Poor man's directivity response measurement technique

[Sokel]

Isn't the reason trying to use MMM in this thread right now because not going to move the speaker or rotate it? Did I get it wrong somethin?
I understood that the writer was trying to find a simple way to do it with a speaker that was already made, and a speaker that was already set up and placed, rather than a speaker design.

Yes,that's the purpose,the difference is that the MMM method on the other thread is to integrate a speaker to a room,not to design it.

 

[thewas]

Two little comments from me:

a) The measuring distance shows also the influence of the baffle step.

b) The main question always is for what purpose the measurements will be used. For most of us not loudspeaker designers but end users we need them mainly for loudspeaker and room correction and for both such MMM (nearfield and at the LP) can be used as I have showed here.

 

[ernestcarl]

Two little comments from me:

a) The measuring distance shows also the influence of the baffle step.

b) The main question always is for what purpose the measurements will be used. For most of us not loudspeaker designers but end users we need them mainly for loudspeaker and room correction and for both such MMM (nearfield and at the LP) can be used as I have showed here.

Would you say 30-50 cm distance would be likely fine for many bookshelf type speakers?

I don’t think I understand your rising measured curves. I mean, my somewhat very nearfield MMM and sweeps do not present such an incline.

 

[thewas]

Would you say 30-50 cm distance would be likely fine for many bookshelf type speakers?

In my experience for the region I correct the direct sound (above 500 Hz) not only for bookshelf since most floorstanders are usually not much wider but just longer.

I don’t think I understand your rising measured curves. I mean, my somewhat very nearfield MMM and sweeps do not present such an incline.

That example was from loudspeakers with a rising treble, for my neutral loudspeakers I also don't have such an incline.

 

[3ll3d00d]

Doctors have a different understanding of measurements to engineers. As a doctor, I know that even imperfect or inaccurate measurements have value. It tells me where I should look, and what investigation I should arrange to give me a better picture. I never start off with the "gold standard" investigation unless the pre-test probability strongly indicates that the test will give me useful and actionable information. I don't order PET scans as a screening test for cancer, I only order it when I already know the patient has cancer.

It's a bad analogy or at least I really hope it's a bad analogy ie a medical "pretest" is a test that is generally considered to provide sufficient signal to warrant being followed up by an entirely different test that is known to be more accurate but is invasive/expensive, in comparison this is just a low resolution version of the actual test where the resolution is low enough to render the results inaccurate enough to be simply wrong.

More practically, the bit I don't understand is, given it's an active speaker, is why you measured it as an entire speaker? and also why you added the moving mic bit which isn't really moving mic at all given it is just moving up and down?

 

[Blumlein 88]

More practically, the bit I don't understand is, given it's an active speaker, is why you measured it as an entire speaker? and also why you added the moving mic bit which isn't really moving mic at all given it is just moving up and down?

If you've read the Magic Beans thread or seen the video with it, you wound have seen joentell, who is behind the Magic Beans software, demonstrate this near field MMM by moving the mic vertically. It didn't look right to me the first time. Which is why I suggested in this thread the vertical needs to be combined with horizontal measures. Or use a different moving pattern. Some sort of figure 8 or spiral would be my guess as to a good pattern.

I'm not convinced this is going to get us anywhere really. While you are reducing the effects of the room, you are increasing the possible problems of speaker directivity without averaging it out properly in my opinion.

 

[kemmler3D]

To the above point something occurred to me about this method: Even if you want to fully average out vertical directivity and look at horizontal only, a straight up-and-down motion is actually varying the mic distance from the measurement axis quite a bit. You'd need to do a curved motion at the right radius to get it exact.

That said, maybe a vertical MMM can be more informative if you keep it at listener / ear height, within a certain smaller range of motion?

 

[wwenze]

Method #1
Move your furniture out of the way.

You can move furniture out of the way? How is this even poor man's method given today's housing market

 

[3ll3d00d]

This thread says poor man's directivity but moving it vertically at specified angles implies vertical directivity is constant or you can't tell these two things apart. If the point is "I don't know how my speaker behaves" then you need to be able to distinguish between those two effects. The varying distance point @kemmler3D makes is also a v good one, there's a reason why the centre of rotation is important when doing this after all.

 

[Keith_W]

It's a bad analogy or at least I really hope it's a bad analogy ie a medical "pretest" is a test that is generally considered to provide sufficient signal to warrant being followed up by an entirely different test that is known to be more accurate but is invasive/expensive, in comparison this is just a low resolution version of the actual test where the resolution is low enough to render the results inaccurate enough to be simply wrong.

Well let's leave the medical discussion out of it for now I would disagree that the resolution is so low that the result can not be interpreted. It was enough to pick up the difference in directivity between my horns and the woofer. And more importantly, the trend from the raw, unprocessed measurements is consistent with what I would expect to see if the woofers have a wider radiation pattern than the horns.

There is something even more remarkable with those measurements. If you look at the directivity index in this graph (green line):

You will see a sudden rise at about 13kHz. This shows the tweeter horn suddenly changing direction from wide dispersion to a beam.

When this was pointed out to me, I got a tape measure out and measured the diameter of the tweeter horn - 12cm. With horns, the lowest supported wavelength is twice the diameter of the horn, anything longer than that is not "horn coupled" and loses efficiency. So we have the lowest wavelength of 24cm, which corresponds to 1430Hz. The upper supported frequency is 3 octaves above the lowest frequency, so 11.5kHz, that is close enough to 13kHz (allowing for errors in my technique, not knowing the exact geometry of the horn, etc). Anything above the highest frequency is not horn loaded either, and it starts to behave like a beam. I was pretty amazed to see that this measurement was able to pick it up.

More practically, the bit I don't understand is, given it's an active speaker, is why you measured it as an entire speaker? and also why you added the moving mic bit which isn't really moving mic at all given it is just moving up and down?

If you are asking why I did not measure the drivers independently, it is because I wanted a rough comparison of the directivity of the horns vs. the woofers. I could do that by measuring the drivers independently, but that would give me twice as many curves to process, and the result would not be side by side in the same graph. I also wanted to see if my very steep crossover causes sudden changes in directivity, which it does. For me, this result is actionable - over the next few days, I will reconfigure the crossover between woofer/horn to something more gentle. I will repeat the measurements and post them here.

As for moving the mic up/down, my understanding of "MMM" is that you move the mic across the area that you want to cover. In a normal MLP MMM, the mic moves around your head. I wondered whether a vertical MMM would give me useful information, which it appears to do.

Look, I am far from an authority on any of this. I am sure that any of you know far more than I do. I am a tinkerer, after all this is only a hobby for me, I do not do it professionally. I often come up with hare brained ideas, which I only find out are hare brained in retrospect and after getting brutally savaged in ASR after discussing them So I don't mind getting savaged, I am humble enough to know that some of my ideas are just plain stupid. If this one is stupid, then so be it. It cost me nothing to do, and it is "backyard science".

 

[3ll3d00d]

I also wanted to see if my very steep crossover causes sudden changes in directivity, which it does. For me, this result is actionable - over the next few days, I will reconfigure the crossover between woofer/horn to something more gentle. I will repeat the measurements and post them here.

Fwiw this is where having individual curves comes in and also using a tool that can show this effect directly as you simulate different crossovers. You can do this in vituixcad nicely (export the xo from acourate as a wav, import to a filter block in vituixcad). I expect I have my data somewhere, if so can post an example to illustrate if you like.

 

[Keith_W]

Fwiw this is where having individual curves comes in and also using a tool that can show this effect directly as you simulate different crossovers. You can do this in vituixcad nicely (export the xo from acourate as a wav, import to a filter block in vituixcad). I expect I have my data somewhere, if so can post an example to illustrate if you like.

Yes please. I have a copy of VituixCAD and I need to figure out how to generate polar plots with it. There is a thread somewhere on ASR that I meant to dig up.

 

[5-pot-fan]

I hope the occasional 'brutal savaging' does not affect your desire to continue with this and future similar threads.
It is useful to me, and I hope to other newcomers, to see the sort of ideas we might have, but dismiss, being offered by someone with knowledge and experience who can also defend his proposals cogently.
Altogether a good example of why I like this site!

 

[MAB]

Yes please. I have a copy of VituixCAD and I need to figure out how to generate polar plots with it. There is a thread somewhere on ASR that I meant to dig up.

Here I think!

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

Quasi-anechoic measurements are a way to capture the 'real' frequency response of a speaker without the influence of a room. They are called 'quasi-anechoic' because they have the potential to get you really close to the results one might achieve with a non-reflective environment like an...

www.audiosciencereview.com

 

[3ll3d00d]

Yes please. I have a copy of VituixCAD and I need to figure out how to generate polar plots with it. There is a thread somewhere on ASR that I meant to dig up.

here's an example - https://drive.google.com/file/d/11U3w-HPyidYJmrIfeXoJPNxPJqKc4pQ_/view?usp=sharing

I would expect to be able to open the project as is, you'll see it has H 0-90 degree data for my M and T and each driver has 2 G(f) filters with an associated text file (which was created by exporting the acourate filter to a wav, opening that in REW and exporting as txt). The 1st one (invert) is the output of driver linearisation step in acourate, the 2nd one is a high or low pass filter created in acourate.

you can look at 1 driver or the other by shorting the other driver

double click the directivity view to maximise it and switch it to polar map (right click)

if the linear filter is enabled, look at it unnormalised
if its disabled, normalised is easier to see the pattern

contour lines are every 3dB, if you look at the mid then you see it narrows rapidly above 1kHz (from ~50 degrees to ~25 degrees) and it gets v messy above 2kHz, it's an AE TD12M so this is expected behaviour. 45 degrees is in the region of 1200-1300Hz

if you look at the T, it's roughly stable at ~37 degrees down to 1500Hz and it broadens more below here and is rolling off hard under 1kHz or so, again expected as it's a BMS4550 on a SEOS15

from this you can see that, based on directivity and driver behaviour, XO is likely going to be somewhere in the 1200-1500Hz range

if you then enable both drivers with the xo & short the invert filter then you'll see the combined response given that particular filter (it's centred on 1200kHz)

you could play with putting different filters (just export a few different slopes from acourate) into that G(f) slot to see the impact it makes