As said before, why not filter the 27khz away? The reply about a notch filter seems a bit much, as why bring back the information both sides of the notch. It's a cheap tweeter. Wouldn't you just use a low order crossover to hand over from this too a super tweeter (by which I mean a resistor) at some point a lot lower than 27khz. Ignoring the idea of flat response past ~16khz as the handover too silence begins.
It seems so easy? But as a metal dome, I won't get involved anyway. Yes.. I'm of that school of thought.
It seems so easy? But as a metal dome, I won't get involved anyway. Yes.. I'm of that school of thought.
The issue with tweeters that have an ultrasonic resonance isn't the resonance itself. Most audio has zero content that high anyway and only some high res stuff might do.
The issue is with the resonance itself acting as a bell.
For example, the motor produces third order harmonics as a natural process. Lets say we have an ultrasonic resonance at 30kHz. If the tweeter is asked to play a 10kHz sine wave then it will produce a third harmonic at 30kHz. As the dome resonates at 30kHz it rings like a bell and amplifies the third order harmonic. So what once would have been a -60dB distortion product, is now amplified to be a -40dB peak.
Of course we cannot hear this third order harmonic product but what we can hear are intermodulations produced from it. So we're feeding the tweeter 10kHz and 9.67kHz. It's now going to be producing two third harmonic products. One is at 30khz and the other is at 29kHz. These two tones are inaudible but will create a 1kHz difference tone. With a soft dome tweeter both distortion products will lie at -60dB and produce a difference tone accordingly in line with those figures. The hard dome will produce one tone at -60dB and one at -40dB, again with the difference tone in line with those figures.
Is this audibly significant? It obviously depends on the size of the bell mode resonance of the dome and how linear the motor is to begin with. Have low enough intrinsic levels of HD and the IMD products will be completely off the audible radar. The same is true if the resonance is less pronounced. Beryllium domes tend to be a bit better in having less intense bell mode peaks and also tend to push the resonance higher up in frequency.
The only harmonics that can be excited aren't in the audible band.
Scenario: the tweeter plays a 9kHz tone, produces a 3rd harmonic at 9x3 = 27kHz with an amplitude of -60dB*. The harmonic tone, sitting on the ~+13dB resonance of the dome now gets 'amplified'/'excited' to -47dB. Is it audible? No, 27kHz is ultrasonic and not audible at any amplitude.
Scenario 2: the tweeter plays a 4kHz tone, produces a 3rd harmonic at 4x3 = 12kHz with an amplitude of -60dB*. The harmonic tone, sitting on the flat part dome's response does not get 'amplified'. Is it audible? Probably not, -60dB is fairly far down and probably not audible.
*Some hypothetical amplitude that I came up with that is representative of a good tweeter.
Only drivers which have significant FR peaks within the audible band (below 20kHz) pose a potential problem because they bring up the amplitude of harmonic tones which lie in the audible band - it makes potentially audible tones even more audible.
Last edited:
Be has a lower-loss value than Diamond. Ironically, higher loss is generally sought to achieve a more damped response (..less "hash" in the raw freq. response). It's somewhat ironic that the lower damped diaphragm tends to present more subjective detail - perhaps owing to the fact that we hear toward the peaks of the response (and not the dips or "average").
Of course Be is also much lower in weight - generally allowing for a more efficient driver.
IMO the best quality depositing diamond or nano-diamond onto the surface (both sides) would be driver longevity (..by FAR).
(..Be is not so great in this respect.)
If it was the Nautilus series with the Kevlar FST midrange driver the mid has a honking big uncorrected peak at 3.5Khz of several dB.
This makes the speaker sound vivid and “revealing” but also ultimately fatiguing. This peak would have been easy to correct in the crossover but they chose not to.
The tweeter may not even be the guilty party.
Maybe this occurs to some degree but it is never a problem in my experience. The distortion products arising directly from the two stimulus tones will be far more severe. Ultrasonic breakup or not those will be there at the same amplitude.
I tested the Peerless 830970 2" driver which has an enormous +17dB breakup centred about 21.5kHz.
Frequency Response:
Closeup of breakup node:
The influence on harmonic distortion is clearly visible in a sweep:
Then I tried a two tone IMD test at 7.3kHz and 8kHz. Some of the intermodulation products land on the breakup node while others fall off to the side.
Resulting spectrum of this:
The only low frequency component of significance is F2-F1 = 8kHz-7.3kHz = 700Hz. The only other audible components of significance are low order IMD/HD components occurring at about twice the stimulus frequencies (2*F2-F1, 2*F1-F2 etc). These are not in any way influenced by the cone breakup node since the 830970 is flat at 14-16kHz.
So I tried again with 7.1kHz and 7.3kHz. This time the intermodulation products should land centred on the breakup node
Again, only 7.3kHz-7.1kHz = 200Hz was significant.
My conclusion is that ultrasonic breakup nodes don't matter, but I'm happy to be proven otherwise. Measurements are knowledge
Last edited:
Probably an 800 series then (802, 3 or 4).
I haven't time to go searching, but speaking from fallible memory, not only did those older models have large uncorrected breakup modes from the Kevlar cone at the upper end of the transition band / lower end of the stopband, the tweeter crossover was a single cap, which isn't likely to let the tweeter perform at its best either. No idea about these current ones, but if they follow the same format... YMMV. If that's the case, it's not how I or many of us here would do it, but I can't honestly knock them either -they're a business after all, with the primary object of making money, same as any other, and this approach seems to work well enough for them. I wish them well.
I haven't time to go searching, but speaking from fallible memory, not only did those older models have large uncorrected breakup modes from the Kevlar cone at the upper end of the transition band / lower end of the stopband, the tweeter crossover was a single cap, which isn't likely to let the tweeter perform at its best either. No idea about these current ones, but if they follow the same format... YMMV. If that's the case, it's not how I or many of us here would do it, but I can't honestly knock them either -they're a business after all, with the primary object of making money, same as any other, and this approach seems to work well enough for them. I wish them well.
Yes I've also heard that the tweeter crossover was a single cap, which is amusing to say the least on a speaker of that price.
I'd love to get my hands on a circuit diagram of the Nautilus 802 as it's a speaker I've heard many times (years ago now though) but apparently a single cap for the tweeter is a "trade secret" and I haven't been able to find a copy online.
Nice speaker in many ways, driver design and cabinet design excellent, let down by a very minimalist crossover that as you say doesn't filter the tweeter adequately, doesn't correct the breakup resonance of the mid, and if I remember correctly, doesn't have adequate baffle step correction either, leading to a relatively forward presentation that is lacking in any warmth. (Which can also contribute to listening fatigue IMHO) Definitely not a flat or neutral sounding speaker.
A complete clean sheet of paper crossover redesign could easily transform them.
Last edited:
How so? All the same distortion mechanisms are present. I've never seen anyone prove that ultrasonic peaks cause distortion of audible amplitude and frequency; cone, dome or otherwise. Fear of ultrasonic peaks seem to be based around the same expectation bias that metal diaphragm drivers impart an audible signature that can't be described by objective measurement.
The theoretical difference tone that would have been produced if the ultrasonic peak was a problem would already exist anyway on the same driver with a soft dome because 3F1-3F2 is a standard 6th order intermodulation product. It's just really far down because decent speaker driver motors have negligible distortion components above 5th order. Increasing a component that is probably on the order of -100dB by 15-20dB is not going to do much. The 2nd and 3rd order components at -50 to -60dB are still going to be the main problem.
Last edited:
B&W schematics are here: http://bwgroupsupport.com/manuals/bw-service
In fairness, the Nautilus 800 series used a 3rd order HF filter. The 800 Series Diamond however... The 800, 802, & 803 Diamond HF filter = 1 x 4.7uF Mundorf Silver-Gold Oil cap, and the HTM2 Diamond HF filter = 1 x 4.3uF Mundorf Silver-Gold Oil. But don't fret: the 804 Diamond used a 4.3uF Mundorf Siver-Gold Oil cap, and a 1ohm series padding resistor as well. The 805 Diamond & HTM4 Diamond used a 5.1uF Mudorf Silver-Gold Oil with a 1ohm series padding resistor & the option of bypassing the latter.
In fairness, the Nautilus 800 series used a 3rd order HF filter. The 800 Series Diamond however... The 800, 802, & 803 Diamond HF filter = 1 x 4.7uF Mundorf Silver-Gold Oil cap, and the HTM2 Diamond HF filter = 1 x 4.3uF Mundorf Silver-Gold Oil. But don't fret: the 804 Diamond used a 4.3uF Mundorf Siver-Gold Oil cap, and a 1ohm series padding resistor as well. The 805 Diamond & HTM4 Diamond used a 5.1uF Mudorf Silver-Gold Oil with a 1ohm series padding resistor & the option of bypassing the latter.
From a paper presented at the 110th AES Convention:My conclusion is that ultrasonic breakup nodes don't matter, but I'm happy to be proven otherwise. Measurements are knowledge
100% of participants in the double blind study heard intermodulation distortion artifacts in a tweeter producing ultrasonic signals they could not hear in isolation from audible signals .
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.