Estimate "hash band" in PWM driven solenoid

[OCPumpGuy]

This is along the same lines as my previous post, but different enough to start a new thread.

I've got a solenoid that is controlled by a PWM driver. I am creating a dynamic model of a valve actuator and I am including a very simple model of the PWM driver and solenoid in the analysis.

I've noticed a very interesting result when comparing my simulation to my actual test results. The solenoid electrical circuit is very simply modeled as 0.15/(0.015s+1). With 500 Hz PWM, I get a "hash band" less than 100 mA. My actual test data shows a "hash band" of around 200 mA. I can adjust my model to match the hash in the test data, but then the response to large step inputs doesn't match (my model becomes faster than reality).

The additional hash in the test data is not from electrical noise. I'm guessing that somehow my simple first order model is too simple???

I read a recent post about hiring a consultant when you don't know what your doing - I really just want an accurate black box for this solenoid that I can include in my mechanical models!! Seems like I should be able to find an academic article or textbook that covers something so simple, suggestions??

 

[Skogsgurra]

I can think of several reasons why you get twice as much "hash" than you expect.

The first is that solenoids, if there is iron in the circuit, produce eddy currents in the iron. That eddy current adds to coil drive current.

The second is that your cable capacitance may be high and draw extra current. You may even have a resonance (or be in the vincinity of resonance) between cable capacitance and coil inductance.

Other possibilities exist. Your parallel diode (if still in circuit) will be conducting more or less continuously if L/R is 150 milliseconds.

Also, if the L/R deviates from assumed value, you obviously get a differing result.

Measurements sometimes are not just that right. There are a few holes in the road to a valid result. Easy to trip.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[OCPumpGuy]

I like 2 of your ideas. Regarding the eddy currents, do you have a reference for how one might model their effects? Also, on the diode, good idea. That is really a key component that needs to be in my model, not sure why I didn't add it in the first place. I assume that this still applies if L/R is 15 ms (you noted 1/R in the numerator by mistake).

In this particular testing, I don't think the cable capacitance is a big deal. The wires are very short, and they don't seem like they could by significant relative to the size of the inductor.

I'll definitely add a diode to my model and see what happens. I'm still interested in finding a good reference for a solenoid model - something that includes EVERYTHING and also comments about what can be neglected when. I'm surprised that I have had such a tough time finding a decent reference. I even spent an evening at the local university library and couldn't find a decent reference.