SE Measurements

[davef]

Ok, so after a few months of work -- I am pleased to say that I have developed a complete computer automated setup for an entire suite of loudspeaker measurements....

What normally took me perhaps 8 hours of engineering work -- can now be completed in under 30 minutes -- and I don't even have to be in the lab

Measurements included in the suite:

*Quasi-Anechoic on-axis full range frequency response

* Horizontal dispersion (15, 30,45 degs off-axis - to right and left)
Vertical dispersion and symmetry (5, 10,15 degs off-axis - above and below)

*Listening window (an averaged response that best represent what the listener hears -- averages both on-axis and off-axis measurements)

* Impedance + Electrical Phase (also called a Bode plot)

* Waterfall CSD plot (Cumulative Spectral Decay)

I think this covers it all -- is there anything else you guys would like to see?

Also, would you prefer I post the graphs unsmoothed or 1/3 octave? Using a 5dB scale or 10dB scale? Understand that if I post measurements of our products using 1/3 octave smoothing and a 10dB scale (like many other manufacturers) -- the response plots will look *almost* like perfectly straight lines...

Personally, I like the completely unsmoothed graphs on a 5dB scale as they are extremely revealing --- however --- much of the general public might not understand this difference in scaling and smoothing and they might wrongfully assume one loudspeaker is more linear than another based on these scaling differences...

Curious as to your thoughts...

Thanks!!!

[Jonnyozero3]

Damn Dave. I need another beer to cogitate on this

Edit - Okay, my Becks Dark thinks that showing both is best. Have a smoothed graph in an obvious place for comparison to other products, but then also have the unsmoothed ones a click or two away so us nit-pickers can check them out

[P Seastrand]

Also, would you prefer I post the graphs unsmoothed or 1/3 octave? Using a 5dB scale or 10dB scale? Understand that if I post measurements of our products using 1/3 octave smoothing and a 10dB scale (like many other manufacturers) -- the response plots will look *almost* like perfectly straight lines...

Personally, I like the completely unsmoothed graphs on a 5dB scale as they are extremely revealing --- however --- much of the general public might not understand this difference in scaling and smoothing and they might wrongfully assume one loudspeaker is more linear than another based on these scaling differences...

I suggest you post both with the "industry standard" as the default for comparision purposes, with a link, or page down option to see the unsmoothed graphs for the details.

[BradJudy]

I don't care much for 1/3 octave smoothing, but I don't know that unsmoothed is right either (too much misinterpretation). Perhaps you can split the difference with 1/6 or 1/12 octave smoothing? I do like the idea of showing a graph "using competitor's standards" and another using "Ascend's high standards" or something like that. Of course, that's most effective if you can demonstrate what an unnamed competitor's speaker looks like in the two scales/resolutions.

Automating that many angles of off-axis down to 30 minutes is great!

I'm curious about the QA full range measurements (feel free to skip this question if you feel an answer would give too much away) - how low are you able to measure accurately? Full range QA would require a lot of free space in my understanding.

[davef]

I'm curious about the QA full range measurements (feel free to skip this question if you feel an answer would give too much away) - how low are you able to measure accurately? Full range QA would require a lot of free space in my understanding.

Depends on how you take the full range measurement... close mic or ground plane...

I prefer ground plane for the low end and "splicing" this into the anechoic measurement. Yes -- ground plane requires a lot of open space --- (the only advantage of running out of inventory ) but even the ground plane measurements I take are only accurate down to about 50Hz (good enough for most loudspeakers). I also use close mic techniques to verify the accuracy and to determine response below the limits of ground plane.

These measurements are definitely NOT automated -- and quite time consuming.. However, once these are taken -- everything else is a breeze and all the splicing calculations are done through macros I developed.

Of course, website upload and image tweaking in photoshop is a tedious process --- but the difficult tasks of generating the graphs and off-axis measurements (I take a total of 144 off-axis measurements) are now completely automated.

One important factor in all of this is that I still have the CBM-170 which was measured by the NRC. This allows me to closely calibrate based on these measurements. If I remember correctly, the NRC chamber is precisely accurate down to 60Hz or so.

[Quinn]

I second giving both and explaining the differences in graph resolution and how to read other companies graphs to determine what resolution they are using.

[curtis]

Mmmmm....Becks Dark...one of my favorites!

Oh....I think posting both would be a good thing as well, but like Brad posted may misinterpret the unsmoothed graph. "Other" companies sometimes don't even tell you what resolution they are using and leave a lot up to interpretation.

[davef]

I like the idea of posting both versions of the graphs -- unsmoothed and using a 5dB scale -and- 1/3 octave on a 10 dB scale. With the macros I developed, this is easy to do -- it is a bit more work with image processing and website uploads but I enjoy doing things differently -- and this is definitely a bit different

I will start with 170 SE measurements and go from there.

Great stuff!!

[Gov]

Dave are you going to include the 200's in this as well?

[debo]

When buying any product I like as much information as possible. However presentation is everything, taking it one step further with both graphs seems like a great idea. Maybe with a statement like "What others won't show you".