The sensor seems to give acceleration instead of velocity. For example, when I turn the sensor briskly CW about 90 deg and then stop, I get the following readings. "Sensor" is the raw reading from the analog port minus the zero voltage (+ is CW, - CCW), and I'm polling at 100msec intervals:
It seems clear to me that the sensor is recording acceleration, not velocity, since the values go positive and then negative. I rotated the sensor in only one direction, if it was measuring velocity the sensor would clearly show values of only one magnitude. Also, when summing all the readings ("Total"), the total returns to nearly zero.
The chip is a vibrating structure type. Anyone else seen this behavior? Would a higher-quality or different sensor give me plain velocity output? I can of course integrate the output to get velocity but the documentation says I should get velocity.
The descriptions says: "The Grove – Signal Axis Analog Gyro is based on an angular velocity sensor...".
It measures the velocity of a rotation.
That is what your values show.
The used sensor seems to be the ENC-03R, there are not many others like that.
Why do you want to use this sensor ? Is there a specific reason for it ?
I'm using this particular sensor module because it is cheap, less than $10. Seeed, Sparkfun, Adafruit all have more expensive gyros but I wanted to start with something expendable.
The sensor values themselves go positive and then negative when rotating the sensor in one direction only at a more or less constant speed (as best as my hands can do), which means I'm getting acceleration, more or less. The values, when integrated (summed, in the "Total" column) do show velocity, more or less, so I can make this work in my application, which is to see how much a robot is turning while using cheap motors without shaft encoders.
Rate gyros do look like accelerometers. The output is in deg/sec. Measured versus time is deg/sec/sec. So, if the rotation velocity is 90 deg/sec after 1 sec it will have rotated 90 deg. Once you condition and scale the outputs you can calculate angles traveled.
That is inclusive shipping. It is an 3-axis accelerator and 3-axis gyro in a chip.
That is 6 axis for less money than your single axis weirdo sensor.
The software to interface it is complicated. You have to use I2Cdevlib.
Or to parody Dilbert, "Here's a nickel kid get yourself a better sensor."
For now I have a heuristic to use the sensor. My 2-wheel robot stops before it turns, then turns by running the wheels in opposite directions. So I propose to ignore the output when not in "turning mode", integrating (adding) the initial "slope" of the acceleration curve, and ignoring the deceleration part, when the slope reverses. This works pretty well on the desktop, we'll see how well when the robot is bouncing around on the carpet. The sensor is pretty good at ignoring linear movement and rotation not on the intended axis.
I kind of like Seeed - their prices are low, shipping is free for orders over $50, and they seem like the kind of guys it would be fun to hang out with and have a few beers with if I was ever in Shenzhen. I'd say they aren't for beginners. They don't exhaustively test all their products, but part of the fun is tweaking the sample libraries and trying to decipher the often-cryptic data sheets. It gives a window into how the China-rest of the world mass-market electronics connection works.