"Pick and hold" circuit

[lloydsp]

Modern small form-factor solenoid air valves (even name brands) seem not to be designed for better than 50% duty cycles.  They all tend to overheat, if held 'actuated' for longer than eight or ten minutes.  It's worse if ventilation is limited or ambient temperatures are high (say, "Florida").

Because the vendors cannot predict what will be the 'quiescent' state of a machine vs. the 'off' state, this just isn't right.  The coils won't last at elevated temperatures.  But there's a way to solve it simply.

It's called a 'pick and hold' circuit.  Basically, it's just an RC network that charges a capacitor large enough to 'pick' the solenoid, but then limits the applied voltage to a 'hold voltage' of (say) half of the solenoid's pick voltage rating.  All it requires is a careful selection of the limiting resistor's value and dissipation rating and the capacitor value, and a slow-enough cycle time so that the cap' has a chance to recharge between 'picks'.

For the ones I'm employing, their coil resistance is about 50 ohms, so a 56 ohm, 2-watt limit resistor with a 1000uf cap works fine.

I could do it with PWM with SSRs, but in this machine the solenoids were controlled by mechanical relays, and PWM'ing them would wear the relays out in a day or two, so the RC networks are the solution.  Simple stuff -- but you wish you didn't have to do it!

L

[Bob La Londe]

I'm not so sure about that Lloyd.  I have 3 machines with air seals.  In order to not waste air, or operate a manual valve (or worse forget) I installed DC solenoid valves with a closed shelf state on all three machines.  They are on/open for hours.  In rare instances days.  The valve also provides air for tool clearing blast and quick tool change pneumatic cylinders inside the spindle cartridge.  When the machine powers up the valve opens providing air supply to the machine for hours.  I think the brand I used is Echo or maybe Ehco.  I have a couple spares in the cabinet since I have machine build plans for more stuff using air seals.  In the years I have been running them I may have lost one valve (but I'm not sure), and I have replaced one power supply for a valve.  When I finish running the machine I like to let it set idle for 5-10 minutes to cool down (and not draw in coolant) before cutting power. 

I think a high cycle rate might be more of an issue than a long activation time. 

ASIDE: One of the machines has a manual (push pull knob) 5 port 4 way valve for tool changes, and the other has an electrically (arcade game button) operated 5 port 4 way spool valve for tool changes. Both are reliable, but are activated for only a fraction of a minute to facilitate a quick tool change. 

Some form of arc damping is desired for coils energized by dry contacts or relays.  A diode or an MOV to protect the contacts.  I learned that ages ago installing electric door strikes.

[lloydsp]

I'm well-aware of arc-suppression requirements, Bob.  I always put a heavy Zener diode across the coil/relay contacts (observing polarity, of course!).

I have some 'authentic' Norgren valves that won't stay 'on' for more than 30 minutes without showing overheating in this 96F ambient temperature.  Those are about the primo of the primo, when it comes to solenoid valves.

A dead-simple pick-and-hold circuit solves ALL of those difficulties, regardless of brand.

Lloyd

[Dragonfly]

It's called a 'pick and hold' circuit.

I've found that this method is widely used in cars for various solenoids. But for holding lower voltage (smaller current) they implement PWM. And there is always a diode in reverse somewhere in the circuitry to protect from back EMF from the coil.

[lloydsp]

Yep, and since I'm building and writing a new controller for the thing, I'd have preferred to use PWM.  But you can't PWM a mechanical relay, and expect it to last more than a few hours, regardless of what you do about arc-suppression.  The mechanical parts just won't take it, and most won't cycle fast enough to keep a solenoid 'held' at such a low frequency. (Reed relays could, but these are armature-types.)

So, in this case, I had to employ the RC networks.  If I were designing the whole system from the ground up, I'd be using SSRs, and I most-certainly would do PWM controls on them.  That's just a 'natural' in the GPIO libraries for an RPi!

Lloyd