closed loop control of solenoid frequency and stroke

I'm taking the liberty of cross-posting this to the controls newsgroup

-- it needs some valid traffic, and this post is spot-on for the group.

This sort of question always boils down to questions like how much money you have to spend, whether you're using custom electronics, what sort of production volumes you're contemplating, how precise it needs to be, how long it can take to achieve precision, what sort of environment it needs to work in, how heavy it can be, etc.

_The_ indicated position sensor for this would be a short-stroke LVDT. It may be hard to get the accuracy you want from it at 100Hz without custom electronics (with custom electronics it's a snap). But LVDTs can be a bit spendy, they tend to be bulky, and most LVDT circuits have a serious bandwidth vs. excitation bleedthrough tradeoff.

Potentiometer feedback is cheap, but it's cheap in both senses of the word -- it won't cost much, but you'll have trouble getting noise-free operation at those speeds.

For that size of a stroke you may be able to get away with an optical sensor -- measure the amount that a transmitter/receiver pair is occluded as the thing travels. This has engineering issues, and therefore high initial cost, but it may be just the ticket for high-volume production. LEDs get funny at temperature extremes, so if you want this to work from Alaska to Algeria you can't just get it working on a lab bench and trust that it'll work everywhere, any time.

There are some linear quadrature encoders that may work -- 2mm is a pretty short stroke, but those things are getting better and better every day. I'd do an A/B comparison between that and an LVDT were I to launch a product doing this. An optical linear encoder suffers from the same environmental issues as the differential optical gizmo I suggested above.

Depending on the solenoid and your preferred engineering expense vs. per piece price, you may be able to do this by using the solenoid itself as a position sensor, as its inductance will change with position. For a one-off this is seriously unwise mad science; for production it makes sense as long as the level of accuracy you needs rivals the production volume: modest accuracy would work with modest production volumes, really high accuracy would require lots of engineering which would only get paid off with high production volumes, etc.

As for control:

Frequency is easy -- just drive the thing at the frequency you want.

Stroke is not too much harder: to a first order, if your solenoid and driver have a high enough bandwidth you can directly servo your position. If you can stand a few cycles of settling you can inject some drive signal through a stage that you adjust with feedback to get the amplitude and phase that you need.

Do you need to regulate the center position as well?

Thanks for the responses. Lots to think about. I have a controls background, but most of my work has been on the algorithmic/academic end of the spectrum so I don't really know much about technology unfortunately.

I guess I can share a few more specifics about what I'd like to accomplish. It's a hand tool for an artist friend, but it has a potentially niche-sized market if it's an decent improvement over what he's using now, so I'm thinking in terms of production.

Minimum requirements:

1) Pretty good frequency control. Any oscillation not noticeable to the user. 2) Very precise control of one end point of the stroke and reasonable control on the other end. 3) Little if any noticeable off-axis vibration.

Really nice things to have that would be probably be necessary for commercial success: A) Low enough power drain to operate off a small battery (3 cm^3 max???) for a decent about of time. Sorry, I don't have more specifics on that yet. B) Production costs for accuator/sensor/electronics of $200 at the very most. Probably needs to be less than that unless this turns out to be so great that people are willing to pay substantially more than the current technology costs.

Other things that would be cool but aren't absolutely necessary: a) Shaping of end effector trajectory. Lower velocity on out stroke then on return stroke. b) Quick adjustment of stroke length.

Force requirements: Pretty small I think. The reciprocating tool is very light, on the order of a couple paper clips maybe and it doesn't have to apply much force to the material you're working on. Force requirements are probably dominated by the friction needed to hold the tool in place.

That's all I can think of right now.

(2) and (a) make me want to use a cam. The potential problems I see with this are that cutting an arbitrary cam profile on something small is not that easy, so (a) may not be that achievable. Expense is also an issue. Small motors are pricey and even a 16 line encoder that small seems to cost as much as the motor. Easily over $100 each total right there which is pushing the limits of what this thing can cost. Accuracy and settling time of speed control is not going to be as good as I would have hoped for, but it's probably acceptable. Power drain? I don't know if a solenoid would be any better, but that voice coil has me intrigued. (1), (3), (a), (b), and maybe (B) and (A) are what got me thinking about a solenoid, provided that (2) could be achieved via some combination of physical mechanism and feedback control.

FYI: engineers don't use democratic words like "big", "little", "better", "worse", "decent", "reasonable", "pretty". Engineers use numbers. Before you have the numbers, there is nothing to discuss.

VLV

I respectfully disagree. In the early stages of a project we most certainly do -- and it's never long before we realize that we need to define them in the context of that project, before all go astray.

In the end those smoky words are the 'real' specifications that you are working towards -- you can hit all the numbers on the nose, but if those numbers don't look decent reasonable and pretty to the customer, if the value of the product you engineer isn't better, then your product will make little money, whoever hired you will thing they could have done worse than to never start things, and everyone will have big problems.

But yea -- you don't want them hanging around for very long at all before you pin them down.

(context restored -- this is USENET)

Brandon wrote: > I have a project in which I would like to produce a "smooth" > reciprocating motion with an adjustable stroke length between about 1 > mm and 2 mm and an adjustable frequency between 100 Hz and 150 Hz. I > think there might be some advantages to using a solenoid for this (as > opposed to a rotary motor and a mechanical design) but only if I can > get good feedback control of the stroke and frequency. In a spring- > mass mass system frequency can be controlled open-loop, but I'm not > sure how well I could regulate the end and start position of each > stroke or even what sort of sensor might be appropriate and have a > fast enough response and a clean enough signal to work at that > frequency. Wondering if anyone has any good electronic (or > mechatronic) ideas as to how this might be done.

Swiss precision motors are pricey; if you can make a Mabuchi or a Johnson motor work reliably they'll come pretty cheap. I think that such a motor, properly used, would last for a good long time. For a motor/cam assembly all you need is a once-around sensor on the motor -- there are a number of cost-effective ways to do this; the two that leap to mind are a magnet and a hall-effect sensor, or a slotted disk and an LED/phototransistor pair.

A speaker coil is probably more efficient by far than a solenoid, and will probably be lighter as well. Better yet, it's something that you can fabricate yourself with a bit of wire and some easily obtained magnets. A motor/cam is going to be a power hog if you really just need to overcome inertia -- but a speaker coil's power requirements go up significantly with load, so if you have to actually _push_ on anything or maintain rigidity, that motor/cam may start looking good.

Of course, if you want to change the motion profile easily, the speaker coil beats the pants off of the motor/cam.

If this is going to be a hand-held gizmo, then an LED/phototransistor position sensing arrangement will have no temperature range troubles -- and such a position sensor is something you can probably womp up fairly cheaply.

You'd have to play with this to find out about power drain, but 25cc is enough volume for a lithium-polymer battery to hold a significant amount of energy.

Spring- mass system witn a resonant mode in the middle of your frequency range, use the reference coil of your lvdt as the driving coil for the solenoid core, the CT sense coil provides position feedback. Drive method of the coil may be a tradeoff. PWM would give the best efficency, but AM signal would probably make the LVDT demodulation easier. Sounds very do-able in the price range mentioned.

• Voice coil designs are rather good in those 3 regards; look at speaker voice coil positioning can be controlled by use of an arbitrary waveshape and hard drive positioning is definitely by arbitrary drive.

Not everything needs to be quantifiable. For a big and expensive project development you'd definitely want to nail down your specifications as much as possible to minimize a costly redesign. For things like communication equipment that have to interface with existing technology, the numbers pretty much define your project. But this hypothetical gizmo is for an artist, so knowing that the force needed is "pretty small" tells me enough. You build a mechanism that can overcome its own inertia by a large enough margin to make you confident that it'll work and then go out and test it. Is it really cost and time effective for me to go out and try to measure this force just to provide a number? I doubt it. I could be wrong, but that's a chance I'll take.

Besides, I'm not asking anyone here to design this thing for me. Just for general ideas. Thanks for everyone's suggestions.