Look at the graph. The -20 happen on top, beaming because of the diaphragm width.
Here are the plots of a very popular CD horn for compression drivers. It also increases the directivity a bit towards 2 kHz. It has more CD on top, yes.
XT1086 - Constant Coverage HF Horn
Last edited:
My labeling my diagrams is not going to change the way other people do things.
Real acoustic measurements are never very smooth, hence they are virtually always smoothed in some manner. How much smoothing has been applied is usually obvious from how narrow the resonances are. There are always resonances and they always vary in bandwidth. When there aren't any narrow ones, as in those plots, then a lot of smoothing is going on.
And having used 18 sound stuff before, I know that they are a little more than loose with there specs.
Real acoustic measurements are never very smooth, hence they are virtually always smoothed in some manner. How much smoothing has been applied is usually obvious from how narrow the resonances are. There are always resonances and they always vary in bandwidth. When there aren't any narrow ones, as in those plots, then a lot of smoothing is going on.
And having used 18 sound stuff before, I know that they are a little more than loose with there specs.
But it might increase your credibility.
I'd like to know what smoothing is applied and what the resolution is. I always try to make sure that my diagrams are labeled correctly. Otherwise data might get misinterpreted. Obviously a lot of manufacturers like people speculating.
Dear Mr. Geddes,
May I enquire as to how much smoothing is applied to the Diagrams you show, out of interest? I am somewhat unbelieving that the curves you show are the raw data, but I'm happy to stand corrected.
Ciao T
May I enquire as to how much smoothing is applied to the Diagrams you show, out of interest? I am somewhat unbelieving that the curves you show are the raw data, but I'm happy to stand corrected.
Ciao T
Graaf, back to the room modes:
Measures of detecting room modes that are subjectively destructive are:
Listen to electronic music which is guaranteed free of resonance bodies.
Take a function generator with fine adjustment and with real knobs.
Calculate the wavelength and see where it appears in the room.
Walk around to confirm.
I found that there is only one narrow, subjectively bad mode, the one between the side walls along the wall where the speakers stand.
I will build a triangular reflector box with the height and depth of the bass modules (45 cm) and will see what happens.
I find physical measures are always better than electronic ones. Even if you flatten the frequency response the slowed-down decay is still present.
Get an old function generator for 10€ at Ebay for subjective testing and a simple soundcard-based measurement equipment for adjusting your parametric, and stay away from hightech room correctors.
Measures of detecting room modes that are subjectively destructive are:
Listen to electronic music which is guaranteed free of resonance bodies.
Take a function generator with fine adjustment and with real knobs.
Calculate the wavelength and see where it appears in the room.
Walk around to confirm.
I found that there is only one narrow, subjectively bad mode, the one between the side walls along the wall where the speakers stand.
I will build a triangular reflector box with the height and depth of the bass modules (45 cm) and will see what happens.
I find physical measures are always better than electronic ones. Even if you flatten the frequency response the slowed-down decay is still present.
Get an old function generator for 10€ at Ebay for subjective testing and a simple soundcard-based measurement equipment for adjusting your parametric, and stay away from hightech room correctors.
Attachments
That is a waste of time and money. Get multiple subwoofers and adjust them for a flat frequency response at the listening position using a high resolution real time analyzer. I use a real time FFT software and white noise. This allows me to adjust everything in real time and saves hours or days of tweaking.
No, it's not. If you correct a modal peak with a EQ in the frequency domain, the time domain will be corrected at the same time. Toole shows some examples in his book. You might also want to take a look at http://www.casakustik.de/forum/index.php/topic,309.0.html (German).
Some "hightech room correctors" seem to work quite well: Audio Musings by Sean Olive: The Subjective and Objective Evaluation of Room Correction Products
OIf course my data has some smoothing applied ALL data has some smoothing applied, even if its raw FFT data.
My curves are smoothed at about 1/10 octave and the dB resolution (another smoothing technique) is about 1/10 dB. The data above is at least 1/3 octave smoothed but judging from the data it looks like a litle more. If I get a chance I'll post my data smoothed to 1/3 octave for comparison.
Dear Mr. Geddes,
Yes, I realise that. The problem is that without knowing how much smoothing was applied to each plot it is hard to compare.
With enough smoothing applied even a Lowther driver starts to look flat(ish)...
This is quite high resolution for published plots. I applaud your honesty in showing them. And they do look very good!
Please do. It will be at the very least a good illustration to many just how much a little judicius smoothing alters the picture.
Ciao T
Yes, I realise that. The problem is that without knowing how much smoothing was applied to each plot it is hard to compare.
With enough smoothing applied even a Lowther driver starts to look flat(ish)...
This is quite high resolution for published plots. I applaud your honesty in showing them. And they do look very good!
Please do. It will be at the very least a good illustration to many just how much a little judicius smoothing alters the picture.
Ciao T
Here's the same frequency responses with different smoothings applied.
No smoothing:
No smoothing:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
As I said, with a little experioence you can guess whats been done as far as smoothing goes. You will notice on my plots that the resonances are all over the place, narrow, wider, etc. You can also see that in Markus data. If you don't see these "warts" then the data has been massaged.
Its quite akin to photographing models. It is my understanding that we do not ever see unretouched photos of our favorite models. Those girls never look that good because its not real. Its quite difficult to tell how much retouching has been done.
If you look at my polar paper you will see some real data for "other" loudspeakers polar responses. Thats what they should look like is fairly shown.
Dear Mr. Geddes,
Indeed. However, many here probably lack that.
Also, sometimes one mis-estimates (I thought you had applied more smoothing TBH).
Ciao T
Indeed. However, many here probably lack that.
Also, sometimes one mis-estimates (I thought you had applied more smoothing TBH).
Ciao T
Yes, his plots tend to look very very smooth. Pretty impressive if there's that little smoothing applied!
I thought as it probably uses the TI/TacT amplifier chips it might also use the TacT/Lyngdorf room correction, but maybe this is a too free association. Maybe it does just a simple EQing, one would have to ask.
Don't know what you're referring to. I linked a thread that clearly shows that correcting a modal peak in the frequency domain corrects the time domain too. A simple parametric EQ was used.
Stimulus (sine burst) and measurement:
An externally hosted image should be here but it was not working when we last tested it.
Stimulus (EQed sine burst) and measurement:
An externally hosted image should be here but it was not working when we last tested it.
Sorry, I thought your "?" was referring to the Sony.
Of course one could do parametric EQing with exactly the same Q as the resonance as done in the discussion over there. But I don't think any of the self-calibrating room correctors does something like this. Do they work with convolution/extremely long FIR filters?