View previous topic :: View next topic |
Author |
Message |
jacqueskleynhans
Joined: 10 Apr 2008 Posts: 109 Location: Cape Town, South Africa
|
accelerometer mounting |
Posted: Wed Aug 27, 2008 12:36 pm |
|
|
So i have successfully written the the code build the board now for the maths. How can i do a vector compensation so that the reading on the accelerometer is not influenced by the natural force of the earth??
My accelerometer is reading accurate till 4mg best i can do for 10bit. Just need to solve this very last problem. To simple subtract the value is not going to work!! So has anybody have any ideas regarding my problem. Help will be greatly appreciated
Kind regards
Jacques Kleynhans _________________ "THE ONLY EASY DAY WAS YESTERDAY" |
|
![](templates/subSilver/images/spacer.gif) |
Guest
|
|
Posted: Wed Aug 27, 2008 12:45 pm |
|
|
Hi,
If you AC couple the output you loose the effect of static G.
Not sure if this is what you are looking for, give more info. if not.
Regards |
|
![](templates/subSilver/images/spacer.gif) |
jacqueskleynhans
Joined: 10 Apr 2008 Posts: 109 Location: Cape Town, South Africa
|
** |
Posted: Wed Aug 27, 2008 1:46 pm |
|
|
The thought of an ac couple has crossed my mind, but dont you lose resolution? I need to monitor slight vibrations from almost 20 hz to vibrations in the range of 1200 hz. Im doing a remote monitoring system for high voltage transformers.
Measuring their vibration at fundamental and at the harmonic intervals of 300, 500, 700, and so on.
Kind Regards _________________ "THE ONLY EASY DAY WAS YESTERDAY" |
|
![](templates/subSilver/images/spacer.gif) |
newguy
Joined: 24 Jun 2004 Posts: 1903
|
|
Posted: Wed Aug 27, 2008 2:08 pm |
|
|
Some thoughts.
It's reasonable to assume that the vibration of a transformer will be small compared to the force of gravity. In other words, you can very likely expect the vibration to be much much less than 1g.
Under this assumption, your signal of interest (the vibration) will be very difficult to see because of the effect of gravity. I also assume you don't care about DC acceleration (gravity) because you want to analyze vibration. Basically, without AC coupling, your signal will be like a very small AC signal added to a very large DC bias. Sort of like trying to determine the mass of a fly by measuring the change in mass of an elephant with and without the fly on its arse. Depending on your accelerometer IC and the A/D, you really won't see a large variation in the accelerometer's values. Good luck extracting useful spectral information from that.
Your best option is to AC couple the accelerometer's outputs, amplify & filter the signals, then feed them into an A/D. Designing a HP or BP filter circuit with a low frequency cutoff of ~20Hz is actually very trivial. Implement a programmable gain amplifier into the signal chain, and you can then implement an automatic gain control function in your product that can automatically tailor the gain to obtain the best resolution without clipping. |
|
![](templates/subSilver/images/spacer.gif) |
RLScott
Joined: 10 Jul 2007 Posts: 465
|
Re: accelerometer AC-coupling |
Posted: Wed Aug 27, 2008 2:10 pm |
|
|
jacqueskleynhans wrote: | The thought of an ac couple has crossed my mind, but dont you loose resolution? I need to monitor slight vibrations from almost 20 hz to vibrations in the range of 1200 hz. Im doing a remote monitoring system for high voltage transformers.
Measureing there vibration at fundamental and at the harmonic intervals of 300, 500, 700, and so on.
Kind Regards |
No, you will not lose resolution if you AC couple your signal. If you want to read 20 Hz signals with 10-bit accuracy, you could use a single-pole high-pass filter with a cutoff frequency of 0.056 Hz. However,...
If 1 bit of A/D input corresponds to 4 mg, then full-scale (1023) corresponds to 4 g. So if you are capable of reading values of acceleration over a 4 g range, then 1 more g (due to earth's gravity) is not going to make much difference. Reduce your amplifier gain a little and read in the DC-coupled accelerometer reading and do the baseline removal in software. The only reason for wanting to AC-couple away the gravity is so that you could amplify a much smaller range of accelerations to cover the full range of the A/D converter and get better than 4 mg / bit.
So, what is the range of accelerations that you need to read (in g units)? _________________ Robert Scott
Real-Time Specialties
Embedded Systems Consulting |
|
![](templates/subSilver/images/spacer.gif) |
jacqueskleynhans
Joined: 10 Apr 2008 Posts: 109 Location: Cape Town, South Africa
|
** |
Posted: Wed Aug 27, 2008 2:48 pm |
|
|
well firstly I'm not using any amplification circuitry on my sensor just the basic LPF described in the data sheet. My sensor has a g range of +-2g and i am utilizing it fully. We did some tests in the lab on diy vibration system running a 100hz sine signal through a amp into a speaker with a makeshift table and connecting the accelerometer to that and it actually shows its vibrating a -+500mg.
Quote: | So, what is the range of accelerations that you need to read (in g units)? | i cant really say previous test have shown +-2 is a good number.
I can add my code if you want to have a look?
Quote: | It's reasonable to assume that the vibration of a transformer will be small compared to the force of gravity |
No not at all those transformers actually vibrate quite heavily at certain frequencies.
Baseline removal was mentioned: the i am not actually involved in analysing the spectral information. That will be done by someone else. My role in this thesis design is to develop an low cost front end vibration and magnetic flux sensor for an crossbow or telosb mote were using.
I just have to get a reasonably actually kinda accurate representation of the measure data so that they can take it apart in matlab. _________________ "THE ONLY EASY DAY WAS YESTERDAY" |
|
![](templates/subSilver/images/spacer.gif) |
RLScott
Joined: 10 Jul 2007 Posts: 465
|
Re: ** |
Posted: Wed Aug 27, 2008 4:24 pm |
|
|
jacqueskleynhans wrote: | ...previous test have shown +-2 is a good number...Baseline removal was mentioned:....I just have to get a reasonably actually kinda accurate representation of the measure data so that they can take it apart in matlab. |
In that case forget about AC coupling. Record the accelerometer readings offset and all. They will remove the offset in matlab when they do spectral analysis. As I said before, the only reason for considering the removal of the DC bias would be to improve the resolution for very tiny signals. But you do not have tiny signals. You have big ones: +/- 2g. So there is not much to be gained by removing the 1g of bias. _________________ Robert Scott
Real-Time Specialties
Embedded Systems Consulting |
|
![](templates/subSilver/images/spacer.gif) |
jacqueskleynhans
Joined: 10 Apr 2008 Posts: 109 Location: Cape Town, South Africa
|
** |
Posted: Wed Aug 27, 2008 4:34 pm |
|
|
Thx for the information.
There must be a way of doing it by means of some sort of vector analysis then compensating for it. I don't know all the maths but surely I can figure it out? _________________ "THE ONLY EASY DAY WAS YESTERDAY" |
|
![](templates/subSilver/images/spacer.gif) |
RLScott
Joined: 10 Jul 2007 Posts: 465
|
Re: ** |
Posted: Wed Aug 27, 2008 7:09 pm |
|
|
jacqueskleynhans wrote: | Thx for the information.
There must be a way of doing it by means of some sort of vector analysis then compensating for it. I don't know all the maths but surely I can figure it out? |
Sure, if you know the exact orientation of the accelerometer. The effect of gravity on each axis will be 1g multiplied by the cosine of the angle formed between the axis in question and the straight up vector. _________________ Robert Scott
Real-Time Specialties
Embedded Systems Consulting |
|
![](templates/subSilver/images/spacer.gif) |
jacqueskleynhans
Joined: 10 Apr 2008 Posts: 109 Location: Cape Town, South Africa
|
** |
Posted: Thu Aug 28, 2008 4:38 am |
|
|
Is there an sort of easy way doing this, or is it going to be hard you whould not now of an tut or something descibing this procedure this. This method is cosine interpolation right. _________________ "THE ONLY EASY DAY WAS YESTERDAY" |
|
![](templates/subSilver/images/spacer.gif) |
SherpaDoug
Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA
|
|
Posted: Thu Aug 28, 2008 7:38 am |
|
|
The accelerometer will have an offset due to gravity. But it will have other offsets too, due to sensor imbalance, temperature drift, aging, etc. The result will be the sum of all these offsets. It does not pay to work too hard on one offset while others are still unknown. I would just output accelerometer readings, and let the Matlab guys identify and remove offsets. _________________ The search for better is endless. Instead simply find very good and get the job done. |
|
![](templates/subSilver/images/spacer.gif) |
FvM
Joined: 27 Aug 2008 Posts: 2337 Location: Germany
|
|
Posted: Thu Aug 28, 2008 7:49 am |
|
|
I basically agree with the opinion, that AC coupling doesn't give much additional sensitivity with an 10 Bit ADC. You didn't yet tell, which device you are using, is it an Analog ADXL? The DC bias can be removed in software as well, and you don't need to know about the exact gravitation component, I think. So there's nothing to calculate particularly.
For a maximum sensitive vibration detector, usually a seismic acceleration sensor (either piezoelectric or moving coil) is better suited. But I guess, you have checked that your device has sufficient sensitivity respectively low noise for your application. |
|
![](templates/subSilver/images/spacer.gif) |
jacqueskleynhans
Joined: 10 Apr 2008 Posts: 109 Location: Cape Town, South Africa
|
** |
Posted: Thu Aug 28, 2008 8:09 am |
|
|
Im using a LIS3L02AL 3 axis -+2g analog accelerometer. with an sensitivity of 660mv/g. This device has very small sensitivity change wrt temperature change. You said this can be done in software, otherwise the matlab oukes are going to have to remove it.
I know the device does not have military precision but i suppose for my application its alright. _________________ "THE ONLY EASY DAY WAS YESTERDAY" |
|
![](templates/subSilver/images/spacer.gif) |
|