Cheap and FAST OB, Literally

How did you guys learn this stuff?

And the polar plot of +90° to 0° to -90°, with a 2.5ms right gate...

Setting up and measures took roughly 1hr, preparing the data and plots, about the same...

If 80dB SPL is to low, I can re-do this set before I make the next cut on the U-shape.

Build me something you are so far beyond my understanding of what and how you are doing what you doing. How do you learn this stuff? Did you just pick it up, are there books on it, do you have a background in electronics? Just curious. So much on this board is above my level of understanding.
 
Founder of XSA-Labs
Joined 2012
Paid Member
Nice work BMS! You really went all out and data over + /- 90 deg. Most people do 90 and reflect same data. You can actually take average of both sides and plot reflected data. I see the dipole pattern and the effect of the U cutout impacting the OB bass extension of the TC9FD. I was able to get fairly flat response down to about 300 Hz before but now it varies all through 1 kHz or more with angle. What does it look like at 1/12 octave smoothing? I think 80 dB is fine. If you simply cover the U channel with tape does the bass extension improve dramatically?

I never made polar measurements of the Finonacci baffle OB with the TC9FD and wonder how it would fare against the U cutout in polar uniformity?
http://www.diyaudio.com/forums/full-range/251310-open-baffle-nautilus-fibonacci-solution-edge-diffraction-3.html#post3841588

403491d1393936045-open-baffle-nautilus-fibonacci-solution-edge-diffraction-fibonacci-ob.png


403492d1393936045-open-baffle-nautilus-fibonacci-solution-edge-diffraction-fibonacci-ob-freq-resp-meas.png


But for reference, here is the polar response of TC9FD in sealed spiral TL with a 5 in wide baffle:

402191d1393333563-3-4-driver-very-good-dispersion-high-xmax-nautaloss-ii-2-drivers-polar-1m.png
 
Last edited:
I'm just picking this stuff up as I go along. The smart people do the maths and create the software, I'm just happy doing the donkey work and learning by doing :)

X, good observations & questions... Think I still have the U-shape cut out & can use Blu-tac [like putty in the gaps] to fill the gap again :) Nearly time for work, so I'll run a few charts off at 1/12 smoothing tonight.
 
2.5 ms as a gate is very short. Everything below 1 kHz will be softened quite much. From what I see I would stop the window at 6.5 ms. Just in front of that next reflection, which can be seen in all IR measurements. Generally try to keep the window 4 ms or larger:

REW polar.gif

And I would turn to 1/6 oct smoothing - always having the diagram a bit more precise than what the ear probably hears. ;)

Your polar plot already looks good. :) You may restrict the dB area to 50-90 dB. I find 80 dB as a measurement level sufficient if you don't live in a very noisy environment. Is the Vifa set off to one side on the baffle?
 
Rudolf, no problem, can adjust the gating to just before that fist reflection and re-plot. Used 0 smoothing for the polar data, but the gated response was smoothed out as a result of the 2.5ms point...

Will run through the changes tonight and post the results :)


The Vifa driver is centred in a 20" wide OB, with a 2" U-shaped cut-out around a 14cm x 21cm inner baffle. Background noise in my apartment normally measures as -70dB to -75dB in REW.
 
Last edited:
The driver height is 34", with the baffle sat on the turn-table, so that wont be it... All that's within 12" of the driver is the baffle itself... which is only made from 6mm MDF which I had spare. A reflection off the baffle, which gets stronger as the angle gets more acute, perhaps?

I did note something similar with the 4FE35, but there was no gap between the Impulse and reflection. The impulse just became more distorted as the angle reaches 90° off-axis.

Just about to start sorting the data.
 
Last edited:
By comparison, here's the 0° to 90° of the 4FE35 in a very similar baffle... just no 2" cut out... gated just in-front of the first reflection and 1/12 smoothing.

See if things improve as the width of the cut-out increases :)

Going to do a LX521 style trial baffle for the A7 soon. In the sealed box I got some distortion peaks at 600Hz and 1.5KHz, adding stuffing tamed these but distortion increasing between 5K & 7K.
 

Attachments

  • fr 0 to 90.jpg
    fr 0 to 90.jpg
    226.6 KB · Views: 93
Opps... forgot to add the updated polar plot for the Vifa...

Only thing I can think off for the slight variance in +/- data is down to room furniture. Also double checked the IR of the 4FE35 and there is the same Initial impulse followed by increasing distortion at 2.5ms... Caused by the baffle edge maybe?
 

Attachments

  • polar plot 2_inch cut.jpg
    polar plot 2_inch cut.jpg
    229.5 KB · Views: 403
Last edited:
Hallo BMS,

is this still your measurement situation:

410618d1396720642-cheap-fast-ob-literally-mmars_6.jpg


The couch is much too near to speaker and microfone imho. Your response graphs in post #308 suggest that you've got a problem there. If you can't change that you may measure to one side of the speaker only. Just look at your IR graphs, which side has the later onset of reflections - a wider reflection free zone.
 
Here's the 0° to 90° and 0° to -90° plots from REW, 6.5ms right gate and 1/12 smoothing...

REW polar 2.gif

It "features" all bad habits of a (too) wide open baffle: Lots of dipole notches (red circles - there must be even more) and a "midrange", where the directivity widens to only about 6 dB difference between 0° and 90° (red arrow). Compare again to the Linkwitz diagram with b=8a. :eek:
BMS,
Those polars are kind of ugly :) I think that cutout idea is not my cup of tea.
Those polars have nothing to do with the cutout. They are ugly because of the large baffle. They do indicate that the cutout does not work as wanted in this stage.
BMS,
Thanks for the measurements - great stuff there! The polars up to 4khz show very good dipole shape.
No, they don't show it! Look how the yellow 2000 Hz curve is only about 6 dB down at 80°. You can call it OB, but it's not a good dipole. At least not compared to the standards that can be achieved with a true 2 or 3-way system.
 
Got to say the results look nicer if you don't gate the measurements...

I changed ends of the room this morning so the speaker is sited roughly where the mic was, and the mic where the speaker was... Still not perfect situation, as there's still room furniture that can't be moved.

Saying that, for two sets of measures done this morning, the IR looks a lot cleaner. With a more defined first reflection at ~6.3ms... +/- 80° & 90° start to suffer from an earlier reflection off a kitchen side unit, however the initial impulse looks less distorted than with the first set of results posted.

Set one: 20" wide baffle, no cut out.
Set two: 20" wide baffle with 2" U-shape cut [again]

Furthermore, increased the cut out to 2.5" ready for the next set of measures.

Again... just about to start pulling the data across to the spread sheet.
 
Last edited:
Ahhh... the 0° to 90° are good, the 0° to -90° not so good, as before... the couch & kitchen side unit strike once more...

With the data I have and going forward, I'll just measure 0° to 90° and mirror the data to the negative side...

I'll run through as much as possible this week and over the weekend. Then post an update early next week, so it's not given in drips and drabs :)
 
Thanks for clarifying why it looks ugly and that a true dipole needs to go to zero at 90 deg. I did not realize a wide baffle causes so many problems. I made a measurement in a 24 in x 36 in baffle once and do not recall all these nasty dips.

A while ago in either this or another OB/dipole thread, I stated that a wide OB has absolutely nothing to do with the speaker being a dipole and did not get a single rebuttal. Theoretically, a dipole has zero width and depth. A practical dipole would be simply a bare driver with as little baffle as necessary to hold the "speaker" together. Any additional width will start to screw up the dipole figure-8.

But, a very narrow OB has no bass. So we widen the baffle to get the bass down to where it can be conveniently passed to a sub. Of course, the ultimate end of this process is a true infinite baffle -- a speaker mounted in a wall with an empty room behind it.

Beside the fact that I didn't like the sound, I went down the OB road for a small stretch. Quite small OB tops over H-baffle subs. I am back to boxed speakers -- MLTL's or BR's over sealed subs.

To each his own.

Bob
 
A practical dipole would be simply a bare driver with as little baffle as necessary to hold the "speaker" together. Any additional width will start to screw up the dipole figure-8.
This holds true if you ask for a "full range" dipole with a single driver. And since this is the Full Range part of the forum ... :scratch1:

Which leaves the question how much screwing would be allowed before you really have screwed it up. ;)

Rudolf