Driving piezoelectric actuators with charge, not voltage

Winfield Hill wrote in news:q77qlt015h9 @drn.newsguy.com:

To make a move, even a few mils, the piezo stack would have to create enough force to perform the move, and the load against which the move impinges.

That force requires work and that work over a time factor too.

So I doubt seriously that there would even be enough energy for the stack to even move itself, much less impinge motion against a load.

We had an 800 element stack that was about 3.5 to 4 inches long (probably 100mm). The total displacement was 1mm. We amplified that via lever to produce 2mm. We fed it with a 2kW application specific audio frequency range amplifier.

With that custom amp and the amplification mechanism, it was possible to machine optical grade finishes on a lathe.

Hitting it with a charged HV cap would likely make it flex for a millisecond or two. Hitting it with a charged up Layden jar would likely not result in any motion at all and deplete the jar immediately.

The only 'work' an ES charged element can do is to destroy tiny PN junctions in the IC chips on a circuit board. Pretty much good for nothing. We spend a lot of time, money and design abating such events.

If it could perform real work, we would be finding ways to store all or part of lightning strikes to power cities.

Nothing is free. To get work done, you need constant force.

Otherwise the movement event you are seeking will only last a couple of microseconds.

But seriously, I know that you already knew all of this 'stuff'.

The charge creates a voltage change on the piezo, just as if you'd applied it with an amplifier. Yes, I'm sure work is being done in the process, but when the dust settles in a few us, the charge and corresponding voltage remains. The mirror has been moved, but as no more work is required, it stays in place.

Yes, very nice.

The resonant frequency of our 20-mm-long AE0505D16 piezo is 69kHz, probably a bit lower with the small mirror, so its time-constant was fast, several us. Each 6us charge pulse made immediate voltage steps, after which I observed small voltage variations for a few us, followed by a small quickly-disappearing droop. Thereafter it remained in place.

That just screams for a publication in Review of Scientific Instruments...

Steve

Winfield Hill wrote in news: snipped-for-privacy@drn.newsguy.com:

Ok, if the load is tiny (a small mirror) I can see where this will 'work'. There likely is enough juice to snap it from point to point.

Yes, thanks, Steve. So many things to publish, so little time. I'm spending a few days to write it up for our AoE x-Chapters book, under deadline now. I have maybe a few weeks to add more stuff. Anyway, hoping to get some insights from an s.e.d. discussion, plus doing some literature searching.

On a related subject, the charge delivered by a pulse into an LC filter equals:

dQ = (t_on^2 V_in (V_in - V_o)) / (2 L C V_o)

where, t_on = on time (say the high side switch in a buck converter with ideal "catch" diode), V_in = supply voltage V_o = initial output voltage L, C = filter components

Assuming dV_o ~= 0, which may or may not be a good assumption for a given application. (Also assuming C is linear; if not, then use C(V_o), and again assuming small dV_o --> small dC.)

I'm guessing the efficiency of a switched circuit is not nearly as important as precision in your application, but perhaps a similar analysis will help with signal generation and driving, including consideration of parasitics.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

Sounds like the piezo is another junky nonlinear ceramic cap. It's interesting that charge makes displacement.

Another approach would be to dump a capacitor, a known amount of charge, into the piezo with just a switch.

I'm aware of folks using piezo actuators to do serious work, like bending fighter jet wings. But my piezo experience has been confined to the other end of the spectrum, with STM, AFM and optical positioning systems, stick-slip walkers, etc. In these scenes, usually the actuator's body characteristics dominates. Sometimes the resonant frequency is lowered an octave or two.

Or an inductor, flyback style.

There's a classic instrument for charge measurement, the ballistic galvanometer, that moves on a charge-based input. Unlike your stack, it is NOT a well-damped system, and the charge is evaluated by stopwatch-and-notebook observations of the moments after the charge event, usually with a human-operated switch opening to let the galvanometer move freely. I wonder how important damping (electrical or mechanical) is, it will affect the timescale.

An optoSCR (maybe just an optoisolator) would be near-ideal as a switch for this application; the dI/dt limit of SCRs doesn't apply to a gateless optical one, and it can deliver charge from CV source rather than integrating current over time.

I remember the charge-injection method from back when I was a postdoc making atomic force microscopes, circa 1988. (It was the first commercial AFM, made by a succession of companies ending with Veeco.)

Using charge instead of voltage made PZT-5H bimorphs competitive with PZT-5A despite having (iirc) 10x the hysteresis in voltage mode.

Cheers

Phil Hobbs

I've only done voltage drive. (using the smallest of those Nec/Tokin stacks.) So you don't see any leakage current? Is there any spring loading on the pzt?

George H.

I haven't observed leakage, but I have observed what many call creep. They may measure it with position sensors, but you can see it with the voltage changing after the delivered charge. It's similar to capacitor dielectric absorption. Things stabilize after that's over. I took linearization measurements after 2 ms. For most of the literature I've read in the last few days, showing perfect charge linearity, measurements were made at very low frequencies.

No spring loading, mounted at one end, and a tiny mirror cemented on the other end.

Huh OK, I've never used it like that. I wonder what the mirror movement looks like? I did the design* (diode laser grating drive.) years ago. But I remember reading it is good to pre-load it (pzt) the right amount. There's a mechanical resonance somewhere. Best to find it, before it finds you. :^)

George H.