ADXRS150EB questions

[velocipenguin]

We are using an Analog Devices ADXRS150EB angular rate sensor as a component of our inertial navigation system. While testing it with an oscilloscope, I've noticed that its behavior differs from what I expected. It is being powered from a very well-regulated and filtered power supply that is supplying 5 V DC at up to 1 ampere. The 2.5V reference output on the rate sensor board measures 2.5 V, and the rate out pin hovers close to that when the device is not being moved. When moving the device, I've noticed that the output waveform is not steady; rather, it looks very much like a high-frequency triangle wave whose amplitude corresponds to the voltage representative of the device's rate of rotation. Is this behavior normal for the ADXRS150? Should I add additional filtering on the output? I don't want to get weird aliasing errors when sampling this data, and I'd love to know if anyone knows anything about whether this sort of output waveform should be expected with this part. Thanks in advance.

[KenWittlief]

if you have the eval board version, the output signal is already filtered with a cuttoff around 40Hz.

Question I gotta ask? what sort of angular rate test fixture are you using to produce a steady yaw rate while you monitor the output?

if you are holding the eval board in your hand and twisting your wrist back and forth, then the strange waveform you are seeing is exactly what your wrist is doing.

It would take a pretty expensive test fixture to move the sensor at a steady 75° per second, so that you could see a nice steady signal coming out.

if you try holding it steady in your hand, even that will show a jittery output- thats not sensor noise, thats your shakey hand :c)

I did the same thing with ours - same device - the signal was all over the place when I turned it in my hand.

We mounted it on our bot, closed the loop, and our bot tracks the sensor signal just beautifully - we are adding the signal to an accumulator to create a compass heading, and to steer closed loop we do the same thing with the signal from the X axis of the driver joystick - we divide it by 2, then add it to an accumulator.

This gives us a desired heading (from the accumulated joystick signal) and an actual heading (from the accumulated yaw rate signal)

then we subtracted the one from the other, and use a proportional error signal as our 'commanded' steering (X axis) input to the one joystick equations.

robot steers like its one of the Hitec servos!

[jacob_dilles]

It would take a pretty expensive test fixture to move the sensor at a steady 75° per second, so that you could see a nice steady signal coming out.

and twisty proof wire lol

[KenWittlief]

hey I just thought of a relatively easy way to make a smooth test fixture for the yaw rate sensor.

get a fairly heavy weight and make a solid pendulum, maybe a couple feet long, and securely attach the sensor to the pendulum

then when you let it freely swing back and forth, the sensor will have a nice and smooth yaw rate, going positive and negative.

Darn - I wish I would have thought of this when I was showing the students how it works :^)

[Mark McLeod]

I suppose you could cobbled one to an old clock work sweep second hand (industrial sized clock motor) and make a swivel connector from one of those telephone handset cord detanglers. Then you'd get a very regular 360 degree movement. Just don't pick a clockwork that "ticks" off the seconds.

[velocipenguin]

if you have the eval board version, the output signal is already filtered with a cuttoff around 40Hz.

I know that. However, the datasheet for the evaluation board states that "the user may add an external capacitor to further reduce the bandwidth and improve the noise floor." I was just curious as to whether anyone would recommend doing that.

Question I gotta ask? what sort of angular rate test fixture are you using to produce a steady yaw rate while you monitor the output?

if you are holding the eval board in your hand and twisting your wrist back and forth, then the strange waveform you are seeing is exactly what your wrist is doing.

I was indeed using my hand to rotate the device. However, the waveform generated was more regular and of a higher frequency than anything I suspect my hand would be capable of inducing.

I did the same thing with ours - same device - the signal was all over the place when I turned it in my hand.

That's very reassuring. That part was expensive enough that the thought of having to replace it was not a pleasant one. I'll try it out with a more stable testbed.

We mounted it on our bot, closed the loop, and our bot tracks the sensor signal just beautifully - we are adding the signal to an accumulator to create a compass heading, and to steer closed loop we do the same thing with the signal from the X axis of the driver joystick - we divide it by 2, then add it to an accumulator.

That is a remarkably cool idea. We're using a trapezoidal integration algorithm to generate a similar absolute heading value. The data from that is being used concurrently with the second integral of the output of a Motorola MMA1201P accelerometer and a few trigonometric lookup tables to allow us to determine the robot's longitudinal and latitudinal deviation (in feet) from its starting location.

[KenWittlief]

full scale on the ADXRS150 is +/- 150° per second

the second hand on a clock moves at 6° per second, which wouldnt be much of a signal to look at

an old vinyl record player would be a good choice.

[Mark McLeod]

an old vinyl record player would be a good choice.
What d'ya mean "old"? I'm still using my record player!

33rpm is probably a little too fast for the 150. Have to use a 300. That would make a nice demonstration though.

[KenWittlief]

hey, your right

you would have to put it on 17.5 rpm :ahh:

[DougHogg]

We finally got a gyro chip (ADXRS150EB) yesterday. They were out of stock at Analog Devices although their web site said otherwise, so we had to get it from Future-Active.

The numbering of the capacitors on the board is different than the numbering on the diagram that came from Analog Devices. Did anyone else observe this? (We are wondering if we got a different version.)

[KenWittlief]

yeah that freaked me out too - the pin numbers are right on the DIP evale board package though