Very true. Another way to think of it: we can think of a speaker as a mass (cone, voice coil, spider, etc) on a spring (spider, surround). Newton's laws tell us the force needed to shake the mass (f=ma), while Hooke's law tells us the force needed to fight the spring (f = kx).
At very low frequencies (particularly below speaker fundamental resonance), the f=ma term is small - so most of the force the voice coil supplies is used to fight the spring (i.e. spider and surround). These springy bits are definitely not linear, and so cause lots of distortion at these low frequencies.
As you raise the frequency, you reach a point when the two forces (f=ma and f=kx) are equal in magnitude. This happens at the fundamental resonance frequency of the driver.
Now raise the frequency even more, and the spring stops being a factor - the dominant force is the f=ma bit. And this is perfectly linear - it does not cause any distortion in the speaker movement.
Tweeters are used well above their fundamental resonance frequency - usually at least a full octave above, often as much as two octaves or more. So they are inherently linear, "fast and clean", as Bentoronto said. So there is little to no benefit to trying to apply servo-feedback to them.
(We're assuming we have a good tweeter - the tricky part of tweeter design and construction is keeping the tweeter cone or dome out of mechanical breakup. It must be very light and very stiff. If breakup occurs, distortion comes with it. But servo feedback can't help with this, since different areas of the dome/cone are flapping about independently.)
The whole situation is reversed when we talk about subwoofers, which, as Ben said, are often expected to work below their resonance frequency. Now the very nonlinear spring force (f=kx) dominates, causing huge amounts of distortion. And this type of distortion can really only be effectively treated using servo feedback.
So it's not just engineering laziness that has led engineers to focus on motional feedback in woofers and subwoofers - that is where there is an actual problem, and that is where a solution is needed.
Motional feedback for tweeters, on the other hand, is pretty much a solution to a non-existent problem.
-Gnobuddy
I'm actually interested in correcting full range drivers.
I would also like to try localized mic feedback (as in the mic is right in front of where you are sitting) but even lasers can't do that. A complicated DSP phase corrector is needed for that and I don't know what the benefits might be but I'm looking into it.
I would also like to try localized mic feedback (as in the mic is right in front of where you are sitting) but even lasers can't do that. A complicated DSP phase corrector is needed for that and I don't know what the benefits might be but I'm looking into it.
Why not correct them only over the low frequency end of the range?
IMO full range drivers are a can of worms, with poor performance at both ends of the frequency spectrum due to the huge engineering compromises necessary. Woofers make very bad tweeters, and tweeters make very bad woofers, so a full range speaker is - more or less inevitably - bad everywhere in the frequency spectrum.
You cannot do feedback with long time delay. The feedback signal has to go around the loop in time to correct errors in speaker cone position as they occur. This is impossible if the feedback signal is delayed by more than a fraction of a cycle.
You can do feed-forward correction with a "complicated DSP phase corrector", like some noise-canceling headphones do. But that is not only complicated, it doesn't provide the distortion reduction that feedback does. And why bother, when you can just use plain ordinary EQ ahead of the amp?
One of the problems with full range speakers is that they tend to beam treble like a searchlight, because they are too big to provide proper treble dispersion. This means they have a different frequency response at every listening position. Even with the magic time-travel machine, which one will you correct?
IMO motional feedback lets you take a good woofer or subwoofer and make it even better. It cannot fix a bad (full range) speaker. It's the old "lipstick on a pig" situation.
-Gnobuddy
Last edited:
This is a rough block diagram of my idea of a phase corrector.
It compares the phase degree of the input signal and the feedback signal and then adjusts the phase of the feedback signal until they are equal. If this circuit is very fast it should be able to correct the phase degree error without a problem. Or am I wrong?
Last edited:
I'm thinking if you put the mic right in front of your chair the NFB from the mic will correct the full range of audio in your local area. Obviously if the mic is meters away from the speaker you are going to run into phase degree issues in the feedback. Thus a phase corrector is used to prevent oscillation. I'm hoping that it will give a full range improvement in your "local" area (where you are sitting) in the same way a motional feedback device mounted on a driver gives localized feedback on the diaphragm.
But I guess now that I think about it more the distance is probably too far for the feedback to work, even with phase correction. Probably a dumb idea.
But there must be some benefit right?
Last edited:
But wait.....What if instead of an active feedback using the mic/feedback corrector idea, you use a prerecorded feedback?
For example if you set up your room and sit in a designated spot, then set up the mic in front of you, then play a song and digitally record the audio on the mic.
Then you save that "setting" for your next playback and align the recorded feedback to match the original recording so they are in phase. Tadah localized acoustic feedback Yes? No?
For example if you set up your room and sit in a designated spot, then set up the mic in front of you, then play a song and digitally record the audio on the mic.
Then you save that "setting" for your next playback and align the recorded feedback to match the original recording so they are in phase. Tadah localized acoustic feedback
Yes? No?
Its not just a matter of phase - if the speaker has already made bad noises earlier, you can't correct them later!
By the time the sound from your speaker has arrived at your microphone, the nasty noises have already been made. It's too late to correct them.
This is why you cannot have a delay equivalent to several cycles of speaker motion, and still have negative feedback.
Any time delay has to be restricted to a small fraction of one cycle of the highest frequency that you want feedback to affect.
To full range drivers? The only benefit is that you don't have to design, pay for, and construct a crossover network. This makes it simple for DIY speaker builders to get extremely crappy sound.
Full-range speakers are cheap, and good enough for some jobs: paging / voice announcements, background music in businesses, electric guitars (which don't go below 80 Hz or above 5 kHz anyway.)
But for full-range Hi-Fi audio? Full range is really inadequate for that, unless you're happy with 1940-era speaker sound quality.
New-fangled "full range" speakers are usually 2" or smaller in diameter. They are terribly inefficient, and can only play at very low SPL, but frequency response and sound quality are less awful than ye olde 1950's full range speaker (big thin flappy cone 10" or 12" in diameter, maybe a thin flappy "whizzer cone" glued to the middle of that.)
Designing a good passive crossover network is actually a huge challenge, and very few DIY speaker builders have the knowledge, facilities, or equipment needed to do a good job. So I can see why people try to do an end-run around the problem and just buy a mediocre-sounding full range speaker.
But these days, one can neatly side-step the passive crossover network design problem by using active crossover networks, and powering each driver (woofer, tweeter, etc) with its own amplifier. This type of active crossover network is much, much easier to design.
-Gnobuddy
Would be putting lipstick on a pig. Still ugly as heck, and I wouldn't want to kiss it.
And it won't work to actually improve the sound, because of all the cone breakup and treble beaming issues we've just discussed. MFB can't fix any of that.
Short version: you cannot start with a lousy speaker (such as a fullrange driver) and make it a good one with MFB.
The best you can do is start with an already good speaker, and make it a little bit better at the low frequency end of its range.
-Gnobuddy