Magnetic Levitation System

[imcndbl]

issues with maglev

I built a maglev system and ran into the same issues as above.

there are multiple things going on here.

1) the control loop being unstable is due to pole locations. gain can change this but to acquire a real stable controller, the pole locations must be adjusted.

2) stable then dropping off is caused by two things a) bias to the system needs to be adjusted so that the drive transistor is in the linear region b) thermal heat up of the drive transitor and core of the magnet can/will cause poles to shift. see #1 above. put heatsinks on the magnetic core and on the drive transistor.

You can model the system in the freq domain and see if the poles are correct or you can put a pot in your compensation circuit between your amps, along with an adjustment on the gain of the output and do the following...

1) adjust gain down until stable
2) tweak the compensation in very little increments
3) watch levitating object ("ball") closely for oscillations.
4) when you turn and oscillations happen, turn back and go the other way.
5) adjust bias so ball more evenly floats in middle range of optical sensor
6) adjust gain down.

repeat in order or not whichever seems best or appropriate. adjsuting gain, compensation, and bias. you can watch the output of the optical sensor on a scope for better judgement of stability.

 

[shortbus=]

What you guys are talking about is called "Magnetic Suspension",the O/P asked for, 'magnetic levitation'. Magnetic levitation is what I've been answering about.

 

[DerStrom8]

What you guys are talking about is called "Magnetic Suspension",the O/P asked for, 'magnetic levitation'. Magnetic levitation is what I've been answering about.

This may be true, but the actual project that the OP is working on is simply a single electromagnet suspending a bolt. I'm assuming, in this case, that "levitation" is being used interchangeably with "suspension." Even if "levitation" isn't the precise term, I'm sure that we can all understand (thanks to the pictures and descriptions) that the OP is actually talking about a magnetic suspension system.
Der Strom

 

[4pyros]

DerStrom; I believe Shortbus was refering to the one or two electromagnet debate. Andy

 

[thecritic]

I built a maglev system and ran into the same issues as above.

there are multiple things going on here.

1) the control loop being unstable is due to pole locations. gain can change this but to acquire a real stable controller, the pole locations must be adjusted.

2) stable then dropping off is caused by two things a) bias to the system needs to be adjusted so that the drive transistor is in the linear region b) thermal heat up of the drive transitor and core of the magnet can/will cause poles to shift. see #1 above. put heatsinks on the magnetic core and on the drive transistor.

You can model the system in the freq domain and see if the poles are correct or you can put a pot in your compensation circuit between your amps, along with an adjustment on the gain of the output and do the following...

1) adjust gain down until stable
2) tweak the compensation in very little increments
3) watch levitating object ("ball") closely for oscillations.
4) when you turn and oscillations happen, turn back and go the other way.
5) adjust bias so ball more evenly floats in middle range of optical sensor
6) adjust gain down.

repeat in order or not whichever seems best or appropriate. adjsuting gain, compensation, and bias. you can watch the output of the optical sensor on a scope for better judgement of stability.

1. Done
2. You mean the capacitor? Done in discrete steps.
5. Not sure what you mean?

Note that the object is not very much unstable. If you watch for first few minutes then you may even swear the object is going to stay forever. But slowly and slowly it accumulates some pendulum like swings and in hour or so, it either drops or snap-attach with the core.
Am sorry, but Lately I have been using Pulse-width modulation at the final stage (In between the last op-amp and the Mosfet) (not mentioned in the op), and I tried adjusting the gain of op-amps (varying those 22K and 220K in the OP),
first decreasing until the field is too weak to support the object and then increasing slowly until it is supported. I also experimented with that 0.1uF compensation capacitor in discrete steps, 1uF, .1uf, .01uf 0.001 uf etc (since I don't have variable cap), but Only 0.01 turned to give out the best result. Perhaps I should get some variable capacitor.

And Perhaps If I use a sphere I will get ultra stable suspension since it can't swing like pendulum like the bolt!.

here is a video (although it may not illustrate the problem)
YouTube - AntiGravity

 

[DerStrom8]

DerStrom; I believe Shortbus was refering to the one or two electromagnet debate. Andy

Thank you, Andy. If this is the case, I apologize for the misunderstanding
Der Strom