switching 11 250mA shape memory valves from arduino

Question

I am trying to switch 11 shape memory valves (SMV), which are effectively valves that open by heating a (shape memory) wire by passing a current through it. The SMV require 250mA, and have a resistance of 3.2Ohms when hot(open) and 4.2Ohms when cold (closed). I'd like to switch these with an Arduino. I can use a MOSFET or PNP transistor to allow the arduino DIO to switch them, but this would require placing a high power resistor in series for each valve that, on a 5V power supply, will waste 2.3W (x11 valves...). I'd like to avoid this but I am not sure how to go about this. I played around a bit with a PNP resistor in LTspice and circuitlab which resulted in the below circuit. it reliable shows 250mA to the SMV if the arduino 5v out is modelled as a battery supply, but I suspect I am making some mistake here, also because the calculated current for the smv is different in LTspice from Circuitlab. Anyone have any ideas how to sort this out? Would a PWM output from arduino be a solution?

Thanks!

^{simulate this circuit – Schematic created using CircuitLab}

Answer 1

It won't work, the power will be dissipated in the transistor and it will burn up. It's also too sensitive to transistor hFE to be a practical circuit, which is probably why you are seeing different results. You can use LTspice to show the power dissipation of the transistor.

You could use a PWM signal and gate it with the control for the individual valve. If a voltage on/off is okay then you only need a single PWM duty cycle. I'm going to assume from here on that a fixed voltage (and not a constant current) is all that is required.

At 0.9V RMS you get the equivalent of 280mA at 3.2 ohms and 215mA at 4.2 ohms, but the spikes from the power supply will be as much as 1.6A per valve, which is pretty significant- especially if 11 were powered at once- requiring careful bypassing. (BTW, I think you've got hot and cold resistances reversed, but that's not critical here).

The duty cycle for 0.9V RMS from 5V would be 3.24%. That's going to be pretty nasty and electrically noisy- fast 18A spikes on the 5V rail, and I don't really think it's to be recommended.

A cleaner method might be to use a small SOT-23 switching regulator for each valve, such as a TLV62568 and control the enable pin with your MCU. Efficiency will probably be in the 80% range so each regulator will draw only about 63mA from the 5V rail when on (and virtually nothing when off), so all 11 are only 700mA. That particular regulator can be used down to 600mV output.