Hi all,
It's my first post and I'm asking for help already!
I'm doing my final year project on motional feedback loudspeakers and as an alternative / comparitive to the system I'm building (based on Jeff Macauleys work) I'd like to mention an optical alternative to an accelerometer.
Is there such a thing? I've seen mention of it on one other thread here but I can't find anything anywhere else on the subject?
Anyone got any ideas or links to ideas? Its not the concept I need help with but it would be interesting to know of any existing examples. I'd be very grateful for any help!
Cheers
Tim
It's my first post and I'm asking for help already!
I'm doing my final year project on motional feedback loudspeakers and as an alternative / comparitive to the system I'm building (based on Jeff Macauleys work) I'd like to mention an optical alternative to an accelerometer.
Is there such a thing? I've seen mention of it on one other thread here but I can't find anything anywhere else on the subject?
Anyone got any ideas or links to ideas? Its not the concept I need help with but it would be interesting to know of any existing examples. I'd be very grateful for any help!
Cheers
Tim
eclipse said:Hi all,
It's my first post and I'm asking for help already!
I'm doing my final year project on motional feedback loudspeakers and as an alternative / comparitive to the system I'm building (based on Jeff Macauleys work) I'd like to mention an optical alternative to an accelerometer.
Is there such a thing? I've seen mention of it on one other thread here but I can't find anything anywhere else on the subject?
Anyone got any ideas or links to ideas? Its not the concept I need help with but it would be interesting to know of any existing examples. I'd be very grateful for any help!
Cheers
Tim
I can predict one problem that you might run into. The sound from a driver is proportional to cone acceleration, not cone position. So, in your feedback loop you will have to take the second derivative of your transducer signal and regulate on this rather than the transducer signal itself. This will amplify the noise from your transducer, and probably you will have to limit the useful frequency range of the system because of this. This is also why accelerometers usually are preferred (at least I would guess so).
Hello Eclipse,
I saw test of a T&A loudspeaker in a german hifi magazin several years ago. They used an optical position sensor in the woofer.
Infact it looked like a moving piece of plastic connected to the membrane. The end of the plastic had a triangular shape.
This "moving triangle" was situated between a optical sensitive resistor and a light source. The whole sensor was placed on the backside of the speaker magnet. The moving plasitc piece was connected to the membrane through the hollow pole piece.
Because I'm just working on a speaker feedback projekt using the voltage of a secondary voice coil I'm very interrested in all available material concerning speaker feedback. You mentioned the work of Jeff Macauley. Are copies of his work available in the web?
Greetings from Germany
Zelter
PS All native speakers - please excuse my English
I saw test of a T&A loudspeaker in a german hifi magazin several years ago. They used an optical position sensor in the woofer.
Infact it looked like a moving piece of plastic connected to the membrane. The end of the plastic had a triangular shape.
This "moving triangle" was situated between a optical sensitive resistor and a light source. The whole sensor was placed on the backside of the speaker magnet. The moving plasitc piece was connected to the membrane through the hollow pole piece.
Because I'm just working on a speaker feedback projekt using the voltage of a secondary voice coil I'm very interrested in all available material concerning speaker feedback. You mentioned the work of Jeff Macauley. Are copies of his work available in the web?
Greetings from Germany
Zelter
PS All native speakers - please excuse my English
You don't need to make it so complicated. The absolute position of the cone isn't relevant, per se. If you use any form of sensor on the motion of the cone, you'll end up with something proportionate to the signal. Just put that in as negative feedback, whether it comes from a mic, optical read, second voice coil, etc.
Optical sensing of the cone is a little more tricky than it appears at first, although it has its good points. Incidentally, the second voice coil strategy is convenient, but you're also going to get crosstalk from the driven voice coil so the feedback signal isn't as clean as you'd like.
Grey
Optical sensing of the cone is a little more tricky than it appears at first, although it has its good points. Incidentally, the second voice coil strategy is convenient, but you're also going to get crosstalk from the driven voice coil so the feedback signal isn't as clean as you'd like.
Grey
Re: Re: Help required regarding optical motional feedback theories
You're right that the sound from the driver is proportional to the 2nd derivative of the cone position. However, every signal can be considered as an infinite sum of sinusoids, and we all know that the 2nd derivative of sin(x) is simply -sin(x) [noting the negative term]. This implies that you can simply invert the position signal and you are given acceleration (out by a scaling factor).
As for the position sensor itself ... I wouldn't imagine that you'd be using this system on tweeters, rather large woofers with decent excursion levels ... thus I'd suggest a thin plastic rod with a small white dot on the end. A CCD (of sort) reading the position of the white dot, and inverting it, would be your feedback.
Svante said:
You're right that the sound from the driver is proportional to the 2nd derivative of the cone position. However, every signal can be considered as an infinite sum of sinusoids, and we all know that the 2nd derivative of sin(x) is simply -sin(x) [noting the negative term]. This implies that you can simply invert the position signal and you are given acceleration (out by a scaling factor).
As for the position sensor itself ... I wouldn't imagine that you'd be using this system on tweeters, rather large woofers with decent excursion levels ... thus I'd suggest a thin plastic rod with a small white dot on the end. A CCD (of sort) reading the position of the white dot, and inverting it, would be your feedback.
Re: Re: Re: Help required regarding optical motional feedback theories
Yes, for a single frequency, that would work. If several frequencies are involved, the amplitude relations between them need to be maintained, and to understand how this relation is affected by a double differentiation we need to take the 2nd derivative of sin(wt) would be -w^2 * sin (wt).
So, in order to acheive a flat response, you'd need a double differentiation of the position signal, otherwise your regulator would make the cone *amplitude* constant (independent of frequency), and the sound pressure would go as w^2 or +12 dB/octave. A simple sign shift wouldn't do.
substrate said:
Yes, for a single frequency, that would work. If several frequencies are involved, the amplitude relations between them need to be maintained, and to understand how this relation is affected by a double differentiation we need to take the 2nd derivative of sin(wt) would be -w^2 * sin (wt).
So, in order to acheive a flat response, you'd need a double differentiation of the position signal, otherwise your regulator would make the cone *amplitude* constant (independent of frequency), and the sound pressure would go as w^2 or +12 dB/octave. A simple sign shift wouldn't do.
I am currenly studying my second year of Electronic Engineering at York university and I was considering this very idea for my final year project.
I had found a company that makes laser-based optical distance measuring sensors, although the sampling frequency of these sensors is limited therefore the frequency at which you could apply this type of system is also limited. The error is also non-linear as they use a system whereby a mirror tracks the laser dot on the object they are measuring and the relative distance is based on the angle of the mirror.
Without looking into the matter further it would appear that using an accelerometer would be the most sensible thing to do, unless of course you intend on pioneering overcoming the problems that come with using lasers?
Best of luck with the project.
I had found a company that makes laser-based optical distance measuring sensors, although the sampling frequency of these sensors is limited therefore the frequency at which you could apply this type of system is also limited. The error is also non-linear as they use a system whereby a mirror tracks the laser dot on the object they are measuring and the relative distance is based on the angle of the mirror.
Without looking into the matter further it would appear that using an accelerometer would be the most sensible thing to do, unless of course you intend on pioneering overcoming the problems that come with using lasers?
Best of luck with the project.
What problems were you thinking of?annex666 said:
I can't think of a single one involving the use of a laser to measure cone velocity (which you could sample at any rate you wanted, from which you could obviously then get the acceleration and position)
Bowers and Wilkins have used this technique before, measuring cone velocity through laser interferometry, so there's no pioneering involved.
In response to my post about measuring the doppler shift:
As a result of this, if you bounce photons off a moving target that is coming towards you, they bunch up, which is the fundamental basis of a doppler shift. Likewise, if the target moves away, they spread out.
This delta-lambda is measurable by interferometry and provides all the data necessary; position (inferred from the direction of motion), velocity (the shift itself) and acceleration (the difference between two velocity samples)
I really don't understand what you are trying to say Are you trying to imply that an invariance in c means that no doppler shift can be measured? I wonder how pulse-doppler radar works then
More importantly, I know that this method has been used, by B&W, in speaker development. There's no point reinventing the wheel.
Yes the magnitude of the velocity in free space remains constant.454Casull said:
As a result of this, if you bounce photons off a moving target that is coming towards you, they bunch up, which is the fundamental basis of a doppler shift. Likewise, if the target moves away, they spread out.
This delta-lambda is measurable by interferometry and provides all the data necessary; position (inferred from the direction of motion), velocity (the shift itself) and acceleration (the difference between two velocity samples)
I really don't understand what you are trying to say
Are you trying to imply that an invariance in c means that no doppler shift can be measured? I wonder how pulse-doppler radar works then More importantly, I know that this method has been used, by B&W, in speaker development. There's no point reinventing the wheel.
Hi Tim,
If you want to go on with the optical approach look for so called "PSD’s", that stands for Position Sensitive Device. Sharp have some ready-made devices available incorporating all the optics for an affordable price. But yes you will need to tackle the noise problem from the double differentiation needed. This will be not an easy task.
When using a secondary voice coil you are measuring voice coil speed and you still need single differentiating. And you have to tackle the cross talk problem.
Best is to measure acceleration directly. Piezo’s are pretty good at this. You can find suitable piezo’s in those “Music post cards” and in piezo tweeters from Motorola.
Cheers
If you want to go on with the optical approach look for so called "PSD’s", that stands for Position Sensitive Device. Sharp have some ready-made devices available incorporating all the optics for an affordable price. But yes you will need to tackle the noise problem from the double differentiation needed. This will be not an easy task.
When using a secondary voice coil you are measuring voice coil speed and you still need single differentiating. And you have to tackle the cross talk problem.
Best is to measure acceleration directly. Piezo’s are pretty good at this. You can find suitable piezo’s in those “Music post cards” and in piezo tweeters from Motorola.
Cheers
Mudge said:
Sorry if the previous post was unclear - the problems I was refering to were with regards to the commercially available distance measurement sensors. There are of course many other ways in which to measure the distance although a commercially available unit would make things a lot simpler.
Also with regards to the sampling rate I am sure this is finite with the measurement system I refered to - the mirror system that has to track the dot and calculate distance by angle of the mirror cannot move instantly therefore sampling must be discete in nature and limited in frequency (the manufacturer themselves quote a maximum sampling frequency - not me!).
One possible problem would be the non-linear error of the commercial units (as I previously stated), due to the mirror mechanism.
Another post on the thread suggested using a sensor that was linear in nature - tracking a dot with a CCD. This seems like a good alternative (infact one that I had also considered) and should be researched more.
Hi,
Found some reading on how PSD’s work:
http://www.mcselec.com/download/appnotes/gp2d02.pdf
PSD’s have the advantage that they work completely analog and as such have infinite resolution, this in contrast to line CCD’s. Also they can be very fast. But if they are suitable for LS motional feedback ???
PSD’s chips are made by various companies. I.e. have a look here:
http://www.udt.com/positionsensitivedetectors.htm
Cheers
Found some reading on how PSD’s work:
http://www.mcselec.com/download/appnotes/gp2d02.pdf
PSD’s have the advantage that they work completely analog and as such have infinite resolution, this in contrast to line CCD’s. Also they can be very fast. But if they are suitable for LS motional feedback ???
PSD’s chips are made by various companies. I.e. have a look here:
http://www.udt.com/positionsensitivedetectors.htm
Cheers
Er the speed of light remains constant......... This is true whatever the situation the light is placed in.
You fire a light beam from a car moving at 100mph and the light will move at 674M mph not 674M+100 mph. Light speed always remains the same regardless of what speed the source of the light it moving at.
You fire a light beam from a car moving at 100mph and the light will move at 674M mph not 674M+100 mph. Light speed always remains the same regardless of what speed the source of the light it moving at.
Sadly the speed of a photon is fixed, but the propagation of a beam of light is determined by the refractive index.5th element said:
Alternatively though, if you have an optical material where the refractive index varies rapidly (as a function of frequency), and you bounce a laser off the dustcap/phaseplug and then catch the reflected beam with a prism made of said material, it will bend the beam of light up and down as the it doppler shifts
Alternatively, with an interferometer you look for changes in the interference pattern when the original beam is combined with the reflected beam, since they are no longer at the same frequency.
The proper way to do this optically is to set it up with a monochromatic source (ie, a laser diode) and an interferometer. In this case, you're just counting fringes and since it's digital, you can massage the data any way you like- there's no added noise from the two differentiations, for example. You would probably use a double beam, with one being in quadrature to the other so that you know when the cone is moving forward and when it's moving backward.
I think it would make a dandy senior project.
I think it would make a dandy senior project.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.