• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Zilch's AK Design Collaborative - Econowave Speaker

DEQ2496 is a 1/6 octave RTA, with 0.1 dB max amplitude resolution, as I recall. It's FFT, but we don't know the actual sampling algorithm. Nonetheless, smoothing is inherent in the number of "bars" it displays and the amplitude resolution, which is adjustable. As Skywave observes, it's an excellent compromise for typical hobbyist use, and more than adequate for professional applications, as well, including basic crossover and system design.

The higher the resolution and the less smoothing you apply, the more detail is displayed. That's a double-edge sword, though, with RTA; for that additional detail to be meaningful, other measurement techniques such as MLS with time-windowing or stepped and gated sinusoidal must be employed to minimize or eliminate contributions of the measurement environment such as room reflections. At high enough resolution, I can see the influence of the location of my coffee cup in the impulse response.

All of that is important with respect to characterizing and working with the response behavior of the system in isolation, quasi-anechoically. HOWEVER, bear in mind also that's not how we listen, typically. In-room frequency response, including the influence of the reflected sound field also matters; it's about compromises, balancing signal and noise. Noise matters, too.

http://www.churchsoundcheck.com/resolution2.html

What I'm saying is that the utility of high resolution depends on the measurement method and conditions as well. If we all had anechoic chambers, it'd be a slam dunk, but we don't. Key to getting the job done is understanding the limitations of each method and working within them. Study D'Appolito's Testing Loudspeakers for a more comprehensive appreciation of these factors.

Note: Most all CLIO curves I post here are shown with 1/6-octave smoothing. Check the legend to be certain. You do NOT want to try to make sense of the unsmoothed response of my typical measurements.... ;)
 
Thanks for your lucid words, Zilch. Your point about in room response is well taken.

In my situation, I'd like to be able to measure quasi-anechoically; since my room is small and I believe that's the only way to measure what the speaker is doing down to a certain frequency. If I had a back yard I'd probably drag the rig outside for testing.

Time to order Testing Loudspeakers.


Having used the DEQ2496 for hobbyist crossover work and speaker "experiments," it's been instrumental in helping me understand (to the extent that I do understand) how drivers sum, room factors, phase and crossover testing, among other things. Beyond that, I do use the EQ in my active system, mainly to help with room modes after optimizing placement. The EQ is not as bad as most say. I'm not saying it's an awesome EQ. The worst aspect is that since I use an analog system, apart from the DEQ, I must suffer with A to D/D to A/ conversions.

Good bang for the buck.
 
Zilch said:
I'm tryin' to figure an easy way to get a bit of upward tilt on the waveguide, as they do sit low, and also a CNC shop to make the legs, now that I have a drawing for those.

Another member is doing Valencias with tilted waveguides. I'm not sure what Jack's planning for his Sevilles, but they have the same issue. If anyone can come up with a clever solution, it's our own pre-eminent E'wave cab adapter Jackgiff.... :yes:
Click to expand...


The Seville cabinets may not be the best ones to work out the details of how to tilt the waveguides. The speaker baffle sits roughly two inches behind the edges of the cabinet, and the top overlaps the edges by another inch. With the baffle being three inches behind the top, any tilt will cause the top to screen out the upward projected sound, unless the waveguide is way below the top.

The pix show that with my projected drivers (LE14, and 6x12 waveguide), the top of the waveguide will only be about 1 inch or so below the curved grille former. Just using swag, it seems like the distance should be roughly 6 inches or more to be clear. What is your guess, Zilch? The 8 inch square waveguide will be even worse off.

The last pic shows the first pair of speaker cabinets that I built about 40 years ago, sitting on the top of one of the Seville cabinets. These cabinets are on their third set of drivers, and sound so good, that I couldn't stand them being so ugly from all the moves they have made over the years. The bashed corners, scarred veneer, and mistakes made using my router to remove the back panel were all patched, and they were veneered in walnut burl. Now they look as good as they sound. The walnut burl will probably be the choice when it comes time to veneer the Sevilles.
 

Attachments

  • Sevilles01.jpg
    Sevilles01.jpg
    78 KB · Views: 113
  • sevilles02.jpg
    sevilles02.jpg
    77.9 KB · Views: 350
  • sevilles03.jpg
    sevilles03.jpg
    82.8 KB · Views: 115
There's certain to be some diffraction effect at that top edge, but make a sketch to see how the geometry plays out. As Dennis suggests, it's not going to take much tilt to make a significant improvement.

You can see how rapidly the SPL falls off outside the 50° vertical dispersion window in the measurements Earl Geddes did for us. By 45° (90° vertical beamwidth,) it's down 10 dB, and the interference node Wayne has discussed here will be attenuating it even more. The node may in fact be controlling in that region, and also limit how much benefit may be achieved via differential aiming of the drivers:

attachment.php

Prototype both ways and measure. The horizontal displacement of the driver acoustic centers and the crossover phase relationships are going to be factors as well. We may get lucky. :yes:

FUN stuff, this...! :thmbsp:
 

Attachments

  • Geddes EconoWave Vertical.jpg
    Geddes EconoWave Vertical.jpg
    88.4 KB · Views: 1,029
Hi Jack, those looks sweet!!! It seems like you could make a new 1" or so motor board and glue it to the existing motore board. In that build you could angle the EWG 3 to 5 degrees and bring it out an inch. Just a thought. Cheers.
 
Anechoic measurements

Zilch said:
If we all had anechoic chambers, it'd be a slam dunk, but we don't.
Click to expand...

Actually, we do - outdoors. Anechoic chambers usually only work down to a few hundred Hertz, but outdoors is truly reflection free. In a large field, the only boundary you have is the ground.

I really like to do ground plane measurements outdoors because it lets me see what the speaker is doing without room influence. At high frequency, pseudo-anechoic is fine. It lets me see what the crossover is doing without dragging the speaker and measurement system outdoors. But to get the final measurements, I like to go outdoors in a large open reflection-free area, truly anechoic.

As I said before, I don't mean that smoothing is necessarily bad, as it shows general trends and spectral balance. That's what is most important.

However, it is not difficult to gather and view the data in finer detail using most modern measurement equipment. You really can see more than you can hear if you set the resolution high enough. For those looking for diffraction and high-order modes, I would think they would want to look at high resolution data, where they might be able to see their effects. Whether or not those effects are audible is another matter, I suppose.

Zilch said:
The higher the resolution and the less smoothing you apply, the more detail is displayed. That's a double-edge sword, though, with RTA; for that additional detail to be meaningful, other measurement techniques such as MLS with time-windowing or stepped and gated sinusoidal must be employed to minimize or eliminate contributions of the measurement environment such as room reflections. At high enough resolution, I can see the influence of the location of my coffee cup in the impulse response.
Click to expand...

Definitely, I agree. If doing measurements indoors, one must do pseudo-anechoic to separate the speaker from the room. In-room measurements are fine, useful for seeing what the end result is. But to know what the horn is doing, or how the interaction between horn and driver, it is probably best to see what's happening without room influence first. And like you said, with high enough resolution, you can see the reflections of your coffee cup in the impulse response.
 
Wayne Parham said:
Actually, we do - outdoors. Anechoic chambers usually only work down to a few hundred Hertz, but outdoors is truly reflection free. In a large field, the only boundary you have is the ground.
Click to expand...
That'd work, if I could shut down I-80. Noise floor's kinda high outside here. :D

Meyer Sound does some of that locally, actually. Maybe Ken knows some secret places that are quiet and have power available.... :yes:
 
Zilch said:
Prototype both ways and measure. The horizontal displacement of the driver acoustic centers and the crossover phase relationships are going to be factors as well. We may get lucky. :yes:

FUN stuff, this...! :thmbsp:
Click to expand...

Easy for you to say. The baffles on the Sevilles are not removable. They need some holes filled, and then new ones cut from a distance of two inches away. Not going to be an easy thing to do. I need to cut some guides to fit the cabinet, and use a long router bit with a wheeled guide. Don't want to do that twice, but I get your drift. Maybe some more plotting before the move is made.

Moving the bottom of the waveguide out from the baffles is an option, but there isn't much room since the grille curves around and the 12" width is pretty much filling the width of the baffle. I will calculate how much can be gained that way, and then maybe some tilt will be available.
 
jackgiff said:
Easy for you to say. The baffles on the Sevilles are not removable. They need some holes filled, and then new ones cut from a distance of two inches away. Not going to be an easy thing to do.
Click to expand...
Don't I recall you removing an existing baffle completely and installing cleats to accept new ones in the same location in one of your earlier retros?

Is that not a workable approach with the Sevilles?

I do see that the baffle is recessed behind the moulding at the sides for the grille retention. Maybe leave a stub of the existing baffle in place, and cleat behind that.... :dunno:
 
skywave-rider said:
Steve, Using the DEQ2496, I've never been able to see the 19kHz spike produced by the BMS drivers we have. Being a DEQ owner I assume you have the same experience. I've relied on Zilch's measurements to see that. I never thought I'd be able to see HOMs or horn resonances with that unit. Having said that, for me, it's been a great piece of gear and very useful.

Planning to move up to SoundEasy when I can, and I hope it can measure with appropriate data resolution.

Thanks, Wayne.
Click to expand...

Hi Skywave:

Just want to be clear that I was in no way disparaging the DEQ2496 or its use for hobbyist type measurements. I own one and use it although I also continue to use my Audio Control SA-3050A even though it’s only 1/3rd octave…mostly because I’m familiar with it and know its quirks.

I too cannot resolve that 19 KHz resonance that shows up in the BMS spec sheet and in Zilch’s measurements of the BMS drivers on either the 3050 or the 2496. What I find curious about this is that Zilch mentions that most of his CLIO curves have been “smoothed” or de-resolved to 1/6th octave and yet they show the resonance quite clearly. There must be something in the conversion process that preserves a degree of resolution that is never available when native resolution is lower to start. Insights anyone?
 
Steve O said:
Insights anyone?
Click to expand...
Click the link I provided at Post #1161. The TEF curves are all smoothed 1/6 octave (16.7%,) but are still displayed with 2048 data points.

It's apparent "Binning" in RTA. Not even the 1/12-octave RTA has sufficient resolution to display the detail shown in the 1/6-octave smoothed TEF curve. Clearly, binning and smoothing are different processes. I don't know what that difference actually is, though, i.e., how smoothing is performed. A running average comes to mind as one possibility which would maintain the original display resolution but smooth the curve. D'Appolito probably tells us what's actually going on, but the chapter on FFT puts me to sleep. :boring:

I'm very likely using the wrong words, but that's the concept.... :dunno:
 
:scratch2:
Zilch said:
Click the link I provided at Post #1161. The TEF curves are all smoothed 1/6 octave (16.7%,) but the curve is still displayed with 2048 data points.

It's apparent "Binning" in RTA. Not even the 1/12-octave RTA has sufficient resolution to display the detail shown in the 1/6-octave smoothed TEF curve. Clearly, binning and smoothing are different processes. I don't know what that difference actually is, though, i.e., how smoothing is performed. A running average comes to mind as one possibility which would maintain the original display resolution but smooth the curve.

I'm very likely using the wrong words, but that's the concept.... :dunno:
Click to expand...

I'm going to have to cogitate on this a while...a long while. However, it's clear that the output of a 1/6 octave RTA is NOT the same thing as 2048 samples over ~ 10 octave range "smoothed" to 1/6 octave over that same 10 octave range. By my rough math, whatever it is you use to obtain 2048 samples is the equivalent of 1/205 octave. High resolution indeed!
 
Data resolution

One other thing about binning, if you're measuring low frequencies a high sampling rate can lead to low resolution. It's an odd side effect of real time processing in some systems, something like the problem of how to deal with aliasing in video.

Measurement systems are very interesting. I've been very fortunate to be able to chat with Keith Larson a lot, he has a weath of knowledge on signal processing. If you start getting into it, contact him because he's very knowledgable and approachable. His WTPro measurement system is a great value, by the way.

I think everyone here is starting to think about the issues, and that's great. The reason I mentioned data resolution is dnewma04 pondered why some horns measure flat but sound bad. My suggestion is the data resolution may not be high enough to see the details. I have never encountered a horn that measured nicely but sounded bad - I can always see the details in greater resolution with Speaker Workshop, LMS or WTPro, beyond the level of audibility.

You can measure in high resolution and then smooth down, or you can measure in low resolution to begin with. Either way, you lose precious detail that could show small details like you would expect from diffraction or reflection. Those both make little blips that are completely lost when smoothing is applied or when resolution is low.
 
Wayne Parham said:
Measurement systems are very interesting. I've been very fortunate to be able to chat with Keith Larson a lot, he has a weath of knowledge on signal processing. If you start getting into it, contact him because he's very knowledgable and approachable. His WTPro measurement system is a great value, by the way.
Click to expand...
I sent Keith a little invitation to come help us out on this one.... :thmbsp:
 
Zilch said:
Don't I recall you removing an existing baffle completely and installing cleats to accept new ones in the same location in one of your earlier retros?

Is that not a workable approach with the Sevilles?

I do see that the baffle is recessed behind the moulding at the sides for the grille retention. Maybe leave a stub of the existing baffle in place, and cleat behind that.... :dunno:
Click to expand...

Well yes, that has been done by me in times past. The circumstances were a little different though. The construction of the front of the Sevilles leaves very little room for error. The baffle is held in with 3/4 inch strips around its periphery, and I would need at least 3/4 inches beyond that to make an acceptable baffle mounting board. That would make the width of the new baffle 13.75 inches, and the cutout for the LE14 is 12.45 inches. split the difference, and that leaves only about .65 inches of material at the sides of the new baffle. Not good odds from my perspective.

I don't want to take a chance of spoiling the Sevilles, since I have waited to get them for so long, and they look so nice. I will experiment on the Aztec cabinets that Storm gave me.
 
Well, just a quick note on where I've been with the Valencias.

I've installed the E-Wave in the cab and sealed it, but haven't done much with the porting. I did also attach a D210 to the 811B and have it laying on the cab. With bananas on the leads from the Ewave crossover tweeter outs I can swap the horns back and forth at will and frankly, I don't notice much difference between the Ewave and the 811s.
Kegger would like to see more adjustability/range on the L-pad as we have to run it almost wide open to match the "stock" Valencia with the supertweeter, so we may adjust the resistors and cap values in that leg of the circuit. As it stands, it looks like I'll be ditching the Altec 806As for the D210s and maybe even the Ewaves horns too as the 811B kill them in looks.
Well, at least in my book

More to follow. Picture below:

attachment.php
 

Attachments

  • altec91808.jpg
    altec91808.jpg
    74.8 KB · Views: 2,272
dnewma04 said:
BTW, i think Zilch just had a heartattack with your idea to stick with 811s...
Click to expand...
Naw, Andyman has demonstrated to his own satisfaction that, even with E'Wave's compensation, 806As don't get it; the Altec midrange is a fiction.

More quality listening, and 811B will reveal its inherent coloration in comparison to E'Wave, even with these inexpensive drivers. Move up to BMS and E'Wave's transparency is even more apparent.

Once the porting is dialed in, the E'Wave Val will kick the stock Val's butt.

Many would prefer to stick with 811Bs for their imaginary gonadotropic effect. In the end, however, "It's the MUSIC...." ;)

Andyman said:
Kegger would like to see more adjustability/range on the L-pad as we have to run it almost wide open to match the "stock" Valencia with the supertweeter, so we may adjust the resistors and cap values in that leg of the circuit.
Click to expand...
Jack worked out the "High-sensitivity" version of the crossover for just that purpose. Use those values.

BUT, recall also the earlier discussion here regarding Val bass response. By my calcs, you cannot "fix" it with more cabinet volume, as I believe you have already discovered empirically. Give up some of the artificial efficiency of 416Z by installing BSC on it, and not only will you have as much as 6dB more headroom on the HF, but more of the extended bass response that is already there will "magically" appear for your listening pleasure.... :yes:

attachment.php

http://audioheritage.org/vbulletin/showthread.php?p=163004#post163004
 
Zilch said:
Jack worked out the "High-sensitivity" version of the crossover for just that purpose. Use those values.
Click to expand...

I think we are; we're using the 95 dB woofer version with the 1.5uf/15/24 circuit.

BUT, recall also the earlier discussion here regarding Val bass response. By my calcs, you cannot "fix" it with more cabinet volume, as I believe you have already discovered empirically. Give up some of the artificial efficiency of 416Z by installing BSC on it, and not only will you have as much as 6dB more headroom on the HF, but the extended bass response that is already there will "magically" appear for your listening pleasure.... :yes:
Click to expand...

BSC???


Dave, FWIW, I WAS wondering if my 811B comment was functional heresy in this thread too :D
 
You must log in or register to reply here.
Back
Top Bottom