A perfect accelerometer calibration

Marco Falchi · Oct 28, 2016

To make a good calibration of the accelerometer, it checks with a spirit level that the surface on which rests the H is perfectly level.

Then I think: but check this with a spirit level, it makes sense?

The accelerometer is in the H fuselage, and the fuselage rests on the landing gear, and the landing gear is on the rubber plugs.

The rubber plugs may be deformed and the legs of the landing gear could not be perfectly located (some unorthodox landing), leaving the fuselage tilted of a few degrees.

In addition to that, I do not know if the accelerometer, inside of the fuselage, is perfectly level respect the fuselage.

All this makes me think that it is totally unnecessary to worry about having a perfectly level plan, because in any case, there may be an error determined by all elements interposed between the accelerometer and the plan itself.

I then proceed as follows: control, of course, the plan with a spirit level (though I delete an error), but then, before and after calibration, control with the GUI that the accelerometer responds well to the movements, and is "level", when to me, visually, it looks "level".

In doing so, today, I had no problems.

I am very interested in your opinion on 2 topics:

1) I'm crazy
2) you as you proceed to calibrate the accelerometer?

Thanks for any response.

PatR · Oct 28, 2016

The system does not require being perfectly level, it only must remain still and undisturbed during accelerometer calibration. If the surface it rests upon is within a couple degrees of level it will be close enough, providing the airframe does not otherwise obviously lean in one direction or another.

Sent from my iPhone using Tapatalk

Tuna · Oct 28, 2016

My understanding is that the point is for the Typhoon to figure out what 'not going anywhere' feels like. You calibrate the accelerometer with the Typhoon still and undisturbed so that when it's flying it knows what readings to expect from the accelerometer when it exactly holds position.

If you think about it, if you taught it that 'staying still' involved being tilted at 45 degrees, it would try to do that and fail - no drone can be still and hold a 45 degree tilt at the same time. On the other hand, so long as it's pretty close to level, then small adjustments in prop power should allow it to sit in one place. Being close to level should be good enough (I think), but it must be absolutely still.

Chilcotin · Oct 28, 2016

Tuna said:
My understanding is that the point is for the Typhoon to figure out what 'not going anywhere' feels like. You calibrate the accelerometer with the Typhoon still and undisturbed so that when it's flying it knows what readings to expect from the accelerometer when it exactly holds position.

If you think about it, if you taught it that 'staying still' involved being tilted at 45 degrees, it would try to do that and fail - no drone can be still and hold a 45 degree tilt at the same time. On the other hand, so long as it's pretty close to level, then small adjustments in prop power should allow it to sit in one place. Being close to level should be good enough (I think), but it must be absolutely still.

They can work that way, but in the case of the H, I think it is working with a "close to level" constant so it can determine its angle.
If it used the method you say, then it would be able to hold still in the wind with gps off, or at least try to fight it.

abinder · Oct 28, 2016

If they are truly being used the way an accelerometer is designed, it senses movement

Sent from my iPhone using Tapatalk

chiloschista · Oct 28, 2016

From other autopilots experiences, one of the most important aspect is to not move the board while calibrating.
Accelerometers should be leveled, gyros not. I guess gyros offset, yes there should be those too, are set before leaving the factory.
I already wrote about leveling the copter. I checked that and found that the copter body leans toward front by few degrees.
Did anyone check motors level? Here too, I guess that the still flying position should be with leveled motors, isn't it? Of course yes, if everything is well balanced!
Motors are not leveled.
Did anyone check for propellers alignment? On mine the propellers are not well aligned as well.
Summarizing: a couple of degrees from the body alignment, a couple of degrees of propellers misalignment, what about the IMU board?
I'm definitely not convinced it is necessary to be that finicky, a little bit yes, but not so much.
Please remember, the most of those device are intended, and have to work for him too, for the average Joe, who does know almost nothing about IMU, PIDs, stabilization, calibration, electronics, electricity, batteries, radio link, antennas, aerodynamics, geomagnetism, solar flares ... (wow, so many things = : D)
Ric

PatR · Oct 28, 2016

Ideally there would be offset capability but such is provided at much higher levels of auto pilots. Too many like to view our little flight controllers as ultra sophisticated, certifiable grade avionics systems. They most certainly are not. They are pretty darn good for what they are but not anywhere near a level where we should ever be totally dependent on them. None of them meet any certified design standards.

Sent from my iPhone using Tapatalk

Typhoon Charlie · Oct 28, 2016

I calibrate mine by placing a level from motor to motor and shim the landing gear so that all motors a level. This is what the flight controller will set as level. With no wind and GPS off I get zero wind.

Marco Falchi · Oct 29, 2016

Typhoon Charlie said:
I calibrate mine by placing a level from motor to motor and shim the landing gear so that all motors a level. This is what the flight controller will set as level. With no wind and GPS off I get zero wind.

View attachment 3882

Thanks for the reply.
You controls alone between 2 front engines, or even between a front engine and a rear engine ?

Typhoon Charlie · Oct 29, 2016

Marco Falchi said:
Thanks for the reply.
You controls alone between 2 front engines, or even between a front engine and a rear engine ?

Place the level between all the motors so that they are all level to each other. The H hangs from the motors when in the air, so you want all the motors to be on a level plane.