Oct 17, 2007, 03:13 AM Registered User Joined Sep 2007 41 Posts Question MEMs gyro accuracy What kind of accuracy should I expect from a rate gyro like this? http://www.sparkfun.com/commerce/pro...roducts_id=396 0.5 degs/s? I looked through the ADXRS150 Datasheet, but couldn't figure it out. The datasheet is here http://www.analog.com/UploadedFiles/...s/ADXRS150.pdf
 Oct 17, 2007, 07:30 PM Registered User Joined Jan 2007 15 Posts If ADI's datasheet is to be believed, 0.5 deg/sec accuracy is pessimistic, if the temperature of the gyro stays constant, or if temperature is compensated for (either in software or by a hardware method of maintaining the temperature of the gyro). One caveat: In the plot in the datasheet entitled "Figure 7. Null Settling Time", it's noted that "0.6 Hz sampling" is used. If it's truly "sampling" time, then that's good, because it shows that the noise peaks are greatly limited in magnitude. However, what they really "meant to say" might be AVERAGING time, in which case the noise peaks might be a lot higher than what they're showing in that plot. In that case, the averaging of the gyro's output over the 1.67 seconds for each point on the plot would tend to lower the high-frequency noise peaks and make it look like the performance of the gyro was better than it really is, for higher-sample-rate applications (sampling at greater than 2 Hz, say, which is not a very high sample rate at all, for the kinds of things we're doing). Dave
 Oct 19, 2007, 03:53 PM Registered User Joined Aug 2007 1,514 Posts Also, for such a slow sampling rate, they must have set the slow bandwidth with large capacitor value on the output analog signal. Personally, I'm interested more in data rates from 100-500Hz, so I use tiny caps like 0.1uF. I then have the freedom do do fast sampling with noise, and mathematically smooth and average the noise to whatever level I desire.
 Oct 19, 2007, 04:46 PM Registered User Joined Sep 2007 41 Posts So do you think the accuracy of 0.5 deg/s or less is practically plausible with MEMs of this class?
 Oct 20, 2007, 07:17 PM Registered User Joined Jan 2007 15 Posts According to the testing of three ADXRS300 gyros reported in this paper (masters thesis) http://epubl.ltu.se/1402-1617/2005/3...X-05307-SE.pdf you could expect to get 0.5°/sec accuracy if you average the gyro's output over intervals longer than 0.03 seconds each. So you can generate output values at 30Hz that have that accuracy or better, if you properly average the gyro's analog output over the entire interval between each 30Hz sample. Longer averaging times (which inevitably lead to greater lag in the output stream, limiting its application in high-frequency control of a vehicle) give you better accuracy. I don't remember offhand exactly what the consequences of the "150" version of the ADXRS vs the "300" version are, as far as noise levels. I think I remember that the sensor elements themselves are different between the two, though, so you can probably plan on roughly half the noise in the "150" version, because the gain on the inertial input is greater. Dave
 Oct 23, 2007, 07:21 AM Registered User Joined Sep 2007 41 Posts Thanks for the replies. I found another rate gyro on sparkfun. Dual-axis IDG-300 http://www.sparkfun.com/commerce/pro...roducts_id=698 http://www.sparkfun.com/datasheets/C..._Datasheet.pdf The rate noise density is 0.014 deg/sec/sqrt(Hz), therefore accuracy is going to be 0.1 deg/sec at 50Hz. If that's true, I get much better accuracy and the second axis for free. Is it too good to be true?
 Oct 23, 2007, 05:56 PM Registered User Joined Jan 2007 15 Posts That does seem to be a good noise density figure, but since you can measure noise density at any frequency, you have to wonder whether this number is an average over some range of frequencies, or at some certain frequency, or exactly what it is. "Bias Stability", for example, is something that might be excluded from this number, but that could have a dramatic effect on the usefulness of this device, depending on what you're trying to use it for. Bias Stability would be a drifting of the DC level of the output over a period of time, without any change in input (angular rate). I think we need more data on the IDG-300 before it's possible to say that it's too good to be true. Dave
Oct 25, 2007, 04:47 PM
Registered User
Calif*
Joined Sep 2006
2,226 Posts
Would forget about IDG300's and focus on the 75 deg/sec gyros, having used IDG300's and ADXRS150's. The IDG300's drifted a bit more than the ADXRS and 300deg/sec is overkill. Even for copters, the vibration is within 75 deg/sec. U can expand the gyro range with voltage dividers.