Looking for a high accuracy angle sensor.
Accuracy : 0.01 deg
Range : +/- 30 deg
On single axis.
Need advice from the community here.
Look forward to some nice discussion/advise.
When you checking angle sensor, do you want to use it in motion status or motionless status? Do you need to test pitch, roll, yaw or only one of them?
If you want to test pitch or roll in status condition, then accelerometer can be used to do so, eg: ADXL345, ADXL362, ADIS16209, ADIS16210. ADIS16210 has the best accuracy accross the full temperature which can reach 0.1 deg accuracy. Please note that 0.01 deg requirement is too tough. Even you can calibrate the zero g bias, sensitivity, cross axis sensitivity perfectly, there are also noise, hysteresis, zero g bias shift over life, PCB bend and twist because of tempco and time... In short, 0.01 deg accuracy cannot meet with our MEMS accelerometer.
If you want to test roll and pitch/roll in motion status, you need gyro or even IMU which integrate gyro + accelerometer + magnetometer, you also need the algorithm. Take ADIS16480 as an example, it integrates gyro + accelerometer + magnetometer + algorithm, its best accuracy is also 0.1 deg. You can check these parts firstly and see whether they can help you.
Thanks for the post. Thanks for the useful info.
Im looking for a angle measurement unit- which can be fixed onto the body and measure the angle on a single axis. For Eg: The sensor is mounted on a rigid body and some roll motion is given, the sensor should measure the roll angle. There might a bit of pitch, which I assume will be ignored, when we use a single axis sensor (correct me if Im wrong about this).The roll motion given will be less than 30 deg amplitude and rate of motion is considerably small. I hope this helps you get an idea of what Im looking for.
ADIS16480 is too expensive for me. I haven't gone through details of other units except for ADIS16210. I would like to request you to explain what relative accuracy means here(as mentioned in the data sheet). As of now, ADIS16210 seems to be an option (not finalised yet though).
I wasn't aware of limitation on accuracy with MEMS sensors for angle measurement. Thanks for clarifying. Are there any alternative to get accuracy levels better than 0.1 deg.
If you just need to detect the roll in static condition, ADIS16210 would be a good option. When it says relative accuracy, this means if you can give a absolute 0 deg or say give a reference position, when you turn 0.1 deg, the ADIS16210 will change 0.1deg too. But the initial output at absolute 0 deg, may be 0.05 deg. When we calibrate the part before shipping to cusotmer, we cannot make sure the reference platform that we offer is absolute 0 deg. So, we use relative accuracy. Please note, the reference platform that you could offer may not be absolute 0 deg too .
Refer to better than 0.1 deg, this is really hard to meet. If you give the constant temperature, did not consider any drift, make BW as low as possible, say update the angle every one second or even every ten second, the accuracy could be better. But, if you consider all the error source that I listed before, 0.1deg accuracy is pretty good already.
In addition to neilzhao's valuable insights, I was wondering if your application needed 0.01 degrees of absolute accuracy or if you only needed to detect that level of movement. Some have referred to this as "resolution?" Detecting that degree of movement, using the ADIS16209 or ADIS16210 might be possible, with 10-20 seconds of averaging, under stable environmental and mechanical conditions. Over the span of the ADIS16209 package size, that equates to <2um (0.000002m) of movement. When you consider solder expansion/contraction, moisture absorption or temp cycling in PCB laminate materials along with many other mechanical sensitivities in standard assembly materials, that is a tough set of conditions to manage, even if you have a perfect inclinometer.
Its not exactly clear for me about the relative accuracy. Excuse me for being too slow to grasp things.
The question I have is, if the relative point itself is changing, how exact are the measurements going to be. I presume, the readings will be with reference to a position, we set as zero.
I haven't mentioned it yet- but the movement to be measured is actually swinging motion about the reference axis.
I hope it is ok to take the case of slow rate of movement as static conditions (roll measurement)!
Thanks for the valuable inputs.
As NevadaMark said, may be it is the resolution of 0.01 deg, that I should look at.
If the relative point change, then the ADIS16209/ADIS16210 output will change too. Say, with the original reference point, the ADIS16210 output is 10deg. The reference point change 1 deg, then ADIS16210 output will change to 11 deg too. After that, any change on reference point will be reflected on ADIS16210 output, it is 11 deg +/- change angle.
If you want to read the tilt angle during the movement, then you need a gyro.
No matter how fast you move or shake the part / your final product. When you want to read the tilt angle, keep the part stable without any movement, then accelerometer is enough. In such situation, as NevadaMark said, better accuracy is possible but the trade off is the response time. You need to do long time averaging, 10-20 seconds.
Is setting the reference point a one time activity - I mean if I fix the sensor onto a body. Can I feed the reference point programatically and make it zero to begin with. Then take the measurements. Or do I have to set reference point each time I power up and take measurements.
Since, I need to take measurements doing a swinging motion, I need a gyro as well. Can you suggest me suitable gyro units.
Normally, you just need to set the reference point only one time. If you need both gyro + accelerometer, you can consider the IMU ADIS16445, ADIS16485. These parts could be a good start for you. The part itself has take care anything for you that related to hardware, you can focus on software with those parts. But in motion status, they cannot give you 0.01 deg accuracy. It is too tough. Around 0.1 deg accuracy is possible.
Seems an expensive option. Any alternatives?
Further,does it come with software for programming/configuring?
We appreciate you posting in this forum. I can see that you and neilzhao have had an excellent exchange; I am glad that he was able to help you with this because he does fantastic work!
Can you help me? I am a bit confused by your last response. You have been very specific about your performance requirements but haven't shared what your pricing or estimated annual usage expectations are, which are pretty important to us when trying to help you understand the performance/price trade-offs that we are aware of. Perhaps as important, what is setting those price expectations? We have many customers that would argue the exact opposite position: that products like the ADIS16445 and ADIS16485 offer them tremendous savings over more traditional IMU technologies that cost >10x. While we understand that there is always a drive to to make things cheaper, solutions tend move forward more smoothly when we understand and optimize the trade space (performance, power, price, package, ease of use, etc) that present technology and personal experience will support. Of course, we are here to help any way we can, but we find that post-sensor software routines are often application specific and present key intellectual property for our customers.
At the levels of performance you are seeking, I would have to agree with neilzhao, dynamic estimation of orientation angle, to an accuracy of 0.01 degrees, seems to be outside of what commercial MEMS technology will support at this time. Perhaps under a narrow set of well understood and controlled conditions, but that is beyond my experience.
With respect to your question on software tools, the EVAL-ADIS comes with the IMU Evaluation Software package, which will enable data capture, register access and some basic functional test features, but it does not come with an API or other official development tools. We have posted some code examples that might help you get started but this platform really wasn't intended for development. Products like the ADIS16445 and ADIS16485 use a SPI interface, which typically connects to embedded processors that have their own development tools.
Yes. This forum has been really helpful especially neilzhao and yourselves. Thanks for the really useful suggestions.
Now,with reference to your question, I would like to let you know that I actually began with a certain price point in mind from a certain product in the market,which is about 250 USD with a similar spec I mentioned (i.e. 0.1 deg accuracy (-10 to 40 deg C) and 0.01 deg resolution.) So, my plan was to develop something equal or better within the target price.
I was never in doubt of the quality of the Analog Devices Product and impressed with the responses of the technical support, especially taking pains to explain to a learner.
Ok, now, it is never a bad idea to seek alternative possibilities/ idea. It would be interesting to see if there are means of achieving the specified ratings using ADIS accelerometer and some good gyro component - I do not know, just an idea, which you can advise, if it helps.
Hope I could explain myself. I will be happy to discuss about the product I was referring, if the forum has a private message option.
I have just one question. If I use ADIS16210 and I fix it onto a body.
When the body is given a swinging motion, I want to know the pattern of readings
from ADIS16210 let us say at 0,10,20,10,0,-10,-20,-10 and 0. Assuming minimal alignment errors.
Im asking this to understand the significance of relative accuracy.
First, if the device is in the act of swinging, it will likely be experiencing linear acceleration, which will influence the accuracy of the tilt measurements.
Second, the "relative accuracy" also means that you will observe
I hope that helps.
Regarding using ADIS 16210 for roll angle measurement during swinging motion, I was of the view that since the sensor has very low bandwidth (50 Hz) and does come with a programmable filter inbuilt, it would be immune to centripetal acceleration. Please correct me if my assumption is incorrect.
You are correct in your observations of attributes (50Hz, programmable filter) but it might be hard to make the leap of "complete rejection of centripetal acceleration." The bottom like is that the filters cannot remove 0Hz behaviors, because gravity would also be excluded. The internal filter is only a single stage averaging filter so its roll-off characteristics are finite. I suspect that this will not be able to 100% reject these influences but we can help analyze if you have magnitude/frequency information.
That is the first thing I had on my mind which I thought would have been helpful. Unfortunately, I don't have the spectral data. I really wish I did. All I can say is that angle is being measured when a swing motion is applied by hand on a rigid body, sorry for being vague. I estimate the rate of swinging motion to be around 20 to 30 Hz and magnitude the lowest possible - its just a push of hand.
I completely understand. That type of information often comes from the development process itself. With that in mind, my personal philosophy would suggest that we should assume that linear acceleration will influence accuracy in ANY accelerometer-based tilt sensor, until proven otherwise, under the conditions associated with a particular application.
Retrieving data ...