In another thread Guss make this post:
"With the woofers coil moving forth and back all the time is it really worth wasting some time to align the drivers to perfection???"
I thought this was a super question and it got me to thinking that the woofer's motion, particularly in a 2 way, is going to "modulate" the lobing pattern and frequency response of the system.
Whether this is audible is another question.
Anyone care to discuss this modultation and it's potential audibility?
"With the woofers coil moving forth and back all the time is it really worth wasting some time to align the drivers to perfection???"
I thought this was a super question and it got me to thinking that the woofer's motion, particularly in a 2 way, is going to "modulate" the lobing pattern and frequency response of the system.
Whether this is audible is another question.
Anyone care to discuss this modultation and it's potential audibility?
Bill Fitzpatrick said:
To get an idea of the magnitude of the problem, let's assume that the cone moves back and forth 10 mm by the low frequencies, and that drivers are separated by 100 mm, and that the main lobe is straight forward, the angle of would vary arctan(10/100)=+/- 5 degrees. Assuming a frequency of 2 kHz (the crossover frequency where both drivers are of equal magnitude) the polar diagram at 5 degrees of two sources separated by 100mm has dropped by cos(pi*f/c*x) or 0.15dB. (x=cone displacement, c=sound vel., f=freq)
So, straight ahead it would result in an amplitude modulation of
+/- 0.08 dB.
Audible? Hardly. At least not using the assumptions above and that I did the math right
.
sreten said:It doesn't occur therefore its not audible.
Its the same arguement than for bass / mids, bass causes
doppler distortion of the midrange, which it doesn't.
And if accepted full range drivers wouldn't have a prayer.
A non-issue,
Are you saying that there is *no* doppler effect, or that it is way too small to be audible?
OK, let's assume Svante got the math right on the lobing; I'm tending to think he's right.
Now, how about frequency response?
If you have response curve X with the woofer mounted on the baffle, you would have curve Y with the woofer mounted forward by, say, 5mm and curve Z with the woofer mounted rearward by the same amount.
So, with the woofer mounted flush and moving back and forth 10mm as a result of some low frequency signal, the response is continuously sweeping from Y to Z. True - Yes. Audible - ?
I thought Doppler might arise - so who says Doppler is not a problem? Klipsch thought it was.
Now, how about frequency response?
If you have response curve X with the woofer mounted on the baffle, you would have curve Y with the woofer mounted forward by, say, 5mm and curve Z with the woofer mounted rearward by the same amount.
So, with the woofer mounted flush and moving back and forth 10mm as a result of some low frequency signal, the response is continuously sweeping from Y to Z. True - Yes. Audible - ?
I thought Doppler might arise - so who says Doppler is not a problem? Klipsch thought it was.
Bill Fitzpatrick said:
The +/-0.08 dB I calculated was the frequency response change at 2kHz. AFAIU the maximium effect should be seen at the crossover frequency, so for the speaker with 2kHz crossover frequency, the effect should be of that magnitude, and no more.
It would be worse at higher crossover frequencies, but IMO not very important anyway.
Let's say that at 100Hz the speaker's cone moves +-3mm, then the velocity of the cone at halfway will be about plus or minus 1.9m/s. Presumably this would modulate a 2kHz tone by +-11Hz, but I think it's just an over-simplification and doesn't really work that way. Essentially the result would still be 2 separate tones superimposed on each other. I wouldn't worry about slight variations in the constructive/destructive interference patterns, where will you be listening to this - suspended in an anechoic chamber?
Why must I explain that the sound coming straight from the speakers is only a small fraction of what we actually hear, doesn't even go in a straight line, and that simplified lobing calculations are pointless? What about room nodes and anti-nodes at low frequencies, or at any frequency for that matter? These are areas of oscillating air pressure and areas of oscillating air velocity. They will interact with other sounds, causing them to refract and shift slightly. Although it's most noticeable when trying to talk to someone in a strong wind, we shouldn't overlook the possibility of it completely ruining our listening pleasure during a little Mozart.
When considering only the midwoofer's cone, what I would be worried about is harmonic distortion at low frequencies. It will mean that a 100Hz tone will cause intermodulation distortion with a 2kHz tone, and thus create new frequencies such as 2.1kHz and 1.9kHz.
As for the room, there may be a low frequency room-node positioned in the path of high frequency on-axis sounds. The oscillating air pressure at the node will create a lens effect that may repeatedly amplify and attenuate the high frequency sounds, thus also causing IM distortion.
While I'm at it, a lot of the bass sounds that supposedly vanish because of baffle-step actually get reflected by walls and such-like, and the lowered dB measurements are at least partly compensated by increased in-room resonances and delayed bass. This is why I'm confused by theorizing that bass levels should be increased with a step filter. What about high frequencies, shouldn't their volume be increased instead, to compensate for reduced off-axis levels? What about both?: A high-pass step filter to compensate for the reduced off-axis dB levels at high frequencies, and a low-pass filter to compensate for reduced on-axis dB levels at low frequencies. I suggest that instead of making 2 mutually cancelling filters, it would be much more accurate to take measurements in the listening room in several positions, and use filters to flatten the measured frequency responses.
CM
Why must I explain that the sound coming straight from the speakers is only a small fraction of what we actually hear, doesn't even go in a straight line, and that simplified lobing calculations are pointless? What about room nodes and anti-nodes at low frequencies, or at any frequency for that matter? These are areas of oscillating air pressure and areas of oscillating air velocity. They will interact with other sounds, causing them to refract and shift slightly. Although it's most noticeable when trying to talk to someone in a strong wind, we shouldn't overlook the possibility of it completely ruining our listening pleasure during a little Mozart
.When considering only the midwoofer's cone, what I would be worried about is harmonic distortion at low frequencies. It will mean that a 100Hz tone will cause intermodulation distortion with a 2kHz tone, and thus create new frequencies such as 2.1kHz and 1.9kHz.
As for the room, there may be a low frequency room-node positioned in the path of high frequency on-axis sounds. The oscillating air pressure at the node will create a lens effect that may repeatedly amplify and attenuate the high frequency sounds, thus also causing IM distortion.
While I'm at it, a lot of the bass sounds that supposedly vanish because of baffle-step actually get reflected by walls and such-like, and the lowered dB measurements are at least partly compensated by increased in-room resonances and delayed bass. This is why I'm confused by theorizing that bass levels should be increased with a step filter. What about high frequencies, shouldn't their volume be increased instead, to compensate for reduced off-axis levels? What about both?: A high-pass step filter to compensate for the reduced off-axis dB levels at high frequencies, and a low-pass filter to compensate for reduced on-axis dB levels at low frequencies. I suggest that instead of making 2 mutually cancelling filters, it would be much more accurate to take measurements in the listening room in several positions, and use filters to flatten the measured frequency responses.
CM
sreten said:
None, at all, thank god, as all our hi-fi's would sound appalling,
Could you elaborate on this a bit, since it is a bit of an unusual standpoint. Or if there is something written somewhere about this, would you have a link?
Just to get an idea of the order of magnitude of the way I think it works I took a typical 6.5" Peerless element, with Xmax=7 mm, maximum cone velocity occurs at 125 Hz and is 4 m/s RMS, or 5.6 m/s peak. So let's assume worst case of this 125Hz tone and 2 kHz tone, this is something that may occur in real life. Now, in my mind, this would mean a frequency for the 2 kHz tone of 2000*345/(345-5.6)=2033 Hz when the cone moves at maximum speed towards me, and 2000*345/(345+5.6)=1968 Hz when it is moving away from me. This is based on the (reasonable) assumption that the velocity from the 2 kHz tone in small.
Are you saying that this frequency modulation of the 2 kHz tone does not happen?
Note that I don't claim anything about if it is audible or not, though. Hmm, this would be easy to simulate and then listen to it... Hmm...
I'm sorry that I can't post evidence that it doesn't occur,
but I defy anyone to post test evidence that it does occur.
You do get intermodation distortion between the tones,
the sidebands of which do change pitch with excursion
level but this is entirely different to doppler distortion.
The physical analogy people are using is say one driver
with another lightweight driver attached, and one driver
modulating the others output frequency which it would.
However for a perfect distortionless single driver the two
signals are simply voltage summed and produced by the
driver.
The apparent position of the higher frequency does not
change one iota, even at practically ludicrous excursions.
sreten.
but I defy anyone to post test evidence that it does occur.
You do get intermodation distortion between the tones,
the sidebands of which do change pitch with excursion
level but this is entirely different to doppler distortion.
The physical analogy people are using is say one driver
with another lightweight driver attached, and one driver
modulating the others output frequency which it would.
However for a perfect distortionless single driver the two
signals are simply voltage summed and produced by the
driver.
The apparent position of the higher frequency does not
change one iota, even at practically ludicrous excursions.
sreten.
Member
Joined 2003
You guys got me curious, so I adjusted tweeter offset 5mm in LspCAD. On and off axis frequency response changed about 1db around crossover (3200), group delay changed a tad, phase changed...impulse response changed quite a bit. This is static offset, but I thought interesting enough to share.
4-way dipoles
4-way dipoles
Of course doppler distortion occurs and it's unscientific to consider otherwise.
How audible it is is another question.
Read John K's paper. There is other stuff out there for those willing to make the effort.
http://www.geocities.com/kreskovs/Doppler1.html
How audible it is is another question.
Read John K's paper. There is other stuff out there for those willing to make the effort.
http://www.geocities.com/kreskovs/Doppler1.html
Maybe a bit off topic.......
But there was a speaker company in Colorado that made a speaker in the late 80s-early 90s, with a Motorola piezo mounted on a 8" woofer.
Yep, it moved with the woofer.
Voices sounded great on it.
But when it got moving with some tunes playing..............
Yeah, Doppler............tons of it.
(I had a pair on evaluation, so I am not pulling anyone's leg.)
Jocko
But there was a speaker company in Colorado that made a speaker in the late 80s-early 90s, with a Motorola piezo mounted on a 8" woofer.
Yep, it moved with the woofer.
Voices sounded great on it.
But when it got moving with some tunes playing..............
Yeah, Doppler............tons of it.
(I had a pair on evaluation, so I am not pulling anyone's leg.)
Jocko
What a fluke that the so-called Doppler distortion frequencies just happened to be exactly 100Hz above and below the 1kHz test tone. I strongly suggest that it was actually intermodulation distortion like I suggested earlier. Otherwise the 900Hz and 1.1kHz tones would change in frequency when the amplitude of the 100Hz tone changes.
Even if the cone moves at half the speed of sound at 100Hz, it still simply causes compression and rarifaction of the air, so that a 1kHz tone superimposed on it will not change frequency even if there is a large physical offset in distance. Several metres away, the movement of the air might be only a fraction of a millimetre, and yet is it still suggested that a 1kHz will be offset by the same large distance?
CM
Even if the cone moves at half the speed of sound at 100Hz, it still simply causes compression and rarifaction of the air, so that a 1kHz tone superimposed on it will not change frequency even if there is a large physical offset in distance. Several metres away, the movement of the air might be only a fraction of a millimetre, and yet is it still suggested that a 1kHz will be offset by the same large distance?
CM
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.