Hey, i am preparing a circuit for LPR530 Gyro
http://www.st.com/stonline/produ...
Now in the datasheet's application hints section, they have 'Recommended' a low past filter and 'Suggested' a high pass filter as optional. Now i'm sure i dont want a high pass filter.
I need a low pass filter. The datasheet says there is an embedded low pass filter of Fc = 140 Hz. So is it necessary to add the 'Recommended' lowpass filter of approx Fc = 2kHz?
If you want a good low noise signal, you should have it.
My first advice is to use it. If you prepare your board to have all the required components, you always have the choice to populate only few of them in order to have a direct connection (using a 0ohm resistor and without any capacitor). From the noise point of view, it is always welcomed.
K thanks. I take it from this that since my internal filter has Fc=140 hz, my external lpf can have any Fc greater than 140Hz (upto say, 2Khz) and it will not matter. coz the external lpf is just double protection
If the internal filter is of the first order, it will lower the signals 3 dB/decade. So at 1.4kHz it will be 3 dB lower. Adding a filter at 2 kHz will add another order of the filter and after 2kHz the signal will drop 6 dB/decade.
There might also be some noise added to the signal after the internal filter that then producers of the chip want you to remove externally. And the internal filter might not be very good, since it has to be very small inside the chip.
Hey, i am preparing a circuit for LPR530 Gyro
http://www.st.com/stonline/produ...Now in the datasheet's application hints section, they have 'Recommended' a low past filter and 'Suggested' a high pass filter as optional. Now i'm sure i dont want a high pass filter.
I need a low pass filter. The datasheet says there is an embedded low pass filter of Fc = 140 Hz. So is it necessary to add the 'Recommended' lowpass filter of approx Fc = 2kHz?
The high pass filter is to remove the quite-significant bias from the gyro. Whether you use one or not depends on your application.
A low-pass filter is required for:
* Noise reduction, and
* Any signals present from the gyro's resonant frequency, which are quite strong and totally useless for any purpose. (Usually > 10KHz, but varies by manufacturer).
-- Damien
Depending on the application I would be very careful using the high pass filter. Filtering lower frequencies means you cant measure these. The bias and gain errors in the gyro is something you will have to compensate for anyway in software, and you will still be able to measure the lowest of frequencies. I used a gyro like this one and measured the rotation of earth using proper bias and gain compensation, even thou it is very very much slower than he initial bias of the gyro.
The high pass filter is to remove the quite-significant bias from the gyro. Whether you use one or not depends on your application.
-- Damien
what is this 'quite-significant bias'?
Is it the result of sudden changes in motion(say, vibration) or is it just one of the unfortunate characteristics of the gyro?
I'm using it for a two wheel bot btw..
damien_d wrote:
The high pass filter is to remove the quite-significant bias from the gyro. Whether you use one or not depends on your application.
-- Damienwhat is this 'quite-significant bias'?
Is it the result of sudden changes in motion(say, vibration) or is it just one of the unfortunate characteristics of the gyro?I'm using it for a two wheel bot btw..
All gyros have a bias that randomly varies with time and temperature. Since it's almost DC, a high-pass filter can remove it.
The problem is, if you need those low-frequency signals at the same time (say, a slow, steady turn), then you'll be filtering that out as well.
The best thing you can do is play with it and collect some data to determine what is best for your application.
-- Damien
I used a gyro like this one and measured the rotation of earth using proper bias and gain compensation, even thou it is very very much slower than he initial bias of the gyro.
Which gyro were you using to measure earth rate?
AgwanII wrote:I used a gyro like this one and measured the rotation of earth using proper bias and gain compensation, even thou it is very very much slower than he initial bias of the gyro.Which gyro were you using to measure earth rate?
It was four years ago, so I don't remember. But how expensive gyro I could have ever gotten back then, the simplest now is probably better. The one I used was just on the limit of being able to measure earth rotation. It was a MEMS gyro.
adwaitsdeshpande wrote:damien_d wrote:
The high pass filter is to remove the quite-significant bias from the gyro. Whether you use one or not depends on your application.
-- Damienwhat is this 'quite-significant bias'?
Is it the result of sudden changes in motion(say, vibration) or is it just one of the unfortunate characteristics of the gyro?I'm using it for a two wheel bot btw..
All gyros have a bias that randomly varies with time and temperature. Since it's almost DC, a high-pass filter can remove it.
The problem is, if you need those low-frequency signals at the same time (say, a slow, steady turn), then you'll be filtering that out as well.
The best thing you can do is play with it and collect some data to determine what is best for your application.
-- Damien
No no no no, the bias does NOT vary with time. At least not more than that it can be disregarded. Only with temperature. (older gyros were quite sensitive to acceleration, newer handle acceleration well) The bias is very systematic. I could not have measured the rotation of earth if I could not very exactly estimate the bias of the gyro.
The bias is a error the gyro does together with your electronics. When the gyro is perfectly still, it for instance say it is rotating 1 degrees/s. When it rotates buy 1 degree/s it sais 2 degrees/s. And when it is rotating -1 degrees/s it sais 0. It is just an error in measurement.
The gain error is a percentage of the measurement that is wrong. For instance if it is rotating 10 degrees/s it sais 11, and when it is rotating 20 degrees/s it sais 22, and -10 degrees/s it sais -11. That´s if the gain error is 10%.
All gyros have these errors.
No no no no, the bias does NOT vary with time. At least not more than that it can be disregarded.
The bias does indeed vary randomly with time, and I will stand by that statement. Most half-decent gyros will publish a bias stability figure.
Whether it can be disregarded depends entirely on your application. For attitude and navigation, this is most certainly not the case. For navigation, integrated gyro bias errors grow very rapidly with time, even for very small biases.
Only with temperature.
Temperature bias is fairly repeatable, except during rapid heating and cooling where you'll get hysteresis, depending on the quality of the device. It can normally be compensated for.
The bias is very systematic. I could not have measured the rotation of earth if I could not very exactly estimate the bias of the gyro.
For current commercial/automotive grade gyros under the $100 mark, the best I am aware of is 3deg/hr bias stability, which is insufficient to accurately sense the earth's rate to any degree of confidence which is useful. Please correct me if there is a device that I don't know about, because we're always looking for different gyros.
-- Damien
Hey, i am preparing a circuit for LPR530 Gyro
http://www.st.com/stonline/produ...Now in the datasheet's application hints section, they have 'Recommended' a low past filter and 'Suggested' a high pass filter as optional. Now i'm sure i dont want a high pass filter.
I need a low pass filter. The datasheet says there is an embedded low pass filter of Fc = 140 Hz. So is it necessary to add the 'Recommended' lowpass filter of approx Fc = 2kHz?
It uses a switched capacitor low-pass filter internally which will probably have a significant amount of clock-feedthrough present at the output. The external low pass filter is to remove that clock signal (probably a few hundred KHz). It is not random noise.
kevin
It uses a switched capacitor low-pass filter internally which will probably have a significant amount of clock-feedthrough present at the output. The external low pass filter is to remove that clock signal (probably a few hundred KHz). It is not random noise.kevin
damien_d, what I used the gyros for was navigation. The purpose was to create a UAV for the military using low price automotive/commercial components.
But we had quite nice, or rather very nice, instruments to calibrate the sensors with. In this navigation system I had to compensate for earth rotation and used a model of earth that was not spherical (earth is a little compressed over the poles).
This article talks about how to measure earth rotation. But its not easy.
http://www.iop.org/EJ/abstract/0957-0233/16/11/024
damien_d, what I used the gyros for was navigation. The purpose was to create a UAV for the military using low price automotive/commercial components.But we had quite nice, or rather very nice, instruments to calibrate the sensors with. In this navigation system I had to compensate for earth rotation and used a model of earth that was not spherical (earth is a little compressed over the poles).
This article talks about how to measure earth rotation. But its not easy.
http://www.iop.org/EJ/abstract/0957-0233/16/11/024
There are a couple of critical assumptions of the paper that need to be taken into account:
Firstly, it relies on the physical rotation of the sensor. Secondly, is best summed up by the paragraph: "The zero offset ψDC in equation (2) is assumed to be changing insignificantly among the measurements in the different positions, because of their relatively rapid change (every 2–3 s)."
Where the IMU is in an arbitrary position and is impractical to precisely rotate and calibate it (as is the case if it is permanently installed in a large vehicle, which is generally what I deal with), gyrocompassing cannot be done with this grade of sensor - the turn-on to turn-on repeatability is not good enough.
Additionally, if the operational time is longer than the bias stability, this method will initially work, then drift away without external aiding.
I am more than happy to be proven wrong - the gyro manufacturer will probably pick up a customer if do* - but I want to see the data first.
-- Damien
* $6000 KVH gyros excluded :) Incidentally, I suspect these are what are installed in these drool-worthy devices: http://www.novatel.ca/products/s...
Not exactly a "gyro schematic", but a picture is worth a thousand words...
http://www.pbase.com/bochie/imag...
I like that gyro!
Back on the original topic - one POTENTIAL reason for both internal and external low-pass filters is that the internal one is a switched capacitor filter. It will have noise at the switching frequency. You will need a low-pass filter (probably rolling off at a higher frequency than the internal filter) to reduce the switching noise. Such a filter could significantly reduce the "variance" of the digitized value. Additionally, the switching frequency is likely to be above the ADC sample rate, in which case, it gets aliased to a lower frequency which you cannot separate out from the "real" data.
Jim
