Active vibration cancellation? - up to ~500Hz

[max_torque]

One of my customers has a motor test rig that has a bit of a vibration issue, because it's not stiff enough, and certain frequencies, the fundamental "once-per-rev" vibration sets up a bad ringing in the rig.

There is the option of adding physical mass and stiffness, but i did wonder about whether an "active" cancellation scheme would be possible?

The frequency is up to 500Hz, so not that fast in the grand scheme of things, and a large audio amp driving a "shaker" type subwoofer (probably need to be mass tuned) and somehow driven anti-phase to the main rotating assembly could work as the actuator.  Anti-phase could be determined through the output of a vibration sensor, or i guess from a simple once-per-rev type sensor, with a calibratable offset as the position of the imbalance shouldn't' change once set up?

Anyone done anything similar?

[ruffy91]

What about doing the same as with a car motor? Make the mount softer so that the resonance is at much lower rpm.

[OM222O]

instead of "active cancellation" easiest solution would be the one used in CNC milling operations: varying the spindle speed by +- let's say 10% so no one frequency has enough time to build up any significant oscillations due to resonance. what's the application (i.e: does it allow for variation in speed or not) and how is the motor speed controlled? (if it's controlled with a digital system, then it's easy to code the rpm change cycle)

[ogden]

If you can find commercial active dampening system that meets the specs and customer is ready to hugely overspend (compared to passive system) - fine. Otherwise just consult noise & vibration analysis company/engineer. Those working in construction business most likely will be able to help or will point you in right direction.

[soldar]

Why is this vibration happening? Unbalanced mass? Because maybe mechanical design could balance the rotationg masses.

Passive measures like adding mass, rubber standoffs, dampers, etc. is going to be more practical, simpler and cheaper.  My clothes washer weighs a ton and most of it is a large piece of concrete attached to the drum.

Now, if you want to go all fancy you can approach this as a feedback design. Have three sensors in three axes which measure deviation and three actuators which can be solenoids or hydraulic.

[max_torque]

As suggested the cheapest simplest route is to just add mass and stiffness, and i already have the kit (3 axis accelerometers) necessary to characterise the vibration etc.

But i was interesting if a more fun solution would be possible!   I cannot control the frequency of the vibration, as that is set by the exact test being carried out on the test motors being testing on the test rig, and by coming up with some sort of active solution then the rig would be capable of stabilising itself across a wider range of frequencies.  I'd only need a 2 axis solution to deal with the primary rotational out of balance forces that are the forcing function in this case.

Any active actuator would need to be able to control it's load in both amplitude and frequency, hence the idea of using modified base kickers ie like this:




They are tuned to frequencies between 50 and 200 Hz normally, but modifying the mass that oscilates would allow a higher frequency (at the same power level, ie less amplitude) and powerfull amplifiers to drive them are not that expensive these days.  The control logic would have to create a pair of Sine waves in quadrature, to be fed into the L&R channels of an amp that drives two of these, mounted orthogonally on the rig, as close to the excitation axes as possible.

Those sine waves would have to be phase lock looped to the fundamental frequency of revolution, and phase aligned to be anti-phase.  At the frequencies we are talking about, up to 500Hz, that doesn;t sound very difficult and could probably be done in software by a small ARM, perhaps even use the ADC to read back the accelerometers and try to close the loop on the vibration magnitude!

Yes, much more complicated that bolting on some mass, but a lot more fun!

[T3sl4co1l]

What about doing the same as with a car motor? Make the mount softer so that the resonance is at much lower rpm.

This is the germ of a good idea (in terms of passive methods) but there are two other variables at play, and varying just one is as likely to hinder as help!

Namely, mass and damping.  We can draw an equivalent circuit with inductance, capacitance and resistance in terms of spring constant, mass and dashpot.  When the ratio sqrt(L/C) = R, it's critically damped.  Moreover, the amplitude of displacement (amount it shakes) will be proportional to that times the applied force (i.e., the mechanical impedance, and Ohm's law).

So, just making the spring weaker, lowers the frequency, but also increases the amplitude at that frequency.  The force may be lower too (we're dealing with an unbalanced load, so the force is proportional to the unbalanced mass times frequency), which would be a wash.  Essentially the damping is a ratio of masses (unbalanced to sprung mass), so the only thing you can really do is increase mass.

Damping is a little harder to do -- in mechanics, we are flush with reactances (masses and springs), and resistance is kind of hard to come by.  Typically we have lossy materials and viscous fluids that mix a bit of both.

Examples:
- Rubber shock dampers: really more springy than lossy, but at least having a fairly low Q -- much like ferrite beads.
- Acoustic absorbers, damping mats, etc.: dissipates free waves, or flexing or stretching energy, in a dense yet porous matrix that has significant viscosity or friction (squishy rubber, fibrous materials), or traps a fluid which does (the gas inside open-cell foam or fibrous materials).
- Dashpots, shocks: viscous loss in a fluid, usually a restricted range of motion (like linear or rotating motion, not free motion or wave energy).

The other drawback is, the range of densities and elasticities isn't all that great, so we can rarely absorb/reflect/filter vibration in a single stage, or even two stages; a stack of alternating high-density and high-elastic materials is needed to act as a lowpass filter (a distributed one at that, because it's very easy to get into acoustic modes -- a consequence of the relatively low speed of sound in materials).

Which is to say: if you can't dampen the oscillation by resistance, or brute-force mass, you may have to let it be, and add springs to isolate that motion from the rest of the rig, which in turn has some mass which the springs work against; and the whole suspension needs to be well damped so it doesn't get super bouncy at some (even lower) frequency.

As for active means -- anywhere you can attach a transducer, you can apply cancellation of some sort.  But doing it at such high frequencies may not be easy.  There are many degrees of freedom.  Just for an unbalanced load: you could use a subwoofer / mass driver to cancel linear motion in one direction -- but you need a pair to oppose both axes.  And that's if the rotational axis is fixed and known.  You need three in general, but even worse than that, you probably need six (in balanced pairs) to ensure you aren't making a rotating moment instead of a linear displacement!  And now you have a six channel controller with interactions between all channels and some kind of transfer function or control program to run it, and... yeah, it quickly gets complicated. 

Personally, I prefer to stick to electrical circuits, where I can place a (one dimensional) transmission line, more-or-less wherever I please, as opposed to a dozen propagation modes in every solid...

Tim

[OM222O]

One of my customers has a motor test rig that has a bit of a vibration issue

Yes, much more complicated that bolting on some mass, but a lot more fun!

I feel bad for the customer ... you're supposed to solve their problem for the lowest price, not trying to over engineer a simple issue just because it's fun!  you do realize they don't care about how much fun you have in the process of solving their issue?