Accelerometer based top balancing

Started by spincakes, February 03, 2024, 07:44:03 AM

Previous topic - Next topic

ortwin

David, maybe you could describe how the strobe light is triggered. I think you said when the acceleration signal is at its maximum? Does your sensor measure the x and the y component of the acceleration separately and then you combine them and calculate the magnitude of the acceleration in the plane? And than the light flashes as soon as that magnitude becomes smaller?
That would be my first idea how to proceed. Probably a bit naive and not the peak of elegance but I am curios how you really do it.


No! On reading that again it does not make much sense what I said. So I think in private a bit more, and wait for your answer.

In the broader world of tops, nothing's everything!  —  Jeremy McCreary

ortwin

Now I don't keep my promise to just think and wait, because now I think I might have understood: you detect only one component of the acceleration in the plane, let's say the x-component. You get some roughly sinusoidal signal. Every time the signal goes through zero, you know that the top is leaning at 90 degrees to the x-direction. Additionally you take care that you flash only at every second time the signal goes through zero, say on the way up.
I am not sure this makes any more sense to you now...


In the broader world of tops, nothing's everything!  —  Jeremy McCreary

spincakes

Quote from: Iacopo on February 07, 2024, 03:03:01 PM
One possible cause, (I don't know if this is your case), are the different proportions of the tops.

That could very well be part of the problem! Might explain a slight difference I've seen at different speeds as well. Thank you for the clear explanation. I will experiment more when I have the lenses for the base. The cardboard dents easily, which might also mess things up.

Quote from: ortwin on February 07, 2024, 03:13:08 PM
David, maybe you could describe how the strobe light is triggered. I think you said when the acceleration signal is at its maximum? Does your sensor measure the x and the y component of the acceleration separately and then you combine them and calculate the magnitude of the acceleration in the plane? And than the light flashes as soon as that magnitude becomes smaller?

Close! But not exactly. It measures X and Y separately, which I then combine into a magnitude (sqrt(x² + y²)) and phase (arctan(x, y) * 180/π). Whenever the phase crosses 0 degrees, I will flash the light. You can see all of the waveforms in the image "phase.jpg" in one of the earlier posts. I have reattached it here. It shows the calculated phase at the top, then the measured X and Y components, and then the calculated magnitude (which still has a tiny ripple, as the movements weren't perfectly sinusoidal. It's referenced to 0 in the image, so is about 0.70m/s² there.) If the magnitude is small (<0.05m/s²) I will not flash the light, otherwise it will keep flashing to the noise of the sensor when there is no top spinning.

Iacopo

Quote from: ortwin on February 07, 2024, 02:53:20 PM
I was of the impression that with the tip method one can in principle correct the static as well as the couple unbalance but not the visual unbalance.

Without the reference of a symmetry axis, even the meaning of the terms "static" and "couple" is relative...

If I have an unbalanced top, A, (the principal axis of inertia not passing through the tip), and I don't care about the symmetry axis, I can fix the unbalance by tilting that principal axis of inertia, (B), or by shifting it, (C).
In B I corrected the unbalance as it was couple unbalance.
In C I corrected it as it was static unbalance.
If there were screws in the top, the unbalance would have been fixed by the vertical screws in B and by the horizontal screws in C.
Both the ways can work.
So, that in A was static or couple unbalance ?
 


ortwin

Yes, for the accelerometer method it seems not important how you correct the unbalance. Once the principle axis of inertia runs through the contact point the method is finished even though the movement of the top might still look terrible. If you have a very well defined symmetry axis that includes the contact point, other methods (for example the paintbrush method) could be a better alternative or at least an additional possibility to improve "visual unbalance".


In the broader world of tops, nothing's everything!  —  Jeremy McCreary