Nearfield cone measurement doesn't appear to reveal system tuning

[chris661]

So I've split off from my thread on a compact 15" PA subwoofer, because I think this is worth some consideration.

First up, sorry, Art, it must've been frustrating telling me to play sine tones to determine the tuning, but I did, and its about 37-38Hz.

Anyway. I observed that the port output of this particular cabinet peaks around 36Hz and then drops quickly. The impedance curve suggests a 36Hz tuning, give or take.
If you put a measurement mic on various locations on the cone, you get notches in output at different frequencies depending on where the mic is.

Here're the relevant bits of of other thread:

So, this is just a bunch of overlays showing what happens when you move the measurement mic around over the cone. You might've thought the centre of the cone would yield an accurate result, and you'd be wrong.

An externally hosted image should be here but it was not working when we last tested it.

Here's a few data points. It'll be location followed by the notch frequency. The cabinet was horizontal.

Edge of dustcap furthest from port, 35Hz
Dustcap nearest port, 33.5Hz
Edge of cone nearest the port, 23Hz and 26Hz (this one is obvious on the graph)
Edge of cone furthest from the port, 35Hz
Edge of cone nearest the floor, 33Hz
Edge of cone nearest ceiling, 33Hz

Looks like there's some kind of cancellation happening between the port and driver, even though the mic was 1/2" away from a 15" cone. That's not something I've come across before.

(snip)

Here's a quick shot I took while outdoor testing.

An externally hosted image should be here but it was not working when we last tested it.

Cheers, all.
Chris

... and Brian's response:

Very interesting results there. Throws the whole approach of determining Fb by measuring driver response into question.

Luckily I always go instead by a 3-second impedance response measurement via WT3 🙂.

So, er, yes. Anyone any idea what's happening?

Cheers
Chris

 

[Chris Daly]

Try turning it upside down, with port at the top and see what differences occur in measurements.

 

[CharlieLaub]

The most common way to find the tuning of a vented/ported system is from an impedance measurement. Why are you using a frequency measurement?

 

[chris661]

Hi Chris,

I'm afraid the cabinet has since gone to be recycled. Storage space is tight here.
The graphs above are all nearfield, with the cabinet on its side. The outdoor testing was to figure out the final rolloff of the cabinet. After some max-SPL testing (~123dB at 37Hz at clip on an NU6000), I decided I was happy enough to build four more when life allows.

Edit - cross-post with Charlie
The paper from Jeff Bagby, "How to Achieve Accurate In-Room Quasi-Anechoic Free-Field Frequency Response Measurements Down to 10 Hz", recommends close-micing the woofer and port in order to sum the two. My results suggest this may not be accurate - none of the notches shown are at the frequency found with sustained sine tones. I respect Jeff's knowledge on the subject matter, but the errors caused by the near-field technique had me (and probably a couple of other forum members) quite frustrated when they didn't line up with other techniques
A quote from the paper: "Here’s a tip – what is the best indicator of the tuning frequency (Fb) from this data? The peak in the port’s output usually appears a little higher in frequency than the actual tuning frequency, so it is not the best indicator. However, the deep notch in the woofer’s response will be right on the actual tuning frequency of the enclosure and port. Do a near-field measurement at your cone and you will find the tuning frequency very quickly."


Chris

 

[weltersys]

Do a near-field measurement at your cone and you will find the WRONG tuning frequency very quickly."

Chris,

I fixed Jeff's observation for you ;^).

Cheers,

Art

 

[CharlieLaub]

It's possible that you are getting the port output picked up when you are doing the nearfield of the woofer cone. This is because at tuning the cone output falls to almost zero while the port output is peaking. If the port is far away from the cone and the tuning frequency higher, they are still somewhat isolated and you can resolve the minimum in the woofer output. In your case port and cone are right next to each other. The curves look similar, but with small differences in the minimum. This could be due to the cone influencing how much port output is being picked up. At least that is my guess that would explain the data.

Again for emphasis, I recommend using the impedance to fine the tuning.

 

[bentoronto]

It's as if nobody ever heard the word "phase" in previous thread. And for all the close attention to sim'ing, not much grasp of BR theory on which modelling programs are built.

So based on classic BR design (which a builder does not necessarily have to adhere to), here're notes that might be helpful.

The impedance curve reflects the cone motion in air and in all enclosures. The big low-end rise is due to what is quaintly called "back EMF" where the VC acts as a generator which gives the impression that the impedance is rising. And some inductance.

Boxes are resonators which are tuned to the driver resonance. So you should see about equal peak on the impedance curve which confirms classic construction. In addition to the need to sequester rear waves, the reason they are resonators is because they clamp down on the driver's natural tendency to fly off into space at resonance... but without using a sealed box to constrain it. And there is hope to use some of the rear wave constructively.

Above resonance, the port pours out sound progressively in phase with the cone and so provides a boost. Below resonance (just where you'd like some boost), the port is out of phase and so detracts from the cone sound (I know that must sound lousy... and it is).

The net sound output is a mix of cone motion and the additive and subtractive port sound.

All mic measurements (esp outside anechoic chambers) are tricky. For a BR box, you have to keep in mind the two sound sources, their phase, and where your mic is in relation to sources and phases.

Ben

 

[weltersys]

Above resonance, the port pours out sound progressively in phase with the cone and so provides a boost. Below resonance (just where you'd like some boost), the port is out of phase and so detracts from the cone sound (I know that must sound lousy... and it is).

Ben,

At Fb, (Box tuning Frequency/Helmholtz resonance) cone movement and impedance is at minimum, and port output is in phase with the cone. Above and below Fb, the output of the port is not in phase with the cone, but port output is minimal. The driver's Fs (free air resonant frequency) is completely independent of Fb, Fb remains exactly the same regardless of driver used, which is why careful attention to "alignment" is required if flat response is desired.

As you have often pointed out, flat response may not be desirable.
There is no need to slavishly design for flat, when high excursion drivers, power and DSP are available to fix any EQ or phase "problems" non-flat designs may present.

Long ports do have an upper "pipe resonance" that can be a problem for "full range" applications, but in a sub-woofer application, the pass-band is usually several octaves below the pipe resonance, so is a non-issue.

Properly designed phase-inversion (the fancy term for bass-reflex) generally sound better than sealed cabinets due to the greatly reduced cone excursion in the bottom octave, and the far less power compression that the increased sensitivity affords.

That said, for low SPL requirements, sealed cabinets are fine, and require almost no design- just stuff the driver in as big a box as you care to use, and keep an eye on the cone excursion, turn it down before it reaches Xlim/Xmech.

Art

 

[bentoronto]

Above resonance, the port pours out sound progressively in phase with the cone and so provides a boost.

Sorry, brain fart. Thanks to Art for waking me up.

Above resonance, the speaker produces more output because the driver is unimpeded by the box; output from the port is out of phase and so it would be destructive but it is inconsequential in volume.

B.

 

[tb46]

Hi Y'all,

Here is a link to the measurements of a typical bass reflex enclosure in Stereophile:

Direct Acoustics Silent Speaker II Measurements | Stereophile.com

I'll attach a screen print of the response. You can clearly see where the outputs are reinforcing, and where they are destructive. Also, the port resonance is particularily obvious in this example.

Bye the way, I think Chris's measurements are from a OD-TL (offset driver transmission line) that was designed for minimum external volume while still meeting low output requirements (with all the compromises entailed):

http://www.diyaudio.com/forums/subwoofers/272492-teeny-tiny-pa-15-subwoofer-2.html


Regards,

 

[bentoronto]

Nice. Clear. Thanks.

But an impedance measurement reflects the motion of the cone - which is of interest by itself. The near-field driver mic trace reflects the cone sound output which needs increasing motion to make sound for a given size of driver as you get lower freq.

B.

 

[weltersys]

The near-field driver mic trace reflects the cone sound output which needs increasing motion to make sound for a given size of driver as you get lower freq.

Ben,

In a sealed enclosure, excursion does need to progressively increase for a given SPL as the frequency decreases.

In a BR enclosure, cone excursion will also progressively increase for a given SPL as the frequency decreases to a frequency approximately 1/3 octave above Fb, then excursion progressively decreases to it's minima at Fb, then rapidly increases below.

BR enclosures are a continuum, an infinite series of "alignments" are possible, each resulting in a different relationship between port and direct cone output and phase.

Depending on the "Q" of the BR "circuit", the response of the cabinet can either look like a "classic" BR with flat response to near Fb, then a 24 dB per octave roll-off, or like the OP's "too small" enclosure, which starts it's roll off several octaves above Fb, but rolls off closer to the rate of a sealed enclosure below Fb.

Art

 

[bentoronto]

I hope readers will notice that in the quote I'm talking about the driver. Art is talking about the speaker.

While there is no legal or moral obligation to adhere to the classic alignment, it does make discussions of principles simpler. And I wouldn't be surprised if sim programs are less reliable as you vary from the classic model.

B.

 

[weltersys]

I hope readers will notice that in the quote I'm talking about the driver. Art is talking about the speaker.
While there is no legal or moral obligation to adhere to the classic alignment, it does make discussions of principles simpler. And I wouldn't be surprised if sim programs are less reliable as you vary from the classic model.

Ben,

Since the thread is about the near-field cone measurement in a BR system, I wrote about the differences between it and the sealed or open-baffle response you mentioned.

BR and TH ("tapped horn" and FLH (front loaded "horn") can all be "tuned" or "aligned" for a variety of different response throughout their pass band, they all share some aspects of design with sealed or open-baffle response, and each alignment of each type of cabinet has individual aspects that can be predicted with a fair degree of accuracy it the T/S parameters of the drivers are correct.

The "classic" BR formulas which use a calculator to predict the response of particular "alignment" are just as "reliable" for non-flat alignments as they are for "classic" "Flat" T/S alignments.

I can't speak for all simulation programs, but Hornresp uses the same "top secret" T/S formulas that Mark Gander of JBL shared with me back in the early 1980's. As the statute of limitations has expired several decades ago, I am sharing that "confidential" material below.
I think you will appreciate the statement at the end of Page 3 (of 6) before or after "doing the math" outlined in the first two pages.

Hornresp makes playing with the different alignments far easier than punching in numbers into a calculator for 10 minutes (right, I'm a slow typist) then converting the numbers it spits out into a semblance of what the response would "look" or "sound" like, but it's the same routine.
And that same routine still assumes that the driver has a response that mimics a flat piston with a uniform BL and suspension stiffness up to and past Xmax, as well as ignoring the actual native frequency response of the driver.
Because of those differences, the sim will get you well onto the ballpark's playing field, but without proper building, actual measurement and listening evaluation, a simulation (whatever the processing power) by no means assures a "home run" each time at bat.

That said, I no longer start to butcher wood without forcing myself to use the disgusting little PC I use to run Hornresp and a DSP program on.

Art