What (and how!) can I deduce about the operation of a particular (e.g., previously empirically characterized) solenoid by observing it's electrical characteristics UNDER OPERATION?
E.g., I can tell if the coil is *open*, obviously enough. Likewise, I can tell if it has been shorted.
[Of course, these assume you can resolve the differences with whatever detection means you employ]
But, can I determine if, for example, the actuator has traveled it's normal length (without encountering some mechanical interference)?
On a related note, how much control (in the hammer driver) can I exert over the actual *motion* of the actuator? E.g., "slow" the motion vs. speed it up? What factors are likely to complicate this, over time?
--
John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Inductance increases as it closes, so with ac drive or a small ac component you can measure v,i and deduce L. You can measure its cold R & present R & work out how hot it is. If you drive it with high enough f you can deduce where it is in its travel from L. A suitable controller can then give you complete control over its position - within its mechanical and control loop limitations at least. You can speed it up by driving it with a whacking great spike or overvoltage - ratings are for continuous operation and there's masses of scope to exceed them briefly. By monitoring spikes you could also tell if the sometimes included diode is faulty, fwiw. You could monitor switch contact waveforms to detect deterioration & predict failure. I don't know whether you could also initiate a switch cleaning cycle with the load otherwise disconnected. And finally you could simultaneously use it as a radio receiver for a suitably low frequency and a moving iron speaker driver coil.
Haven't done this with solenoid, but with stepper motor control to get much of what you asked for. Don't know how far you want to take this, but at the terminals of the coil you have a 'spectrum of sensing vs frequency', ie. the effective reactance [inductance], the effective resistance, AND a bit of voltage generated by the motion. Think generator, or motor kicking back, that type of voltage generation.
We used to 'quick step' AND 'slow step' a stepper motor by doing what you asked.
Over time? don't know, but far more problems with variations caused by temperature. THAT can induce hysteresis and a LOT of non-linearity into your monitoring/control. From memory, doing this reduced power requirements and we could get the SM to run way faster than spec. ...without slipping a cog.
You need to know how much a normal stroke takes -- but of course, you need to calibrate for any of the things listed -- open/short, or if you wanted to know coil temperature (copper resistance, tempco), so I'm going to take that as a given.
A solenoid is reluctance driven, so its inductance goes up as it closes. If you drive with a constant current, you will observe a positive hump in the terminal voltage, with "area under the curve" (aka flux) corresponding to the actuation travel (not proportional, because all sorts of near field rules apply, but you could calibrate the flux vs. travel curve if you really wanted).
Mind to subtract I * DCR first, so your flux integrator isn't drifting constantly. :)
It'll be harder to do from a constant voltage supply, but current will dip by the corresponding amount; the result is, instead of V = I * DCR, I drops and EMF rises momentarily (or vice versa, since this all works in reverse, too).
I suppose the limits of detection, as far as indicating position goes, will be on magnetic hysteresis, since they probably use crappy mild steel in these things, not like, high grade nickel-iron...
Tim
--
Seven Transistor Labs
Electrical Engineering Consultation
Website: http://seventransistorlabs.com
"Don Y" wrote in message
news:mcu68a$909$1@speranza.aioe.org...
> Hi,
>
> What (and how!) can I deduce about the operation of a
> particular (e.g., previously empirically characterized)
> solenoid by observing it's electrical characteristics
> UNDER OPERATION?
>
> E.g., I can tell if the coil is *open*, obviously enough.
> Likewise, I can tell if it has been shorted.
>
> [Of course, these assume you can resolve the differences
> with whatever detection means you employ]
>
> But, can I determine if, for example, the actuator has
> traveled it's normal length (without encountering some
> mechanical interference)?
>
> On a related note, how much control (in the hammer driver)
> can I exert over the actual *motion* of the actuator?
> E.g., "slow" the motion vs. speed it up? What factors
> are likely to complicate this, over time?
>
> Thx!
Open/short are easy to detect (for a particular "class" of coil). Beyond that, I think there is too much "finesse" required -- esp if you have environmental issues that factor into the observations.
But what if it encounters an obstacle in the way? Something that prevents it from "continuing its travel" -- yet still results in power being consumed, etc.?
I can accept a result that tells me "yes, the actuator made its full stroke -- even if it was sluggish or impeded along the way" vs. "no, the actuator got hung up mid-stroke". And, handle open/short elsewhere.
But, I suspect -- short of adding a limit switch or other physical detector -- getting any sort of reliable result will be tenuous?
L changes with position. R changes too but is a separate property, measurable separately. I'm not clear why that would make a measured L value tenuous.
But, you're having to do that "quickly" -- i.e., the stroke takes a small fraction of a second to complete.
I suspect that would be hard to do reliably over large temperature ranges (e.g., 100F). I would be happy with a *static* appraisal of its "final position" (actuated and released). The goal here is to determine when it (or the mechanism with which it interacts) has failed (fully or partially). Doing so without "mechanically" instrumenting it.
Yes, one of the firms where I worked used back emf as a means of sensing when/whether the motor had physically taken its step. Allowed us to use steppers as DC servos without requiring an encoder.
Exactly. That's a ~100F range of temperatures *before* factoring in any self-heating effects. I'd be leary of any conclusions beyond "failed open" or "failed shorted" in that environment.
OTOH, a *relative* measurement could be useful. Note characteristics over time and watch for "changes".
E.g., yesterday, it behaved thusly. Today, it is behaving...
If you're trying to watch/control the *motion* of the actuator, it pays to be able to react faster than *it* does! (last part of my initial post)
Maybe *you* can?
Two "identical" actuators:
Unit 1
------
Time Inductance (H) Q-Factor R (ohms)
----------------------------------------------------- Ambient 80.8m 1.73 293 (S) 107.8m 1.73 391 (P)
6 Hr 76.1m 1.73 276 (S) 101.1m 1.73 367 (P)
12 hr 75.8m 1.74 273 (S) 100.8m 1.74 364 (P) Ambient 80.8m 1.73 293 (S) 107.6m 1.73 390 (P)
Unit 2 (held at ambient throughout)
------
Time Inductance (H) Q-Factor R (ohms)
---------------------------------------------------- Ambient 79.7m 1.69 296 (S) 107.4m 1.69 399 (P)
12 hr 78.6m 1.65 299 (S) 107.4m 1.65 409 (P) Ambient 78.6m 1.65 299 (S) 107.4m 1.65 409 (P)
R is manually calculated. S/P are series and parallel equivalent reports from the bridge. Excitation 1KHz.
After initial readings, unit 1 was placed at Ambient-40F (refrigerator) for 12 hours. This should represent the change experienced over the course of a day -- but less than half the temperature differential expected to be encountered over the course of a year! Measurement at
6 hours was followed by another at 12 hours to get an idea as to whether or not the unit had reached thermal equilibrium.
At 12 hours, unit 1 returned to ambient. Final "ambient" readings at
24 hours.
Unit 2 remained at ambient throughout.
Note both units appear (at ambient) similar -- manufacturing tolerances. And, unit 2's readings remain consistent, over time (i.e., the instrument is not likely introducing the differences reported by unit 1)
Note, also, that unit 1 returns to roughly the same readings after temperature is restored to ambient.
Yet, while chilled, there was ~5-6% difference in measured inductance for unit 1. Q factor remained essentially the same.
Unfortunately, I can't manually access the plunger/core in these particular units to see what the bridge reports with the core in different positions. As such, I can't say how *that* change will compare with the temperature induced change, above.
Remember, this isn't an "inductor" but, rather, a "solenoid" (i.e., subject line of post). I'll repeat the experiment when I have access to a higher "ambient" (e.g., 100+F) so I can see what a 100 degree swing represents.
I'll also see if I can "isolate" some of the other types of actuators so they can be subjected to the lower temperatures. And, if I can manually manipulate their core positions.
Its great to see some physical data. But what we need to see is figures with slug out & in, to draw useful conclusions. If I get some time later I'll see if I can find some solenoids to test, they're not something I normally stock.
The permeability of iron and ferromagnets in general is temperature dependent. As the iron or whatever gets hotter, you get less magnetism per unit of magnetising current.
It's tolerably easy to measure the inductance of the drive coil by driving a small AC current through it at a known frequency and measuring the out-of-phase voltage at that frequency induced across the coil.
The catch is that if the moving actuator is electrically conductive, it will act as a shorted turn and kill the inductance of the coil (or a fair bit of it).
Measurements at a coupe of frequencies might allow you to separate out the shorted-turn effect and the magnetic core effect. It would be messy, but might work.
The point of my numbers was to show that temperature *does* have an effect. Of course, with enough measurement precision and "smarts", I can tolerate large tolerances relatively easily. But, only if I can "see" things in sufficient detail and at sufficient rates.
[E.g., I don't have to be able to deal with 100F temperature change over the course of an hour. Or even a *week*. I can track performance incrementally and compensate for (daily/seasonal) temperature variations in the observations]
It's not a question of how "some solenoids" behave. I need a scheme that I can apply to *any* solenoid/mechanism. E.g., some actuators retain the plunger entirely within the coil when actuated/released.
I've looked at some of the other actuators and been disappointed to see that most of them make the plunger/armature difficult to manually manipulate. Or, require the actuator to be removed from the mechanism to force it into any particular position.
So, I'll have to fabricate a drive circuit and take dynamic measurements instead of just relying on static analysis (e.g., with a bridge).
The question really is whether slug in v slug out makes more difference to L than temp variations. If it does, life's easy. If not one would need to check temp by measuring R, then use the relevant L figures to see if the core is in/out/partway.
I never thought I'd see the day a cpu was used in driving a solenoid, but I guess that's where its at now.
Driving the solenoid is easy! The hard part is figuring out if it actually moved AS INTENDED is the tough part -- without adding mechanical instrumentation to detect that! (and, thereby placing other constraints on the mechanism: "You can only use one of these 'approved' actuators...")
"Did the door *actually* lock when I commanded it? Or, do I have to go back and physically verify that? What if it is remotely located? Do I dispatch someone to check it for me??"
Magnetic materials' behaviour can be complex. Depending on composition, mechanical and thermal history, the permeability vs. temperature can have all sorts of shapes.
or so, which for that core would almost double the permeability, providing almost an extra octave of bandwidth. I dropped the idea, which was just as well.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.