AVS › AVS Forum › Audio › DIY Speakers and Subs › Close mic measurements of a passive radiator sub
New Posts  All Forums:Forum Nav:

Close mic measurements of a passive radiator sub

post #1 of 22
10/5/12 at 3:52pm
Thread Starter 
I have a sub I built years ago with a Tumult and 4 Adire 15" PRs that has been sitting in storage until I got my hands back on it today. Frankly speaking I've always been unhappy with its performance. I'm trying to figure out why it sucks. Step one is to measure it and compare it to the predicted model. Dragging it out into the middle of a field and measuring it ground plane isn't practical. So... I can close mic measure (with REW) the driver with and each of the 4 passive radiators and then add them all together and get a reasonably accurate measurement of the sub (assuming I'm consistent with the distance between the mic and each driver / PR) right?
Reply
Reply

AVS Top Picks

post #2 of 22
10/5/12 at 6:23pm
Thread Starter 
IE: Can I combine all these:

individuals.png

and get this:

combined.png

and consider this a valid measurement of the subwoofer's overall frequency response?
Reply
Reply
post #3 of 22
10/6/12 at 6:35am
If done correctly, close mic will tell you the anechoic response (what it would measure out in the field, speaking loosely) but it won't tell you what it will sound like in a room where response is dominated by room-modes/standing waves.Yes you can add responses from diagram to contributions from the ports or passive radiators. In each measurement the mic distance has to be not only constant but as close as possible to the source to limit contributions from the other sources.

The close mic response dosen't include reflections but, if the sub doesn't sound good, chances are its the reflections (leading to cancellation notches and standing waves) that are the cause. What you need to do is measure it in room with mic at the listening position and then try different positions for the sub and also try to correlate notches with distance to nearby floor/ceiling/wall boundaries.
Reply
Reply
post #4 of 22
10/6/12 at 7:25am
What steps are you taking to get the combined response? It is possible to measure each portion separately and sum the two to know what the system is doing, but you have to do an actual summation that also takes the phase information into account. The combined response you're showing doesn't seem to be showing that.

Do you have the winISD model that predicts the expected performance?
Reply
Reply
post #5 of 22
10/6/12 at 7:25am
Thread Starter 
Quote:
Originally Posted by JackNC View Post

If done correctly, close mic will tell you the anechoic response (what it would measure out in the field, speaking loosely) but it won't tell you what it will sound like in a room where response is dominated by room-modes/standing waves.Yes you can add responses from diagram to contributions from the ports or passive radiators. In each measurement the mic distance has to be not only constant but as close as possible to the source to limit contributions from the other sources.
Thanks for the reply.

I realize that the room is the other dominant factor and that close mic response / quasi anechoic ignores that. My first goal is to find out if the sub is anywhere close to the predicted model in the simulation software. In the graphs I posted above the mic distance was perpendicular and a little over 1" away from the surface of the center of the driver or PR. It was as consistent between the driver and PRs as I could get. I will try again later today with the mic closer and see if anything changes.

My concern is that the software simulation says I should get this frequency response:

220lsimulation.gif

But, I measured this:

combined.png

Not exactly close. mad.gif
Reply
Reply
post #6 of 22
10/6/12 at 7:43am
Thread Starter 
Quote:
Originally Posted by MBentz View Post

What steps are you taking to get the combined response? It is possible to measure each portion separately and sum the two to know what the system is doing, but you have to do an actual summation that also takes the phase information into account. The combined response you're showing doesn't seem to be showing that.
I used the mathematical addition function in REW. I'm pretty sure it considers phase since the result of adding two traces together shows phase. I just didn't show the phase in the graphs I posted.

Here's a graph showing both SPL and phase: (Click the thumbnail for full size)



Quote:
Do you have the winISD model that predicts the expected performance?
WinISD, no. Unibox, yes.
Reply
Reply
post #7 of 22
10/6/12 at 1:48pm
Thread Starter 
So, I went back and measured again today. This time the mic was about 7/16" from the driver / PRs and was as consistent as possible. I also calibrated the sound card in between yesterday and today though the effect is basically negligible.

Day 2 Results: (Click the thumbnail for full size)



Comparison of yesterday and today: (Click the thumbnail for full size)



I feel pretty comfortable that this is the net frequency response of the sub and it's nowhere near the predicted response. frown.gif
Reply
Reply
post #8 of 22
10/6/12 at 2:03pm
The measured response is acting like the Vas is a little smaller, fs a litte lower, and the Qts a litte higher than spec'd.
Reply
Reply
post #9 of 22
10/6/12 at 4:18pm
Do you have all xovers disabled?
Reply
Reply
post #10 of 22
10/6/12 at 5:12pm
You generally can't just sum up the responses, you also have to take into account their area. The microphone when close measures only the pressure right where it is at, but that pressure generally exists across the entire area of the driver or passive radiator (or a port), and its the pressure times the area (related to the amount of air that moves) that counts.

So if the moving area of the port or passive is P and the area of the driver is D, then you have to correct the passive's contribution by adding the value (in dB) to it's curve before summing:
10*log(P/D)
If the passive has less area than the driver, then you'll be adding a negative amount of dB.

If all the passives are the same, measure one of them, and in the correction formula, make P equal the sum of all their areas, then just sum those two values (driver and the combined passives)
Edited by bwaslo - 10/6/12 at 5:20pm
Reply
Reply
post #11 of 22
10/6/12 at 5:12pm
could all that be caused by very large inductance? inductance can create a peak as well as rolloff the top end. i just can't remember seeing it that bad.

ilkka measured one in a sealed.



http://www.hometheatershack.com/forums/subwoofer-tests-archived/5753-diy-adire-audio-tumult-15-mk1-sealed-100l.html
Edited by LTD02 - 10/6/12 at 5:26pm
Reply
Reply
post #12 of 22
10/6/12 at 6:03pm
Thread Starter 
Quote:
Originally Posted by MBentz View Post

Do you have all xovers disabled?
I fed the output of the REW system straight into my QSC PLX 2402. No crossovers, EQs or filters were in play.
Reply
Reply
post #13 of 22
10/6/12 at 6:06pm
Thread Starter 
Quote:
Originally Posted by bwaslo View Post

You generally can't just sum up the responses, you also have to take into account their area. The microphone when close measures only the pressure right where it is at, but that pressure generally exists across the entire area of the driver or passive radiator (or a port), and its the pressure times the area (related to the amount of air that moves) that counts.
So if the moving area of the port or passive is P and the area of the driver is D, then you have to correct the passive's contribution by adding the value (in dB) to it's curve before summing:
10*log(P/D)
If the passive has less area than the driver, then you'll be adding a negative amount of dB.
If all the passives are the same, measure one of them, and in the correction formula, make P equal the sum of all their areas, then just sum those two values (driver and the combined passives)
Thanks for the reply. I will do this later when I get back on the system with REW and the measurements.
Reply
Reply
post #14 of 22
10/6/12 at 7:34pm
Umm, I think I made a goof there.
Should be 20*log(P/D).
Reply
Reply
post #15 of 22
10/6/12 at 8:41pm
Thread Starter 
Quote:
Originally Posted by bwaslo View Post

Umm, I think I made a goof there.
Should be 20*log(P/D).
So, the passives are 824cm^2 and the Tumult is 749cm^2. So assuming I understand your post correctly... Per the formula I should add 0.83dB to each passive's measurement before summing the responses.

Without further ado, here's the new composite (blue trace) and the old composite (red trace): (Click the thumbnail for full size)



It only amounted to .9dB at the largest point though.
Reply
Reply
post #16 of 22
10/6/12 at 9:29pm
Thread Starter 
Quote:
Originally Posted by LTD02 View Post

could all that be caused by very large inductance? inductance can create a peak as well as rolloff the top end. i just can't remember seeing it that bad.
ilkka measured one in a sealed.

http://www.hometheatershack.com/forums/subwoofer-tests-archived/5753-diy-adire-audio-tumult-15-mk1-sealed-100l.html
It'd have to be double digit inductance like 20 mH to cause that sort of roll off.
Reply
Reply
post #17 of 22
10/6/12 at 9:50pm
measured at what frequency? like impedance, inductance varies with frequency and is commonly measured at 1khz, which is kind of meaningless for subs.
Reply
Reply
post #18 of 22
10/7/12 at 5:03am
Thread Starter 
Quote:
Originally Posted by LTD02 View Post

measured at what frequency? like impedance, inductance varies with frequency and is commonly measured at 1khz, which is kind of meaningless for subs.
Well, I was simply increasing the Le in Unibox until I saw similar roll off characteristics.
Reply
Reply
post #19 of 22
10/7/12 at 10:03am
Thread Starter 
Okay, so I pulled the driver and close mic'd it free air and got this: (Click the thumbnail for full size)



I guess there's a massive amount of inductance. confused.gif
Reply
Reply
post #20 of 22
10/7/12 at 10:18am
Thread Starter 
While I had the driver out I measured it with my DATS with the coils in series and got this: (Click the thumbnail for full size)



Feeding the measured T/S parameters back into the Unibox model gives me this: (Click the thumbnail for full size)

Reply
Reply
post #21 of 22
10/7/12 at 10:33am
...in addition non-linearities in the inductance. inductance creates back emf. if the back emf is significantly different at different frequencies, the frequency response will be non-linear.

if you eq it for flat response, does it still suck?
Reply
Reply
post #22 of 22
10/7/12 at 3:48pm
Thread Starter 
Quote:
Originally Posted by LTD02 View Post

...in addition non-linearities in the inductance. inductance creates back emf. if the back emf is significantly different at different frequencies, the frequency response will be non-linear.
if you eq it for flat response, does it still suck?
I haven't tried that. I did replace the Tumult with an AE AV15-H D2 and was pretty impressed. It sounds much better and not surprisingly measures much better too.

Comparison between the two drivers in the same 220L box with the same PRs: (Click the thumbnail for full size)



I guess trying to buy the biggest and baddest driver around doesn't mean you're going to get the best performing sub. Or, in this case even a decent performing sub. $384.84 down the drain after shipping (~10 year ago) and a lesson learned. redface.gif
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: DIY Speakers and Subs

AVS Top Picks

AVS › AVS Forum › Audio › DIY Speakers and Subs › Close mic measurements of a passive radiator sub