the idea is to put a couple of sensors in my system, excursion metering (and motor temperature metering). i want to know whats going on in my subwoofers. has anyone experience with this? links to diy metering projects much appreciated.
Interesting project.
Temperature should be relatively easy. Measure resistance cold, than let 'r rip, then measure resistance hot.
There's probably a nice equation on line to get delta-T from delta-R assuming copper wire.
Excursion is harder.
What comes to mind is an IR distance measuring module with an analog output so you can look at it on a 'scope.
The pro's would use a laser, but that's $$$.
You also could make a contact between the cone and a probe on the outside, move the probe until the contact is just open, than measure the distance.
Jan
Temperature should be relatively easy. Measure resistance cold, than let 'r rip, then measure resistance hot.
There's probably a nice equation on line to get delta-T from delta-R assuming copper wire.
Excursion is harder.
What comes to mind is an IR distance measuring module with an analog output so you can look at it on a 'scope.
The pro's would use a laser, but that's $$$.
You also could make a contact between the cone and a probe on the outside, move the probe until the contact is just open, than measure the distance.
Jan
Perhaps only indirectly related but you may want to look at Philips motional feedback speakers developed in the '70's.
I'm sure you can find schematics.
Hugo
I'm sure you can find schematics.
Hugo
Here's the link to a more complete wiki page than the English one:
https://nl.wikipedia.org/wiki/Motional_Feedback
Hugo
https://nl.wikipedia.org/wiki/Motional_Feedback
Hugo
Jan, sorry but this is not quite so easy in practice. It's easy to measure the driver Re when there is no signal present. Now try doing that when you are pumping lots of low frequency signal into it. You have to filter, filter, filter with a very low cutoff frequency. This will give you a decent reading but will cause both a "time averaging" effect and a lag or delay (it's an inherent property of such a filter). The lower the HP filter cutoff the longer the delay and averaging effects but the smoother the output signal. I looked into this once using TI-TINA when I had a project where I wanted to temperature correct a feedback circuit around a driver.
There are some crude ways to measure the excursion. Siegfried Linkwitz used a triangular piece of paper glued to the edge of the cone, for example. You have to tilt your head and visually guesstimate, but it is a way to do it. LED sensors and accelerometers are other ways, but none of them is "simple" if you are not familiar with this sort of technology. If you want something approximate, the idea of directing an LED onto a mirror glued to the edge of the cone and then watching the reflected image (on the wall or ceiling) move will work the SL method, but has the advantage that you can make the image larger. But you will not get quantitative info. That is actually pretty difficult.
Also, good luck trying to use motional feedback circuitry to obtain cone position!
If you want a practical example of how to do Re measurements you can take a look at this article doing just that, published in of all places Stereophile:
https://www.stereophile.com/reference/1106hot/index.html
The author used a 6th order filter with 1 Hz cutoff frequency, and there was still some bleedthrough of the music signal.
https://www.stereophile.com/reference/1106hot/index.html
The author used a 6th order filter with 1 Hz cutoff frequency, and there was still some bleedthrough of the music signal.
Klippel measures driver's temperature through the change of the voice coil impedance using a very low frequency (<10Hz as far as I remember) while playing the stimulus. This is part of their "Power Test".
I've done some simillar testing on transistors to find bias changes with temperature.
I measured with a sample & hold at xover, and at the same time threw a solid state relay to disconnect the load to give the S&H enough time.
No signal, and fast enough to catch the actual temp.
But yeah, not something you do between lunch and dinner!
Jan
I measured with a sample & hold at xover, and at the same time threw a solid state relay to disconnect the load to give the S&H enough time.
No signal, and fast enough to catch the actual temp.
But yeah, not something you do between lunch and dinner!
Jan
Last edited:
I have been working on a IR distance sensor for a while.
Just a bit lack of free time and some other things unfortunately. 🙁
The biggest issues are that they are very sensitive for temperature changes as well as not having a linear output (excursion vs voltage). Both can be compensated for.
Lasers aren't THAT expensive anymore these days.
Around $1000-1500 should be fine.
I don't agree with that the temperature is the "easy" part.
Most of the discussed techniques here are to slow to get a good sense of how the temperature act with time.
Especially with smaller voice coils (like tweeters), these can sometimes jump up pretty fast (faster than like once a second or so).
Additional problem here is not just resolution, but also the averaging effect when using a lower resolution.
Depending on the kind of woofer, but 10Hz is definitely NOT low enough to determine the Re very well!
I like your zero-crossing detector idea though.
I am gonna give that some more thought! 🙂
Btw, for just measurements, things can be performed a lot more (quasi) static.
Doing multiple things on the fly is a very nice thing to have, but not always necessary.
A simple way to measure excursion is to glue a small magnet to the cone, together with an analogue hall-effect sensor, like the WSH49E. However, since magnetic field is inversely proportional to the square of the distance, the input-output relationship would be non-linear, and it maybe better to obtain an interferometer meant for a laser turntable.
It also adds mass.
Not a problem for a big (sub)woofer, but just not as handy for smaller or lighter speakers.
Also the residu of the glue isn't so nice.
Plus like you mentioned, the non-linear output.
I have tried that in the past, but it's really a pain to compensate well.
It seems we need 'cube' in place of 'square'.
Did you use a microcontroller / LUT ? What curve did you try fitting (video) ? A possible means to (more) linear output would be to use only a small quasi-linear portion of the curve, as also shown towards the end of the video, which I think is quite interesting, considering the simplicity of the method.
Last edited:
No, just measured straight into the computer and did all the compensation there.
But back than we quickly dropped the idea because sticking/gluing magnets on things was just not very practical.
You also needed to be very aware of the value of the Mms and be careful not to change it to significantly.
What part are you referring too?
Because I don't see any real linearity shown? Just some initial graphs, but that doesn't say much at all.
1:32 onwards.
Quasi-linear, as I said before, since the magnet is not taken too close to the sensor in the second test, thereby using only a small portion the curve ...
Right, so you kinda approach it as linear.
In that case I would still compensate it, but it all depends on the resolution and accuracy.
Yes, of course, there needs to be compensation if perfect linearity is the criterion. Resolution is an issue only when A/D conversion is done 'as is', without using filtering or amplification. With an amplifier, I'm quite sure that a regular (hobby grade) sensor could be used and there'd be no need to get the (possibly expensive) brand mentioned in the video.
Do you need real-time physical monitoring?
Otherwise you could measure both parameters once, model a simplified equivalent and derive the data from the input signal.
(Edit: you would need to monitor surrounding temperature and air pressure for more exact results, of course!)
Last edited:
For just "knowing what's going on", one could also use poormans Klippel method.
Which is measuring the impedance with a constant voltage source (very important, so NOT the resistor attenuation method) and keep an eye how the frequency shifts of the impedance peak, as well as how smaller this impedance peaks gets, as function of output voltage.
The shift in frequency says some about how well the compliance behaves, the height of the peak says something about the BL.
Which is measuring the impedance with a constant voltage source (very important, so NOT the resistor attenuation method) and keep an eye how the frequency shifts of the impedance peak, as well as how smaller this impedance peaks gets, as function of output voltage.
The shift in frequency says some about how well the compliance behaves, the height of the peak says something about the BL.