Hi,
I want to get angular rotation using a Dexter gyro but I have more 5-10 degree error in 90 degrees rotation.
This is how I use dps to get rotation:

I'm not sure if that's the correct way to get rotation. If that's the correct way, I guess the inaccuracy is related to third last line of my code. If I divide x_val by a number near 17 the error reduces to 1-2 degrees. (1000.0/loop_delay)=18.5 approx.

Please tell me if I am doing something wrong.
Thanks,

For an integration loop, it's best to loop as quickly as possible (maybe a 5ms wait). Also, I think using the 250 dps range will give you much higher resolution.

Thanks for your reply.
5ms wait doesn't make any improvement to 50ms.
with 250dps range, with gyro being static, the rotation starts to increase or decrease very fast. 2000dps gives much lower error in this case.

p.s : If i manually rotate the unit 90 degrees. total x_val will be 1520-1540. 1530/90 gives us 17 and if I divide x_val by 17 the output gets accurate.
Now the problem is that output keeps increasing or decreasing with respect to the direction of last rotation. I assume that happens to all gyroscopes..

You should not initialize time_temp to zero. That would introduce a random error to your integration because nPgmTime doesn't start at zero. Instead you need to do this:

There may be some other issues with the calculation but not knowing the spec of the gyroscope you are using, I wouldn't know.

That's correct. The code i wrote does not take into account the time before starting the loop.

Thanks for the tip !