Hi,I understand the issues about using simple rate gyros with an autopilot - i.e. you can't unless teamed with accelerometers etc - but could one use a head-lock gyro for stabilisation via aileron control ?Max flight time would be about 10 - 15 mins.Simon

Interesting comments !
bGatti's original response reminded me of the occasion I watched a model jet aircraft being expertly flown straight and level down a runway, whilst continuously rolling around the x-axis; not too sure then how you'd use the gps vectors to correct the roll axis gyro.
The Sparkfun gyro is c $30, add an integrator plus assoc electronics and we're looking at around $70 which compares reasonably with the Co-Pilot IR solution at c $75. However I had in mind an ESky EK2-0704B head lock gyro which is promoted as incorporating drift cancelling. It doesn't claim 'zero drift' but certainly seems, on initial testing, to be pretty stable. I guess I'll just have to try it.

Given that U-Nav doesn't own the idea of a turn-rate wing leveller... I am REALLY shocked that there is no DIY-spawned open source or competing product at a much more reasonable price. It would be the World's simplest test case for a simple Kalman filter in aircraft stabilization: predictions are done by integrating gyro in yaw axis, and the update cycle comes with each GPS update. Simple. $400 for a Pico-N board??? I think NOT! The LISY-AL300 gyro is only around $10, and Sparkfun sells it now.

Even head-lock gyros drift over time. A heli only needs to hold a vector for a few seconds. A plane needs to know where "down" is for for many minutes.

The better Analog Devices gyros have a noise density of 0.05 deg/sec/sqrtHz.
To be generous, assume your aircraft has a pretty slow response frequency of 15Hz, so you only need to sample your gyro at 30Hz. This means a noise density of about 0.25 deg/sec. You need to integrate the output of your rate sensor to get your rotational position (which is what you are controlling for), which basically means to find the worst-case drift you would just multiply the noise density by the time in seconds. 15 degrees per minute is not a negligible error. Luckily, three-axis accelerometers are getting very inexpensive thanks to hard drive protection mechanisms in laptops, and gyros are thanks to image stabilization in consumer cameras. We may be getting close to a tipping point of affordable inertial stabilization.

## Comments

bGatti's original response reminded me of the occasion I watched a model jet aircraft being expertly flown straight and level down a runway, whilst continuously rolling around the x-axis; not too sure then how you'd use the gps vectors to correct the roll axis gyro.

The Sparkfun gyro is c $30, add an integrator plus assoc electronics and we're looking at around $70 which compares reasonably with the Co-Pilot IR solution at c $75. However I had in mind an ESky EK2-0704B head lock gyro which is promoted as incorporating drift cancelling. It doesn't claim 'zero drift' but certainly seems, on initial testing, to be pretty stable. I guess I'll just have to try it.

If we define level flight as flight resulting in forward gps progress, can we not use gps vector to correct the roll axis gyro?

The better Analog Devices gyros have a noise density of 0.05 deg/sec/sqrtHz.

To be generous, assume your aircraft has a pretty slow response frequency of 15Hz, so you only need to sample your gyro at 30Hz. This means a noise density of about 0.25 deg/sec. You need to integrate the output of your rate sensor to get your rotational position (which is what you are controlling for), which basically means to find the worst-case drift you would just multiply the noise density by the time in seconds. 15 degrees per minute is not a negligible error. Luckily, three-axis accelerometers are getting very inexpensive thanks to hard drive protection mechanisms in laptops, and gyros are thanks to image stabilization in consumer cameras. We may be getting close to a tipping point of affordable inertial stabilization.