How to Design Loudspeakers without Performing Measurements

[fntn]

Re: How to Design Loudspeakers without Performing Measurements

How do you obtain accurate offset data without "measurements"? :D

Maybe a ban on crossovers higher than 1500hz would make that whole issue go away.

[ripcard]

Re: How to Design Loudspeakers without Performing Measurements

Way to go Dave! Too bad we didn't make the time at the DIY meet for you to present this. Maybe next one?

[[email protected]]

Re: How to Design Loudspeakers without Performing Measurements

Added to The Speaker Building Bible. Thanks for the time and effort that went into this!

[gowa]

Re: How to Design Loudspeakers without Performing Measurements

Thanks again Dave.

[DDF]

Re: How to Design Loudspeakers without Performing Measurements

Hi folks,
The typo was corrected and Charlie has uploaded the update to his site.

Dave

[DDF]

Re: How to Design Loudspeakers without Performing Measurements

Hi all,
FYI, a small typo was found in one of the slides:

Was:
6.5” to 7” driver (24mm) = 0.017ms; 5” driver ~ 0.061ms (21mm)

Should be
6.5” to 7” driver (24mm) = 0.07ms; 5” driver ~ 0.061ms (21mm)

I'll send Charlie the update for the download later today

[gowa]

Re: How to Design Loudspeakers without Performing Measurements

Thanks Dave.

[DDF]

Re: How to Design Loudspeakers without Performing Measurements

I urge everyone to determine offsets as best as possible, get the frequency response looking good on axis, and then explore a few off axis angles within a +/- 30 degree window to make sure there are no nasty surprises in there. Maybe this is what people do, and maybe not, I am not sure.

-Charlie

Absolutely right Charlie. I didn't get into that level of design detail in the presentation as it was meant as a tool training aid, not a design tutorial. We could fill many pages herein discussing that topic.

FWIW, in design, I put most importance on the direct axis response, followed by the response at the angle that corresponds to the first side wall reflection, followed by the angle that corresponds to the floor bounce, followed by a general +/- 45 deg "wellness" view. Dips are relatively OK, peaks, not so much.

One other thing i need to stress is that it's important to do a good job on the diffraction modeling and splicing. Changes in <150Hz level relative to 1 kHz of even 1 dB has a profound effect on the tonal balance. Thin and brittle results from too soft a mid bass.

Dave

[charlielaub]

Re: How to Design Loudspeakers without Performing Measurements

It's really not the same either way. You can't simulate off-axis lobing in PCD without the x,y,z offsets being entered, so you give up a powerful tool in seeing the response family over a wider window.

If you mean also entering the driver position on baffle (x,y,z), then I 100% agree. I didn't get into that. Personally, I don't use PCD that way.

If one is using manufacturer supplied measurements and impedance curves modeled from MFG T/S parameters, I think DDF's suggestions about offset are valid because the level of approximation about the phase and amplitude data for the drivers is already high.

But for those who do a full blown design, I'd like to chime in on the point that Jeff made, so please read on...

It is a mistake to optimize a speaker based solely on the on-axis response, although that is a good top priority. This is because some topologies can result in severe off-axis lobing (nulls or peaks in the response) in the crossover region while the on-axis response remains mostly flat. This is especially true for low order (e.g. 2nd order) crossovers, which have more driver FR overlap. Off-axis anomalies will color the sound. Off-axis energy is reflected back to the listener from the room and perceived as part of the over all tonal balance of the speaker, even when listening on-axis.

Jeff's point is that, in order to model accurate on and off-axis responses in PCD, you need accurate data on the offsets. Without good offset info, you can't utilize PCD to its fullest potential.

I urge everyone to determine offsets as best as possible, get the frequency response looking good on axis, and then explore a few off axis angles within a +/- 30 degree window to make sure there are no nasty surprises in there. Maybe this is what people do, and maybe not, I am not sure.

-Charlie

[DDF]

Re: How to Design Loudspeakers without Performing Measurements

It's really not the same either way. You can't simulate off-axis lobing in PCD without the x,y,z offsets being entered, so you give up a powerful tool in seeing the response family over a wider window.

Don't get me wrong, this is my only qualm. I think your tutorial is very well done, and thank you for creating it.

Jeff

Jeff,
dlr's response clarified what you were getting at and please see my response to him. I could always add this in a future edit.

[Jeff B.]

Re: How to Design Loudspeakers without Performing Measurements

Jeff,
I think this is one of those tomato, toe-matt-oh cases, either way works. I recommended the same offsets in the slides (slide 13), but personally feel its better if the actual frd file contains the delay offset. This replicates what you would get from real measured files.

The user can then still dial in additional delay via the PCD to see the lobe move over various vertical angles.

Dave

It's really not the same either way. You can't simulate off-axis lobing in PCD without the x,y,z offsets being entered, so you give up a powerful tool in seeing the response family over a wider window.

Don't get me wrong, this is my only qualm. I think your tutorial is very well done, and thank you for creating it.

Jeff

[rogoll]

Re: How to Design Loudspeakers without Performing Measurements

Just a suggestion, but what I did was download and extract the zipped files, and then, in the same sub-directory, I "right-clicked" on the link to the tutorial/PDF file and selected the "save target as" to save the PDF along with all the sample files. This way you have the tutorial locally on your hard disk.

[DDF]

Re: How to Design Loudspeakers without Performing Measurements

That was very nice of you to do that, Dave! I hope it enlightens a few more budding designers!

Thanks Paul, that's my goal!

[Paul Carmody]

Re: How to Design Loudspeakers without Performing Measurements

That was very nice of you to do that, Dave! I hope it enlightens a few more budding designers!

[DDF]

Re: How to Design Loudspeakers without Performing Measurements

I don't know how much difference it makes in typical cases, but I agree with Jeff because I believe it's the more accurate model if one is going to examine the off-axis.

dlr

If you mean also entering the driver position on baffle (x,y,z), then I 100% agree. I didn't get into that. Personally, I don't use PCD that way. I could always add it to a future version of the presentation, but no time in the near future to do so.

I think the presentation offers alot of value as it stands.