[Saurav]

Dr. Toole thinks my measurement protocol is not good enough for designing a speaker, I'm sure he wouldn't advocate designing a speaker by ear!

Agreed. Discussions like this often end up in baby-bathwater scenarios IMO, where if something isn't perfect, then it gets dismissed as being completely useless. There's a lot of middle ground between those extremes.

[DanTheMan]

Very true Saurav.  Well said.

Dan

[planet10]

'Fraction of a decibel' is resolution in the amplitude (y) axis. Dave was talking about resolution in the frequency (x) axis, even though he said 'y' in his post.

Thanx for reading what i was thinking not what i typed   I've fixed that now.

dave

[rjbond3rd]

So Dan, what have you designed and built?

[DanTheMan]

an OB 2-way and a sealed/WG 2-way--well since I've learned to measure.  The others aren't worth mentioning since I have nothing useful to provide about them.

Oh, and I've put a Silver Iris in a u-baffle, but that was done w/ little knowledge.

Dan

[rjbond3rd]

Thanks for that! How did you sim them beforehand?

[DanTheMan]

For the OB I used the edge for baffle contribution, published prams and Zaph audio for the other significant stuff.  The rest I had to measure--where I found the downfall of the design.

The sealed 2 way is an ongoing process and slow moving at that.  To get the directivity at the crossover I used this chart as a rough guide:



Then the manufacturers anechoic response overlaid on that with actual measurements of the waveguide at hand.  It works rather well.  Certainly not a Dr. Geddes caliber design, but it does justice to the basics.  Look here for more:
http://www.audiocircle.com/index.php?topic=72549.160

Dan

[rjbond3rd]

Interesting and cool that you are using the old-school techniques of the audio pioneers.  Why not just sim though? (The Edge isn't really a sim, obviously, more like a baffle diffraction calculator with no sense of the driver's response.  Still a good tool for the toolbox though.)

[DanTheMan]

Interesting and cool that you are using the old-school techniques of the audio pioneers.  Why not just sim though? (The Edge isn't really a sim, obviously, more like a baffle diffraction calculator with no sense of the driver's response.  Still a good tool for the toolbox though.)

I don't have enough tools to sim.  My XP PC only speaks Japanese and it's a PITA to use.  The edge is easy enough though and it gives you what you really need to know for OBs.  Now that I know what I'm doing with OBs, I'd do  lot of things differently.  First time around has a steep learning curve.  I did everything for flat on axis response.  How foolish I was.    You can see I've seen the err of my ways.  The whole picture needs to be accounted for.

I should use more of these: http://www.rjbaudio.com/Audiofiles/FRDtools.html

That's not you is it?

Dan

[D OB G]

Hi DanTheMan.

I agree with you in large measure.

If we measure a clean impulse response (polar responses), we can derive a lot of information.  We can predict a good deal of the performance, "sound", of the speaker.

But it seems to me that we may not have the tools, yet, to predict everything that matters (or to know what matters).  And as Geddes says, we may need to "learn" to like what "matters".

I'm not trying to be fancy here, but since we are talking about "shout", and you are a participant on the Horn Honk thread on DiyAudio,
 




it seems to me that the Dayton 10" waveguide (you mentioned using it on another thread) may have problems of some sort below 2.5 kHz- if I'm reading the measurement correctly. 
(I'm now using the 12" Dayton, and it measures best by the wavelet technique and "sounds" best, satisfies my preferences, above 3 kHz ! ).

How can the specs, only, reconcile these measurements with the physical size determinants of performance, for example?  (can they  ?).

It seems to me we need to make choices, as I think you say, even if we account for the whole picture, and even if we are set on a certain paradigm (I wouldn't choose a rising response, but then I haven't heard the B200 in an "optimum" implementation).

David

[DanTheMan]

Shout and honk I think are 2 different thing, but we don't really have definitions for either yet.

Is that response normalized?

Any comparisons to a less honky horn under identical conditions?  It would also be interesting to know the conditions and see the impulse response.  Preferably by percent.

Thanks,

Nice wavelet.

Dan

[DanTheMan]

Oh, one more think I forgot to ask: How did you get that much output that low?  I was never able to do that.

Thanks again,

Dan

[D OB G]

Well.......I've tripped myself up with "objective" measurements.

It wasn't the 10" !!

It's all centred around Post # 328 in the Horn Honk thread.  I didn't credit the graph.  It's JMMLC's CSD from panomaniac's measurements.

Yes, shout is different to honk.

The point I was clumsily trying to offer was that different parameters MAY be in conflict.

Nice set of web references by the way!

David

[DanTheMan]

no worries.

Check out this OB wavelet from MigeO:



Last I read on that thread, no conclusion had been made to what honk is.  I don't know as I haven't been keeping up.

Dan