If you want to make an error correction signal proportional to displacement, yes, you have to double integrate. That doesn't get rid of the signal, it just gives it a second order slope in the frequency domain. Effectively, it limits the usefulness of accelerometer servoing to low frequencies.
But can you set the integrator poles high enough to not filter the feedback signal too much while still having the f^2 correction?
Also, doesn't the integrator have to be compensated for, or the system will be unstable?
I just recived my accelerometers from AD!
That reminds me; the guys at AD didn't compensate for the f^2 factor. Is that the reason for their mediocre results?
/Marcus
Also, doesn't the integrator have to be compensated for, or the system will be unstable?
I just recived my accelerometers from AD!
That reminds me; the guys at AD didn't compensate for the f^2 factor. Is that the reason for their mediocre results?
/Marcus
No, you want to set the integrator poles pretty low, lower than your desired cutoff. It reduces the feedback at higher frequencies, but that's where you have lower displacement (lower distortion) anyway. I think a lot of this is discussed in the Velodyne patents- you might spend a few minutes at the uspto site and download/read them.
The AD guys (it was actually a college project, not something done by AD staff) did not use a double integrator. Nor proper phase adjustments. It looked like it was a hurry-up job for a summer semester credit. We hobbyists can be a bit more patient and thorough.
How much do those AD accelerometers cost? I lucked out, have a drinking buddy who's a big shot analog guru there.
The AD guys (it was actually a college project, not something done by AD staff) did not use a double integrator. Nor proper phase adjustments. It looked like it was a hurry-up job for a summer semester credit. We hobbyists can be a bit more patient and thorough.
How much do those AD accelerometers cost? I lucked out, have a drinking buddy who's a big shot analog guru there.
I know this is an old topic, but is a constantly repeatable topic so... kick!
A few replies back there was a theory why a capacitor mic could not be used.
But my theory is that most recording are done with capacitor like microphones. and we audiofreaks want to reproduce this!
The ideal situation is as following:
There is a sound named "A" if we record the sound with a cap mic (most cases) we get the speed of the air! *1
The recording we now have is signal "B" (differentiated signal of "A")
So we buy the recording in the store (cd or some thing like it) and it contains a differentiated recording!
So we want to reproduce the signal "A" and not signal "B"
And most speakers do that because the acceleration is at highest when the voltage is at peak. But as the frequency gets lower the speaker will sooner get to the ideal point when the resistance of the speaker hanging is equal to the voltage that is offered. So the integrating stops at that point!
This is the same reason why you need a very high Xmax for a Servo Subwoofer! Because the feedbackcircuit won't accept the speaker from stopping when there's still air to be moved!
(Example) so now the mic comes in when the sound "A" was recorded at some point the speed of the air was very high. To reproduce the same sound the speed of sound must be at the same level. So we place a capmic in the speaker and we want the speed of the air exactly the same as it was during the recording!
This way you should reproduce exactly the same sound!
In short:
1 Sound music
2 recorded Sound (differentiated by the mic)
3 sound out of the speaker (integrated by the speaker/feetbackmic)
4 real music!
*1 i tested the with a mic, a constant airspeed gives a dc offset
A few replies back there was a theory why a capacitor mic could not be used.
But my theory is that most recording are done with capacitor like microphones. and we audiofreaks want to reproduce this!
The ideal situation is as following:
There is a sound named "A" if we record the sound with a cap mic (most cases) we get the speed of the air! *1
The recording we now have is signal "B" (differentiated signal of "A")
So we buy the recording in the store (cd or some thing like it) and it contains a differentiated recording!
So we want to reproduce the signal "A" and not signal "B"
And most speakers do that because the acceleration is at highest when the voltage is at peak. But as the frequency gets lower the speaker will sooner get to the ideal point when the resistance of the speaker hanging is equal to the voltage that is offered. So the integrating stops at that point!
This is the same reason why you need a very high Xmax for a Servo Subwoofer! Because the feedbackcircuit won't accept the speaker from stopping when there's still air to be moved!
(Example) so now the mic comes in when the sound "A" was recorded at some point the speed of the air was very high. To reproduce the same sound the speed of sound must be at the same level. So we place a capmic in the speaker and we want the speed of the air exactly the same as it was during the recording!
This way you should reproduce exactly the same sound!
In short:
1 Sound music
2 recorded Sound (differentiated by the mic)
3 sound out of the speaker (integrated by the speaker/feetbackmic)
4 real music!
*1 i tested the with a mic, a constant airspeed gives a dc offset
If you integrate a sine function two times you get a... sine function with reversed polarity.
So you take a piezo tweeter, an impedance buffer amp (OP) close to it. Mount the tweeter to the element, near the voice coil, and as stiff connection as possible.
The signal is filtered, feedback adjusted. And thats it!
I will put up some schematics one day...
So you take a piezo tweeter, an impedance buffer amp (OP) close to it. Mount the tweeter to the element, near the voice coil, and as stiff connection as possible.
The signal is filtered, feedback adjusted. And thats it!
I will put up some schematics one day...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.