Back EMF and snubber
Electrical engineering students are taught about magnetics and that energy is stored in an inductor when it is energized. This can be presented in a theoretical sense and may be applied to topics like switch mode power supplies. What does not get a great deal of attention is the fact that there are inductors embedded in many devices and that the rules of energy and energy dissipation still apply there. The real world is bristling with relays and it is quite likely that the nascent engineer will be faced with activating some. Success! It turned on just fine, but wait a minute… when it turned off the micro went nuts. What happened?
Well the relay, like the solenoid, the contactor, the motor, and who knows what else are inductors (aka coils) and they store energy when energized. But what happens to that energy when the coil is de-energized? The coil produces a back EMF (electromotive force) that is given by V= -L(di/dt). First of all, note that it is negative and secondly early in the process dt is small, so V will be a large number. Translated this means that you get a voltage spike at the coil terminal that is being driven by the electronics. Two things can happen: The voltage may be high enough to damage or destroy the electronic driver; and the spike may radiate and upset the operation of the micro (and other electronics) running in the vicinity. In extreme circumstances it may even cause a hard fault. It may also generate enough emissions to cause an RF emissions certification problem.
Since getting rid of all the inductors is not possible, we have to suppress this back EMF. The most common technique with DC switching is to use a diode in parallel with the relay coil (aka freewheeling diode) as shown in Figure 1a. When the load is energized the diode is back biased and has no effect. When the load is de-energized, the diode becomes forward biased. The stored energy is dissipated back through the coil and the voltage at the driver is clamped to 0.7V above the relay supply voltage. The diode should be suitably specified based on the following. It must have a PIV (peak inverse voltage) equal to the supply voltage. The current through the coil cannot change instantaneously so that the current flowing through the coil when activated is shunted through the diode on de-activation and then decays, so the current flowing through the coil is the maximum current expected to pass through the diode. The freewheeling diode technique is so common that you can buy relays with the diode embedded (did you ever wonder why the pins to a coil were polarised?), or semiconductor drivers like the ULN2803 that have the diode integrated onto the silicon.
However the freewheeling diode is not a panacea. In some cases, depending on the load, the flow of current in the reverse direction may keep the coil activated until the energy has sufficiently dissipated. Let’s assume that the solenoid was controlling a turnstile. Access to the turnstile release may be delayed for a several hundred milliseconds and when that happens at my gym it is quite irritating when you bump into that locked gate. I have seen the diode replaced with MOV (Metal Oxide Varistor) or a bipolar TVS (Transient Voltage Suppressor) but I am not fond of that since both types of devices seem to age over multiple breakdowns and finally stop working. In the case of the MOV it shorts across its leads when the back EMF exceeds its breakdown voltage, so it is no better than the diode. The TVS keeps a fixed voltage on breakdown and the time the current flows for is reduced.
It is also possible that the freewheeling diode can damage the relay. The app note “Coil Suppression Can Reduce Relay Life” from TE says that a diode and zener in series work better than the diode alone.
I have also seen MOVs, TVSs and fast Zener diodes used in the configuration shown in Figure 1b. My reservations on the aging of the types of devices hold here as well.
Figure 1. Different Back EMF suppression techniques for DC. (Source: Author)
Until now this was all about DC powered coils. What about AC? Well, there is a technique that works for both AC and DC and it basically involves connecting a resistor and capacitor across the inductor as shown in Figure 2a. This combination is known as a “snubber”. It is not unusual to find a snubber connected across the energizing switch as in Figure 2b. This has the advantage that it reduces arcing across the switch and since it is more likely to be on a PCB, it is easy to add the resistor and capacitor to the layout. In some cases with long wires it may be necessary to use both circuits.
However there is one drawback with the snubber across the switch - whenever there is a capacitor there is an AC path. Aside from the fact that there is a voltage that may surprise the unsuspecting engineer or service technician, there will be a leakage current across the switch and in some cases it may even activate the load. One counter-measure is to add a bleed resistor or capacitor across the load to sink the leakage.
Figure 2. Snubber connections to suppress back EMF. (Source: Author)
When the coil is de-energized, the circuit is in a damped LCR configuration. Since the coil and capacitor are reactive, most of the energy is dissipated as heat across the resistor. In the real world there will be some heat generated in all the components. But of course this begs the question: what values do you use? In the field you will find that panel makers and other proponents of the industry have rules of thumb and will stick to those values, no matter what different experts say. (These are the suspiciously convenient 120R and 0.1uF.) Given that there are so many variables it may be impractical to design for every specific case. In researching this topic, I found that there is even a book on the topic “Snubber Circuits for Power Electronics” by Rudy Severns although I suspect it is aimed more at the switch-mode power supply designer. However the best guide to value calculations I found is an Application Guide for Snubber Capacitors from Cornell Dubilier.
You can of course take a short cut and simply purchase commercially available coil suppressors, but in all events don’t give yourself an unpleasant surprise next time you try to control something in the real world. Plan for back EMF!
Embedded Systems Design using the Rabbit 3000 Microprocessor: Interfacing, Networking, and Application Development by Kamal Hyder and Bob Perrin.
The application of relay coil suppression with DC relays: Application Note from TE Relay products.