Strange current profile of a solenoid

[dexym]

I've been trying to figure out how my 2T bike controls the oil pump and I've noticed something that I can't understand. The pump is a simple solenoid that as far as I can tell is driven low side.

I've attached the ina219 sensor on the high side to measure the current as I don't have access to the current probe.
I get a signal with a strange "bump" that is right after the plunger "bump". This was not what I was expecting to see. When I drive the same solenoid pump with a simple low side mosfet with arduino I get the signal that I expect to see, where I have two "bumps" that I can correlate with the plunger movement. I also recorded voltage signal with mini scope and the voltage signal also looks normal.

Is there a different way to drive a solenoid that can explain this current curve or is there something defective with the way the bikes ecu drives the pump?

[T3sl4co1l]

What V, I scales?  So the first one is on the bike, second one is Arduino?  Can you show your circuit for it?

I would have to see exactly what all is inside the solenoid, but the bumps are likely where the armature hits end stops.  It's possible it's hitting, and bouncing just once.

Tim

[dexym]

So the first one is on the bike, second one is Arduino?

Correct.

You can see scoped voltage profile of how the bike drives the solenoid in the last picture of the previous post. This is how I expect it to be. I didn't record voltage signal of arduino driving the solenoid but it is nothing out of the ordinary. I can do that also when I have more time. Max current in both cases is about 1A, and the duration of the signal in both cases is 100ms. When I connect the solenoid directly to battery I get the same signal that I get when arduino drives it.

With arduino I use the same 12V battery that the bike uses. When tested with the bike ecu the engine is not running.

I've attached the arduino circuit.

[coppercone2]

solenoids often have a trigger and hold voltage range, if you are not familiar with it the circuit can be confusing, it is made to reduce how hot it is after it is actuated. That is another solenoid trap.

[T3sl4co1l]

Ah.  Okay, what you're seeing in the first waveform is, the voltage is fixed by transistor (pulling down), so the current varies due to dynamics.  You see the armature bounce or whatever it is, and for whatever reason, it apparently bounces exactly twice.  (Maybe this depends on what's in, or connected to, it?  Maybe it varies from shot to shot?)  At turn-off, there is no clamp diode* holding coil current, so current falls rapidly.  In the third waveform, we see the reflection of this: voltage is allowed to freewheel quite high, and voltage varies due to dynamics.  We see one bounce.

In the second waveform, voltage is fixed during both phases, and so the bounce is visible in the current flow, in both cases.  Evidently, this time it only bounces once, even though the voltage, current and rate seem to be the same.

*It would be quite irresponsible to have no clamping whatsoever; what's normally done is, either a higher voltage TVS clamping the excess voltage to ground, at a voltage somewhat below the breakdown rating of the switch; or a protected switch, which turns on the switch a little bit if voltage goes too high -- serving the same purpose, clamping excess voltage to ground.  As you can see, this only lasts a few ms, or fractions of a ms, it's not a huge amount of energy.

MOSFETs apparently wear due to breakdown, even though you'll find some rated for one-time or even repetitive avalanche.  The current flow mechanism is fundamentally different between breakdown (Vgs off, Vds > Vds(max)) and conduction (Vgs > Vgs(th), Vds < Vds(max)).  The latter is reliable in continuous operation (granted that thermal and stability limits are observed), the former is not.

The difference to the diode clamping method is, fall time is as fast as can be.  This affects mechanical speed to some extent, which is most exaggerated on very small and fast actuators, like fuel injectors.  Consider: if the armature only takes 5ms to travel, but it takes 10ms for the current to discharge, is it really going as fast as it can?  Of course not.  So, it can be a good idea to use a zener/TVS or the like.

It's also cheaper, when using protected switches which are also protected against current and temperature; so they use a lot of those in automotive applications.  At least, I'm guessing that's what's in the ECU.

Tim