Geddes on Waveguides

My "on-listening-axis" curve is flat. The power response is falling.

Thats the classic response of a piston source.

Yes, CD speakers do tend to sound bright compared to this is the listening axis is flat. I generally roll off the top end of my personal system a dB or two because of this.

Its obvious why this is - there is simply no type of sound source with this characteristic in nature in a real room. We are used to the HFs rolling off in the reverberant field because of absorption. In small rooms this doesn't happen as much and so a nearly flat power response is deemed as "bright".
 
Thats the classic response of a piston source.

Yes, CD speakers do tend to sound bright compared to this is the listening axis is flat. I generally roll off the top end of my personal system a dB or two because of this.

Its obvious why this is - there is simply no type of sound source with this characteristic in nature in a real room. We are used to the HFs rolling off in the reverberant field because of absorption. In small rooms this doesn't happen as much and so a nearly flat power response is deemed as "bright".

Yes, the natural stuff. I like to get flat first arraival, the rest may be rolled off (if controlled). Just my opinion after doing this some years :). My room is about 40 sqm but I don´t know if that´s small by Your standards.

BTW, my HF is a "waveguide" from Italy.
 
Thats the classic response of a piston source.

Yes, CD speakers do tend to sound bright compared to this is the listening axis is flat. I generally roll off the top end of my personal system a dB or two because of this.

Its obvious why this is - there is simply no type of sound source with this characteristic in nature in a real room. We are used to the HFs rolling off in the reverberant field because of absorption. In small rooms this doesn't happen as much and so a nearly flat power response is deemed as "bright".
I think it's necessary to look at the CSD. Flat tends to sound right to me. If brightness is due to the room, then probably the best solution is to treat the room. I am very curious though, brightness caused by the room seldom happens unless a large portion of surfaces are metal, glass, or some resonating object in the room.
 
CSD!? HUUM seems you have suggested that before:rolleyes:

What you "hear" when you sit in a room is a combination of both the direct and reflceted sound. As far as spectral balance goes its more heavily weighted towards the reflected sound or "power response" (imaging is more strongly affected by the direct sound for comparison). So it isn't "the room" per see, it's the way the room deals with the power response of the speaker. All of my speakers have a great deal of HF power - far more than is typical, because of the CD nature of the design. This is why they can appear "bright" if set to "flat" on axis. I don't see CSD having anything to do with it.
 
Inherent in the GedLee argument, is sitting far from the speakers. This increases the power of reflections and brings about the ills that his patented approach rectifies.

Parenthetically, ESL listeners typically put their speakers about 1/3 down the room and hence far closer to mid-area seats. This gained room liveliness without introducing echo-dominance that Earl rightly criticizes.

Sitting a distance from speakers was essential long ago when base noise levels were intrusive especially from over-powered amps and/or pre-amp outputs insufficiently throttled at the amp input.

Likewise, for home theater environments, you can't ethically set up for a single-person listening area. So again, the GedLee rediscovery of toe-in is helpful whereas in a single-person up-closer room it would not serve the same social purpose.

My system does some upper range beaming and it beams right to me and in the power spectrum I like. Sound from the walls (which are very irregular surfaces) slightly adds to the sound in all kinds of wonderful ways (some before and some after the magic 20 ms interval) and it is all softer and following receiving my very direct beamed sound.

CD combined with toe-in can help a lot in some cases. In fact CD doesn't make a lot of sense without toe-in. But for a single-listener in a properly set-up room, not so much.
 
Last edited:
Some things that I've found in listening tests:
1. If bright sound occurs, and the image depth is lost such that the sound seems to come from the plane of the speakers, and in some occation seem to shift to one speaker only, then normally something within 2K~6KHz can be found causing this. If the sound is a bit on the harsh side, then something above 10K~25KHz is causing this.

2. If bright sound occurs, and the sound seems closer to me than the speakers, then it could be either a specific humb in the SPL or strong early reflection.

3. If bright sound occurs, and the sound seems to come from all around, I have only heard this in a room with more than 30% glass walls(two glass walls facing each other). Sounds really bad due to the high frequency modes.
 
But for a single-listener in a properly set-up room, not so much.

While most of what you say has some merit, the idea of design for a single location, IMO, does not. But yes, as one gets closer to the speakers, the direct field begins to dominate and the polar response and room matter much less. But this kind of "into the recording" imaging does not please me, single listener or not. So the fact is that I don't sit close to the speakers because I don't like the effect (it's kind of "headphoney") - albeit some do.

But let's not forget that near field listening requires small speakers and small speakers have dynamic problems (thermally related). And then there is the near field itself - not at all a stable environment.

I've been through all these "alternative" views and setups. I didn't come by my approach by chance. It was well thought out, investigated, implimented and refined and all I can say is that everyone who hears it is impressed. I'm not saying that "I" like it - I'm saying that everyone else likes it.
 
Inherent in the GedLee argument, is sitting far from the speakers. This increases the power of reflections and brings about the ills that his patented approach rectifies.

I wondered when someone would pick-up on this. ;)

..it's worse than that actually. (..or rather it *can* be.)

It doesn't just create a problem and then solve it, rather it creates a problem and then potentially compounds it. (..or perhaps is even likely to compound it.)

The problem is one of cross-reflections, often near in time to direct sound.

In other words your right speaker (referenced to you) is aimed at your left side wall. Your left speaker is aimed at your right side wall.

So you have a first reflection bouncing off of your left wall from your right speaker, and the other first reflection bouncing off of your right wall from your left speaker.

This isn't really a problem IF most of each reflection's energy is reflecting from a position a fair bit BEHIND you, where it can "reverberate" out and marginally give the impression of the recreation of envelopment. (..though note that you need a fair bit of room behind you for this to occur, otherwise you can get some pretty strange effects.)

Chances are though that each listener is NOT in a position for this to occur, because again - they are further away from each loudspeaker. In fact just the geometry of the room, (in relation to the angle of the loudspeaker's radiation under this setup) is pretty limiting on where you can sit and still be "in front" of the dominate first reflections. Worse, because of the distance this cross-reflected sound could be arriving closer in time to the direct sound than other setups (..particularly those loudspeakers with nulls at their sides that are well-positioned).

Psychoacoustically this potential new problem tends to create a "limit" on horizontal "spread" or re-created venue expansion and particularly source localization within that venue.. (..i.e it's a higher freq. problem where intensity dominates.) . In fact, it can be a bit odd to hear performers "boxed" in your room (..or sometimes out of it in the depth plane and only marginally expanded horizontally), and yet sense a massive space well beyond the confines of your room because of the information presented with low and extremely low freq.s (..at very low spls). I would note however that this is absolutely ideal for front L/R channels for movies.. needing the *restriction* to match the motion on the "screen".
 
Last edited:
It was one of the Truth 203X - other than that I don't know.

Nice attention to low diffraction for a commercial product.

I should mention that in a recent blind listening test it beat out the Orion. The data supports that result.

Dr. Geddes thank you for putting a low cost solution in the measurements program.

Is there going to be some sort of protocol to allow other speaker measurements to be added?
 
If anyone in the Portland area wants to hear some Summas and a couple of tapped horns, I may be doing a demo on Saturday. Send me an email if you're interested.
Aren't you going to tell them which Portland it is? I almost ended up in the wrong city once. Good thing I looked at the departure and arrival times carefully.:D
 
Last edited:
Dr. Geddes thank you for putting a low cost solution in the measurements program.

Is there going to be some sort of protocol to allow other speaker measurements to be added?

What I have not decided is if I am going to make this program free, charge for it, or keep it proprietary. So I have to hold to the most restrictive of those for the time being and that is proprietary. I have offered to add any data to this list that someone sends me and I will honor this up to the point where the list just gets unreasonably long and then I will reserve the right to cull the list to those designs that are the most important.

What I definately want to show is what influence the different design choices have on the critical aspect of speaker design - the polar response. I'm lacking any real three way and four way designs so I'd be very interested in those. I have quite a few smaller two ways now.

So send me the data and I'll add it to the list, but the data has got to be in a very particular format. This format is almost automatic in Holm (hoping for now that the dropped data point issue is not predominate). PM me seperately if you would like the exact format and would like to enter a data set. Basically you have to have "clean" impulse responses (at least 5 ms. of reflection free data) every 7.5 degrees (+- can be accomodated, but every 15 degerees is not good enough) out to 90 degrees, 180 is better. Export from HOLM as a single .txt impulse response file in 32 bit floating point - NO sample numbers. I need pre impulse data - about 300 - 500 points and the total data should be 8500 points. SO the text file needs to be comma delimited with all angles on one line for every sample point.
 
Last edited:
Hello Dr. Geddes,

When I run your program om my pc (Win XP pro SP3 Eng) I get incorrect polar plots (see attached screenshots).

I did some experimentation, and found the problem disappeared if I changed the regional and language options from Belgium to us (see screenshot).

I believe you can force dotnet to use a specific region and language (in your case US), instead of using the system default (by setting the UICulture I believe). The bug probable originates from the different use of points and comma's in the nummerous calculations.

After I made this change you software ran perfectly. I can imagine some of the other problems reported here can also be resolved by changing these settings. Normally you should be able to simulate this bug by changing you settings to Belgium.

What I would very much like if you could make you software available (as as exe) that allows to imports measurements made with holmimpulse (I suspect you use the function "export all measurement" with somekind of naming convention for the different measurements). This would be a fantastic supplement to the holmimpulse software, and we would then be able to compare our own prototypes (and other commercial speakers) with the same tool.

Best Regards,

Wim
 

Attachments

  • polar map summpa.jpg
    polar map summpa.jpg
    309.5 KB · Views: 381
  • polar map abbey.jpg
    polar map abbey.jpg
    310.4 KB · Views: 377
  • Regional and language options.jpg
    Regional and language options.jpg
    51.7 KB · Views: 363
Thanks Wim, that makes perfect sense. The data uses "." and "," for completely different things here than there!! Now all that needs to happen is for you EUs to change! :)

But I will check out the Regional settings. Thanks

My original idea was to offer the software as a link to Holm. But now I am not so sure. Holm seems to have changed, Ask Bojessen no longer seems to be arround and the software has not updated in a long time. This is precisely why I objected to its being free. The idea of "free" software just goes against my capitalist inclinations. There has to be an incentive to keep the software alive or it will die. There is not a single piece of software that I use that is free (Ahh - except HOLM). The free stuff just doesn't work right. (A few small apps maybe, but nothing substantial.)

If HOLM doesn't become supported, then I will just go ahead and write this part of the project myself as well.

Yes, I do use the "export all curves" aspect of Holm. No special naming convention is required. You just need to be able to find the file.

Thanks again for that note on "," & "." - that is certainly the issue.
 
This is precisely why I objected to its being free. The idea of "free" software just goes against my capitalist inclinations. There has to be an incentive to keep the software alive or it will die. There is not a single piece of software that I use that is free (Ahh - except HOLM). The free stuff just doesn't work right. (A few small apps maybe, but nothing substantial.)

Earl, I highly recommend you stay away from any most any Linux distribution. Failure to do so may force you to recant. :rolleyes:

Also, I think allowing folks to send you data to publish in your app is a bad thing. You'll have even less control over its quality than you do now. I assume you're familiar with the term, "GIGO".
 
I don't understand your first point, but then since the sentences aren't syntacticly correct, I guess that's not surprising.

Its very easy to tell good data from bad data when one sends the impulse responses. There is no fooling an impulse response in this regard.

No one has to date taken me up on my offer. I think that the reason is obvious. No one takes data this way. But, to me, thats a serious problem. Many people ask "what measurements are important" - a good set of polars is a "minimum", but no one does them. Its just like the THD mantra " I know it doesn't mean anything, but they are so easy to do!"
 
Any differences must be in your perception because the data set is exactly the same in all the plots. The data is a matrix - the top plot shows rows and the side plot shows columns - exact same data in all of them. No smoothing done in one of them thats not in the others. The smoothing is 1/20th octave in frequency.

Yes, perception. But when I look at your contour map for the un-normalized Nathan the black line seems to indicates a pretty constant beam width of around 40 degrees at about -6dB. However, when I look at the polar plots at 1k, 2k, 4k and 8k what I see is not particularly uniformity with frequency. And at 8k Hz the polar response looks particularly unsettling with 3.2 dB peaking at 18 degree off axis relative to the on axis level. Normalizing the contours by the on axis response would certainly make the -6dB beam width look worse (more pinched in at some frequencies and wider at others). All of your designs seem to have a range between 5 and 10K where the polar response show depressed on axis response with some off axis peaking at about 15 to 20 degrees. Constant DI on axis, even if achieved, doesn't necessarily mean constant polar response. The prime example of this is a dipole and a cardioid. Both have the same on axis DI but the polar response is completely different.

Don't get me wrong, uniformity of the beam width is certainly better than most other designs I have seen, but I think uniformity of the polar response could be better. I am actually some what surprised that it isn't.

I'm also wondering if your Abby data might not have the same problem at the Orion data, at least on axis, since the tweeter level is depressed overall by about 6dB on average above 5k Hz. Seems strange to me. That just shouldn't sound right if true.
 
I don't understand your first point, but then since the sentences aren't syntacticly (sic) correct, I guess that's not surprising.

Good thing we don't grade on spelling errors since "syntacticly" is spelled 'syntactically' (not too surprising).

Remove the first occurrence of "any" from my sentence and it *should* make perfect sense. Anyway, my point is that open source (free) software makes a lot of sense to a lot of folks.
 
Last edited: