Discovery Lite AHRS

[bvolcko]

Flying yesterday, all systems nominal, fat dumb and happy, glass smooth air, when suddenly my horizon showed a 15 degree bank to the right. My wings were level, flying straight. After landing horizon was back to normal. !!!!

[mikeh]

Was the engine running? Vibration often causes this.

[rainier]

Was the engine running? Vibration often causes this.

Yes correct - this typically happens if a resonance is hit but basically this is what happens:

The gyro vibrates in one or more axis - the movement is a fraction of a mm usually but in bad cases can be more. The movement more or less approximates a sine wave (sometimes with multiple harmonics - think of a prop via a gearbox with non-integer ratio).
The movement's frequency can be quite high - several 100 Hz to several Khz.
The internal algorithm looks at the gyro a fixed number of times a second - usually 100Hz is a popular frequency in the AHRS World.
The vibration/movement causes two issue - the gyro itself is a small vibrating structure (usually vibrating at around 20 to 40 KHz depending on type) and this vibration is measured - the actual position of the vibrating structure relative to the surrounding silicon chip is used to measure a force - this force is proportional to the rate of rotation we are interested in - but now there is another force and if the frequency of that force starts to have a relationship with the gyro's own frequency we start seeing a rotation that does not exist.

So the algorithm in the AHRS thinks in this case you are busy rotating into a bank. But there is something else also measuring force - the accelerometer - this works very similar - also a vibrating force but rather than looking for a twisting displacement of the vibrating mass we look for a linear displacement (we would like to use this to find out which direction the planet is relative to us). So we use this to measure G-force and the direction of that force. This we can use whenever we can sense that we are not in an accelerated flight maneuver (i.e. we going in a straight line, speed constant) - to get an idea what the horizon position is. Whenever we can't do that we rely on our previous known attitude and whatever rotations we measured since then - it's obvious we need to measure that with extreme precision or else our attitude (which is now an estimate) is going to be horribly wrong very quickly.

Now - we can see that the accelerometer derived attitude can be used to force our estimate towards the correct picture while the gyros give a small error. However, exactly the opposite can also happen - the accelerometer gets it wrong as the vibration components happen to be linear rather than rotational.

Our current crop of AHRS modules we put into the Lite's and MX1 as well as the smaller Vega and Blaze AHRS indicators have a new algorithm developed with help of a very large bass speaker that has been modified to become a versatile vibration source - via a powerful amp and a synthesizer we can create any shaped vibration and frequency combinations possible. This allowed the creation of a more clever sampling method for the accelerometer that all but eliminates the effects of vibration on the accelerometer. It does not work for the gyros - here the problem is more complex. But the gyros are more affected by high frequency vibrations which can usually be dampened fairly easily. They do however not like low frequency vibrations that have large rotation components - the only real solution to that are ring gyros such as used in our SP-9. Here these kind of vibration effects cancel each other out due to the way the ring oscillates. But this is still quite expensive.

The bottom line for any AHRS sensor to perform well - it must be mechanically tightly coupled to the airframe and a rigid part of the airframe at that but in a way that any high frequency signals do not get to the sensor. Vibration is actual physical movement and we need to measure the movement of the airframe with great accuracy without having to deal with any other movement on-top of that.

[bvolcko]

Was the engine running? Vibration often causes this.

Engine was running, but it has been running many times before without this happening.

[bvolcko]

Rainier,
This is very interesting. How can I keep this from happening in the future, or correct it when airborne? Changing engine rpms didn't seem to make a difference. I have flown for approx 70 hrs and this is the first time this has happened.

[rainier]

If it is related to RPM this tends to happen only at one particular RPM setting and goes away shorty after changing RPM. It would be repeatable every time you hit that RPM.

If it is not related to RPM it could be an unusual operating temperature (that affects older "Lites" more than the newer models as the newer models that have the plug in AHRS module use a discrete accelerometer made by Analog Devices which is pretty stable).

Other than that I could only imagine a case where on startup it did not read correct calibration values - that however is done only when you apply power.

[bvolcko]

Thank you. I will do more testing. My sn is i dj71100250