Gary, can you normalize your plots for level so we can see them all overlayed? It looks like if you use 4" as d1, you could go off of 20*log10(d2/d1). So you would add these values to each of the curves:
0.00
7.96
12.04
16.90
22.28
25.58
27.96
29.83
Or if you wanted to get fancy after you normalized for level you could normalize all the curves to the 10' curve to look at just the difference.
0.00
7.96
12.04
16.90
22.28
25.58
27.96
29.83
Or if you wanted to get fancy after you normalized for level you could normalize all the curves to the 10' curve to look at just the difference.
Agreed on the Hom issue, the impedance curve would show if that is mechanical (resonant) while two separate path lengths could also cause what is shown. The path length difference doesn't have to be very large physically to cause a notch up that high given the wavelength.
Best,
Tom
OK, I should not be allowed to play unsupervised in the lab after a long day at work... The measurement fixture was set for impedance measurements when the distance series was taken so there was a 35 ohm resistor in series with the driver causing the lumpier than expected results.
I re-did the measurements this morning with the fixture set correctly for frequency measurements. Not going to make guarantees though... Bumbling around in the lab at 5:00am before coffee has it's risks also.
John, if you want to play around with the measurements I can send you the ARTA .pir files via email. They add up to 1.9M zipped. PM me if you want them.
Tom, yes, there is a impedance wiggle at 8K so a mechanical resonance is likely.
New distance series-
And the impedance curve-
I re-did the measurements this morning with the fixture set correctly for frequency measurements. Not going to make guarantees though... Bumbling around in the lab at 5:00am before coffee has it's risks also.
John, if you want to play around with the measurements I can send you the ARTA .pir files via email. They add up to 1.9M zipped. PM me if you want them.
Tom, yes, there is a impedance wiggle at 8K so a mechanical resonance is likely.
New distance series-
An externally hosted image should be here but it was not working when we last tested it.
And the impedance curve-
An externally hosted image should be here but it was not working when we last tested it.
Oh, I was lazy - sorry. Well, i do tweek my crossovers for better response in farfield.
If it were a pure mechanical issue, it would not change from 0 to other distances, generally something like that would spark more measurements. The little wriggles in the impedance generally could be caused by the driver internal cavities. I recall during tweaking on driver, the cavity between the pole and dust cap caused problems noted during listening tests, and would show up as a little wriggle in the impedance around 1KHz bearly noticible. After putting some vents in the voice coil former, the issue disappeared, but we had to also do an FEM analysis to look at the modes which confirmed our suspicion. Looking at the impedance plots Gary presented, I think they seem similar, best do a vacuum chamber impedance plot to compare (learned from Dr. Klippel in one of his lectures).
I have no experience with compression drivers, but this looks much more normal. What I am curious about is with a resonance at such a high frequency, why use such a large horn? If it were driven by a current source amplifier, I would understand.
Ohhh, I thought the distances you listed were moving horizontally off-axis. Sorry about that.
Hi Gary
Ok the up side is it shows up strongly in the impedance and so, if you can design the network, can be mostly / entirely fixed electrically.
A couple thoughts, the response does change when you are close to a horn or other loudspeaker and if one needs to get a measurement which can be used to scale at large distances, one must measure past the point where the response stops changing.
You can see in your horn that up close, the low end is much stronger. Some of the measurement “grass” above a few MIGHT be reduced if you average say 5 or more of the sequences (in the setup).
Also, the sequence length could be increased to be more revealing at the lower end. The yellow bar at the bottom shows where the valid measurement stops (looks like 600Hz) and so at 1200Hz etc has limited resolution. This can be adjusted with the sequence length. It probably would be good to try one like that to see if or how much the bottom end of the response changes.
For making a crossover, one would ideally want to get the raw measurements from a distance more like the listening distance than up close.
Two meters is often enough for a small speaker or close driver spacing and this doesn’t appear to be a large horn.
What are you going to use for / arrive at a crossover?
I love horns, I hope you’re having fun!
Best
Tom
Simon, your comments usually sound reasonable, but I find this one untenable. How do you substantiate such a claim? That if you turn up a 17.5 kHz signal loud enough you can detect something? That's what most people do and it is invalid. I hope you have better support for your claim.
There are two possible explanations for this, one is interaction with the amplifier, which I had also experience at 20+KHz where reducing the cone breakup significantly improved the listening experience. Another is personal hearing sensitivity, and I also experienced this when I was playing around with a hearing test app on the iPad while playing 15KHz at max volume through Apple EarPods; I could nearly hear the signal while my wife sitting a few meters away from me was annoyed by it. These are some of the fun things we discover in real life.
I also remember when I was dating a girl, and I was testing one of those ultrasonic mosquito repelling devices, for some reason I could hear it turned on, but I could stick in right up to my date's ear and she could not hear it.
Last edited:
Looks more like 2dB peak to peak to me, but that's quibbling. A sharp pointed 2dB zig zag in the treble is quite audible on some material and inaudible on other material. It all depends on how the music excites the resonance or not. On the right music it can be quite irritating.
No I'm not easy to please. I've gone out of my way to notch resonances smaller than the one shown at the same frequency. High Q is a problem - don't fall for the "high Q resonances are less audible" fallacy, its simply not true of mechanical resonances which become more audible as the Q increases.
Huh ? An amplifier issue ?
Given that you haven't seen any measurements of the system in question this is just a load of uninformed guesswork.
The resonance in question is a cavity resonance in a ribbon tweeter and does not show up in the impedance curve at all. (BL product of a ribbon is far too low and moving mass far too low to reflect the resonance in the impedance curve)
The EQ I applied to correct it was line level before the main amplifier anyway, not that it matters in this case.
No need to invent exotic interactions with the amplifier to explain things, its a mechanical cavity resonance that can be corrected with an appropriate PEQ, simple as that. And yes it was audible on the right music even though it was only a 1dB peak.
I was surprised because I listened to that peak for a number of years without thinking it to be a significant problem before trying to correct it, more out of idle curiosity than anything else, and because digital PEQ made the change so easy to test.
Only then did I realise what I thought was the occasional bad recording was actually that small resonance. Suddenly some "bad" recordings were good, and "good" recordings were still good. I sure learnt a lesson from that.
Last edited:
How do I substantiate that I can hear to 17.5Khz, is that what you're asking ? How exactly am I supposed to prove that on an Internet forum ? C'mon.
I've always been able to easily hear the 15.625Khz (PAL) line frequency from the line output transformer in a CRT TV - not just barely, but to the point that it is irritatingly loud and bothers me anywhere in the same room. Thankfully CRT TV's are almost a thing of the past.
I don't have to turn the volume up to hear 17.5Khz - I can do a sine sweep at a moderate listening level and it goes up to 17.5Khz and then completely disappears. It still sounds like a tone to me as well, up until the point where it disappears.
I can hear 17.4khz easily, I can't hear 17.6khz even if I turn the volume up another 10dB or more. Its not a gradual rolloff like a normal low pass filter, its more like a brick wall filter. I don't think being able to hear up to 17.5Khz is particularly unusual either, so I see no particular reason for you to disbelieve me.
My point is that a resonance at 15Khz is audible to me as its well within the limits of my hearing, and my tweeters do have a small resonance at 15Khz which I have also notched to audible benefit.
I'm the first to admit that I find a resonance at 15 kHz far less noticeable than one at 8khz, it is rather subtle but it is still noticeable and its absence is noticeable on particular material when it has been removed. Imaging is also noticeably affected by its presence or absence on some material. On other material notching it out or not makes no difference.
Take what you will from my comments, I offer them as is.
Last edited:
That is extraordinary, since a single semitone above 17.4 kHz is 18.43 kHz. Your hearing acuity drops off more than 10 dB within a fifth of a single semitone.
I think hearing to 17 kHz is unusual in an adult, but perhaps that is entirely a function of exposure to noise in our industrial environment. Is there anything unusual in your upbringing in terms of exposure, or lack of it, to loud sounds?
Just curious.
..and running out of retail stores with poor lighting (..that don't seem to bother anyone else).
-switching power supplies and their cap's screaming aren't a pick-nick either.
Last edited:
How do you explain the results of Toole which clearly contradict this belief?
Simon's claim would put him in the 1 in a thousand or even ten thousand or even more. Well actually it is completely without precedence that I know of. My wife is an audiologist and I have seen hundreds or thousands of audiograms. They all are dropping at 8 kHz and above. To anyone in the hearing business above 10 kHz is non-existent, which is why there are no tests that test that high.
Detecting a tone that could well have sub harmonics is not the same thing as having exceptional hearing at 17 kHz.
I will have to put it down as untenable. Certainly unprecedented. Absolutely unproven.
