYA so I am sharing it here...

Sooo, to me, the main take-home message of your tests is that the ground-plane method easily yields consistent results at low frequencies, up to 500 Hz or so. Normal indoor gated measurements will go down to 300-400 Hz (anechoic). The ground-plane measurement might therefore be a useful method to extend this low frequencies (that's what the theory always told us, but now we KNOW).

Yes. But to reiterate, it makes more sense to perform an indoors ground plane measurement instead of the speaker-on-a-stand measurement for the reasons I listed previously. At least for me.

Sooo, to me, the main take-home message of your tests is that the ground-plane method easily yields consistent results at low frequencies, up to 500 Hz or so. Normal indoor gated measurements will go down to 300-400 Hz (anechoic). The ground-plane measurement might therefore be a useful method to extend this low frequencies (that's what the theory always told us, but now we KNOW).


This leads me to the following questions and ideas:
How does the ground-plane measurement compare to near-field measurements? This will tell us more about converting from the nearfield to the farfield.
How does the ground-plane measurement compare to "microphone-in-box" measurements [1,2]? In theory, the "microphone-in-box" method yields a 2pi farfield SPL response curve. How well does this work out in practice?


Would it be possible for you to take some nearfield and mic-in-box measurements of your speaker, and compare these to the ground-plane measurements?

The DUT I have been using is one with a passive radiator so I cannot perform the MIB test. However, I had previously captured nearfield response of both the woofer and the PR.

FWIW, I am using the sd of these drivers (as they appear to be the same ones used in the Buchardt S400, with assumed differences in electro-mechanical properties; however the physial dimensions are almost certainly the exact same):

SB Acoustics SB15SFCR39-8 5x8" Paper Cone Woofer

SB Acoustics SB15SFCR39-8 5x8" Paper Cone Woofer

www.madisoundspeakerstore.com

SB Acoustics SB15NBAC30-4 5" Black Aluminum Cone Mid-bass

SB Acoustics SB15NBAC30-4 5" Black Aluminum Cone Mid-bass

www.madisoundspeakerstore.com

With the above stated, the nearfield method has a maximum frequency accuracy dictated by the size (as discussed in the above audioexpress link). The 5" woofer is limited to about 980hz and the 5x8 inch passive radiator I will limit to the 8 inch dimension and therefore is invalid above about 600hz. Just spit-balling here; not using the actual effective radius. But it's close enough.



... On with the results ...

As you may know, when capturing near-field data it's hard to get the exact same level relative to distance of mic from cone and therefore the result is usually "eyeballed" to overlay them together, by using the lowest frequencies. I have done that below. What you see below is:

Woofer Nearfield (orange)
Passive Radiator Nearfield (green)
Summed Nearfield Response of the Woofer + PR (blue)
*Note the y-axis scale is 10dB. Using 5dB is too hard to read.

And here is the summed nearfield response above vs my ground plane (at 2 meters) measurement:

Understanding, of course, I had to "eyeball" where to overlay these in frequency, you see the presumption regarding NF responses showing a bump in the low frequencies due to the fact that it effectively ignores baffle step is indeed true: the 2 meter ground plane measurement is lower in amplitude on the low end than that of the NF components & summed response. This is to be expected and you can find this quote on why why stated in the above audioexpress link:

First, Keele assumes all radiating surfaces are mounted on an infinite baffle. Under this condition, the radiation is into a “half-space” or a solid angle of 2π. However, most loudspeakers have relatively narrow baffles so they become omnidirectional at low frequencies. For this reason, the Keele approach may over estimate the low-frequency sound pressure level.

The farfield ground plane technique is more accurate as you can see (and as was expected). The NF summing technique gives you an idea of the low end response but isn't entirely accurate to that of a far-field response AND it requires care in the measurement, assumptions and final calculation. Honestly, I prefer the ground plane measurement method simply because it's less prone to personal error. Meaning, the more ports and woofers I have to measure individually and then add together, the greater chance for a human error to occur. Whereas a ground plane measurement is just sticking the mic a few meters in front of the speaker, running a sweep and then you're done. I would only use a nearfield method if I had no other choice due to area available or terrible weather.

 

[KEW]

Just wanted to let you know somebody is following along.


Cheers!

Ditto ... although I think mainly because I'm infatuated with watching just how deep your OCD goes!

 

[Verdinut]

Very interesting and good testing.

By the way, I also like the SB Acoustics drivers. I have built several bookshelf speakers using the SB15MFC30-8 5 inch mid-woofers. They are solidly built and produce a rather smooth frequency response.

 

[ErinH]

I had someone PM me to ask if I modeled diffraction in the above NF summed response results. I think it's a worthwile question so I'll share my reply here as well:

I did not.

Ughhhh.... Ok. Curiosity got the best of me. But I swear, I AM DONE after this!....



Now, let me state: I have never used VituixCAD for this purpose but I *think* I did this correctly...
I opened the Diffraction Tool.
I loaded my NF curve in the Half space response section
I checked "full space"
I modeled the enclosure with the woofer in the appropriate place on the baffle.
I modeled the microphone about where the tweeter axis is.
I set the mic distance to 2 meters.
I saved this response.

I repeated the same process for the passive radiator (using a rectangular driver since it's an oval).

I then used VituixCAD's Calculator Tool to sum the those (2) curves. What you see below is a comparison of the original 2 meter far field in Red vs the different NF sums. In blue is the summed woofer & PR response *WITHOUT* diffraction modeled. In black is the summed woofer & PR *WITH* diffraction modeled.

As I posted previously, the simple NF sum of the woofer + passive radiator (blue) shows a boost on the low end that was expected. HOWEVER, if you apply the diffraction effect to each of the components and then sum them (black) you get a combined response that follows the 2 meter GP measurement (red) quite well. Close enough, in fact, that I would feel comfortable using this method in future tests if I don't have the ability to physically measure outdoors to get the LF extension in response I need. It isn't perfect but I think it shows the necessity for modeling the diffraction effects for those of you who are dealing with how best to accurately gather data for your reviews or your own designs.


Okay. That's it! I'm done!!! I have to quit obsessing over this topic. Thanks for joining me in this journey. But I've gotta bail.

 

[Danzilla31]

Ughhhh.... Ok. Curiosity got the best of me. But I swear, I AM DONE after this!....



Now, let me state: I have never used VituixCAD for this purpose but I *think* I did this correctly...
I opened the Diffraction Tool.
I loaded my NF curve in the Half space response section
I checked "full space"
I modeled the enclosure with the woofer in the appropriate place on the baffle.
I modeled the microphone about where the tweeter axis is.
I set the mic distance to 2 meters.
I saved this response.

I repeated the same process for the passive radiator (using a rectangular driver since it's an oval).

I then used VituixCAD's Calculator Tool to sum the those (2) curves. What you see below is a comparison of the original 2 meter far field in Red vs the different NF sums. In blue is the summed woofer & PR response *WITHOUT* diffraction modeled. In black is the summed woofer & PR *WITH* diffraction modeled.

As I posted previously, the simple NF sum of the woofer + passive radiator (blue) shows a boost on the low end that was expected. HOWEVER, if you apply the diffraction effect to each of the components and then sum them (black) you get a combined response that follows the 2 meter GP measurement (red) quite well. Close enough, in fact, that I would feel comfortable using this method in future tests if I don't have the ability to physically measure outdoors to get the LF extension in response I need. It isn't perfect but I think it shows the necessity for modeling the diffraction effects for those of you who are dealing with how best to accurately gather data for your reviews or your own designs.


Okay. That's it! I'm done!!! I have to quit obsessing over this topic. Thanks for joining me in this journey. But I've gotta bail.

You'll be back tommorow LOL

 

[ErinH]

Just an update...

I have been in talks with Klippel about my test method results. They have a module called ISC (In-Situ Compensation) and the long story made short is that I can use this module to get anechoic results in my garage based off a single outdoors measurement (per DUT). Some early tests proved incredibly useful. It is flat out amazing what can be done with this module. After talking some things over with their R&D tech I have gotten specific instruction on how I can use this in my case and am good to go.

I also ordered a few parts to build a computer controlled turntable so I can rotate the DUT at any angle I desire. I
built the prototype today. I figure I’d share a little of the progress. The attached videos show this thing in action. I will need to tweak the design a tad but overall it’s a success. It supports at least 185 pounds and as you can see, it has no problem spinning my kid around.


Once I complete it I'll do a little video in case anyone else wants to copy the design.

Hopefully I can be fully operational by next weekend and start knocking out some of these tests.

 

[Gmoney]

You'll be back tommorow LOL

I see you have Canoas to huh?

 

[ErinH]

I got started yesterday on actually completing the review of the Buchardt S400. For now, I am using my "manual" turntable to get measurements indoors until I have time/money to finish the motorized platform. The stand height is about 42 inches off the ground with the ceiling at 10 feet. With Klippel's ISC module you can use an anechoic "reference" measurement with an indoors measurement and subtract the room out of the result. You can read about it here: http://www.klippel.de/products/rd-system/modules/isc-in-situ-compensation.html

I will be using the outdoors ground plane measurement as my "reference" and getting anechoic data for the DUT. It's really impressive. The reason why I chose to go this route instead of using the ground plane measurement for all is due to a few things:
1) Accuracy in aiming. I showed in an earlier result that the difference in just a couple degrees can equate to relatively significant "errors" in the measurement. For instance, if the tilt of the speaker isn't lined up perfectly with the microphone the Buchardt S400 measurement was off as much as 3dB within a 5 degree tilt window. This issue is more trivial but it's something to keep in mind. Measuring on a stand in my garage allows me to laser-align the mic position relative to the location I want to measure (i.e., tweeter axis, between waveguide/mid, etc).
2) Noise. Ground plane measurements are subject to wind in high frequency. Aligning my available time with no wind and perfect weather isn't really feasible. So, using the GP method for lower frequency response helps alleviate this issue.
3) Temperature. High frequency drift is common in extreme temperatures. My garage is insulated and I also have a 220v heater that warms the garage quickly in the winter months. So as long as the doors are closed the garage stays within about 10-20 degrees of the house temperature which is enough to keep drift from occurring.
4) Comfort. With the ISC module I can create a room compensation curve and conduct additional tests such as distortion measurements indoors. No need to stand outdoors and conduct additional measurements and sweat/freeze.

Without Klippel's ISC module I would likely conduct my tests outdoors in the ground plane method only. But with the ISC module I am afforded some additional flexibility without loss in accuracy (actually, potentially better accuracy) which makes my life a bit easier.



Here's some shots I took of the measurement rig with the Buchardt S400 being tested. And, yes, there is a rear leg on the stand; it's just blocked in this angle.

 

[ErinH]

After about 140 different measurements and nearly 2 miles of walking between the computer and the DUT, I have completed a full-spin set in 20hz resolution of the S400.

In presenting the data, I tried to stick to what is in Dr. Toole's book for the curves (dashes, dots, etc) but realized they likely did things the way they did because they didn't have colors. So, I built the graph the way I think it looks best. Matching sound power and SP DI curve colors (blue) and the same for early reflections (red). Dashed lines for only DI curves. Solid lines for everything else. I added the y-axis on the right to mimic what is in Dr. Toole's book. But I prefer the legend outside of the graphic because it's distracting otherwise so I shot for the way Klippel provides it at the top below the default (Matlab doesn't have a default for this which is stupid so I had to code that up specifically).

Anyway, here's my first full set of CEA 2034 curves. Took nearly 2 solid months of testing methods, understanding trade-offs, etc etc but I finally arrived at a data set I am proud of. And now that I've gotten the method for testing down as well as the scripts written to populate the graphs in a manner I like, I can hopefully start knocking more of these out. First up, though, I have to actually do a real review for the Buchardt S400 and get the data on my site.


A couple notes:
I provided two versions of the spectrogram: one is the full 360-degree view (from -180 to +180), with the frequency axis extending from 20 - 20kHz. The second version is a 'zoomed' version with a tighter window of ±90° and frequency axis from 200 - 20kHz. I will probably always do this. I also added some labels on the y-axis to indicate the direction the measurement was taken relative to the speaker. I think this helps gives newcomers an idea of what they are seeing.

I have also provided two different versions of the predicted in-room response: one with the target curve (1dB/octave) starting frequency at 100hz and the other starting at 200hz. In this speaker's case it doesn't much matter, though you can see a slight difference higher in frequency. However, if I were to have a speaker with a high-Q bass hump around 100hz (which I've seen before) then that would obviously effect how the predicted curve lines up with the target line. Plus, given the room is dominate below about 200/300Hz, I am more inclined to go with the version that starts the target line at 200hz. But your feedback and rationale is welcome.

I purposely provided the ±40° vertical response to mimic each other (same color for + & -). Also, adding all the curves to this graph creates a cluster in one's brain (same for the horizontal plot if I were to extend beyond 90 degrees). Really, to get an idea of directivity one should just look at the normalized spectrogram. I typically look at the on-axis and then the spectogram to see how the off-axis response corresponds to the on-axis. But I can make some changes or additions if enough people really see the need.

If you spot anything that you don't necessarily like feel free to make a suggestion. Just keep in mind I have been at this for days and may resent your feedback. LOL.

Now, here's the alternative version of the Predicted In-Room response vs a target curve starting at 100hz.

 

[ryanosaur]

@ErinH
Considering you were using Amir's testing as a benchmark of sorts... how do you feel in comparing/contrasting what you have achieved now vs your earlier measurements?

 

[ErinH]

@ErinH
Considering you were using Amir's testing as a benchmark of sorts... how do you feel in comparing/contrasting what you have achieved now vs your earlier measurements?

Well, one thing to note specifically with my test of this speaker is that Amir used the tweeter as his reference angle. I used the midpoint between tweeter and woofer (per the manufacturer's direction).

I am quite satisfied with my data. Its high resolution provides detail that cannot be seen when the typical gating of 3-5ms is applied. It even provides higher accuracy than when I was using an 8 foot stand and getting 10ms+ of reflection-free window time (10ms ~ 100hz resolution which means frequency resolution of 100hz; you would miss the detail high-Q peak that shows up around 500hz in my data with 100hz resolution and you would miss it entirely with 200 or 300hz resolution). A near-field measurement creates a low-end bump where the speaker is transitioning/transitions to 4-pi down to 2-pi so that's an alternative one but a misleading one if this isn't explained or otherwise compensated in the data. I am deliberately avoiding these common practices and am therefore using Klippel's ISC module, using the outdoors ground plane measurement as my reference to obtain room correction and then using the more accurate 4-pi measurement with the room correction applied. Though, I performed this test outdoors for 4-pi with the speaker about 7 feet off the ground. I will create a video regarding the different methods I use at some point. But you can read about how the ISC module works and you can watch the webinar if you are so inclined: https://www.klippel.de/know-how/education/webinars.html (#4)

Anyway... I feel more confident in what I have now. Those earlier measurements were all me playing around with different methods and trying to understand and weigh the pros/cons. In some cases I didn't even use the calibration file for the mic (because it wasn't of concern at the time). Until something is actually "published" it is not done. That is why I have yet to publish any data as complete. This set of data is nearly there once I iron out some of the graphic cosmetics (ie, font size, colors, etc). Once I get that nailed down I'll complete the review, post the data to my site and be done with this speaker and ready for another test.

 

[KEW]

When can I drop off a Canton Vento for you to measure?

 

[ErinH]

When can I drop off a Canton Vento for you to measure?

Not sure yet. I am waiting to hear back from a couple others regarding their stuff.

 

[shadyJ]

Looks very nice. I remember when I started getting good data after all the effort, it's like scaling Everest. Have you sent your measurements to Buchardt, and if so, did they confirm that your results matches theirs? You have very good similarity to Audio Science Review although not a perfect match. It's possible that your measurements could be more accurate. I should send you a speaker to see how close our respective measurements are.

One suggestion I have is that you might make the lines a bit darker. Some of the lighter colored lines in your dispersion graphs are a bit hard to make out against the white background.

 

[ErinH]

Looks very nice. I remember when I started getting good data after all the effort, it's like scaling Everest. Have you sent your measurements to Buchardt, and if so, did they confirm that your results matches theirs? You have very good similarity to Audio Science Review although not a perfect match. It's possible that your measurements could be more accurate. I should send you a speaker to see how close our respective measurements are.

I haven't sent Mads the final graphs but I have sent him earlier ones when there was concern about the discrepancy between his and ASR's. Which, as you can guess, means neither Buchardt's nor ASR matched each other even though they were both measured using Klippel's NFS (which also may explain/alleviate concern for why mine also doesn't match Amir's). Additionally, as I stated above, Amir measured on the tweeter axis where I followed the manufacturer's suggestion to measure between the waveguide and woofer. Given the number of variables I don't expect the results to ever be the exact same between setups; even the CEA-2034 spec provides some leeway in this paragraph:

The following techniques are useful in assessing the frequency response of a loudspeaker in the absence of an anechoic chamber. Many of them work well. All measurements have errors, including measurements in anechoic chambers and outdoors, so it is important to validate your own techniques by comparing the results to a measurement made in a free field environment. A good way to do this is to submit one of your speakers to a recognized testing laboratory. An agreement of ± 1.5 dB is considered to be good.

This helps take some stress off.

One suggestion I have is that you might make the lines a bit darker. Some of the lighter colored lines in your dispersion graphs are a bit hard to make out against the white background.

Noted. Truth be told, I am only providing those for the people who will pitch a fit if I don't show them. They only show a percentage of all the radial measurements; the spectrogram is what should be used to determine how the off-axis response fares against on-axis.