Thanks, Kerim, for pointing out how the Single Pulse Avalanche Energy spec
can eliminate the need for a protection diode across a solenoid. I hadn't
considered that before now.
I'd point out that the energy that needs to be dissipated when the solenoid
is turned off comes not just from the coil inductance, but also comes from
the mechanical energy stored in the spring (or gravity, etc.) that restores
the solenoid core to the inactive position. Some of that mechanical energy
will be returned as electrical energy that adds to the energy in the
inductance. The mechanical energy is average force times distance traveled
by the solenoid core.
Let's try to figure out if external protection diodes are needed if using
an IRL540N with a 24V 1A solenoid valve!
First of all we need to know the inductance. To get an estimate, I measured
the inductance of my lawn sprinkler valves. (The sprinkler controller has a
24V 1.25A power supply, so presumably the valves run from 24V and can draw
up to 1+ A for each sprinkler zone.) The inductance ranged from 90mH to
100mH for the various zones. The valves were not activated when I measured
the inductance. I've noticed, when previously playing with solenoids, that
the inductance increases hugely as the core is moved into the solenoid. It
has to do with the reluctance (which you can think of as "magnetic
resistance") seen by the magnetic field. I don't remember the actual ratio
of inductance for core in / core out, but I think a factor of 10 is not
unreasonable. So maybe these valves have 1H of inductance when activated.
(1/2) * 1H * (1A^2) = 1/2 Joule. Oops, that's already over the Single Pulse
Avalanche Energy spec for the IRL540N which is 310mJ.
What about the mechanical energy? Well, the distance traveled is probably
less than 1 cm, and the force... I don't know, but let's say 1 pound ~= 5
Newtons. So the mechanical energy is only 50 mJ, not so bad.
So, depending on the inductance of Chuck's solenoids in the active state,
the IRL540N may or may not survive the turn-off events, if there is no
Much easier and more worry-free to just put a diode across the solenoid
On Sat, Apr 20, 2013 at 12:57 PM, Kerim F <ahumanbeing2000@...> wrote:
> --- In Electronics_101@yahoogroups.com, "rtstofer" <rstofer@...> wrote:
> > --- In Electronics_101@yahoogroups.com, "Kerim F" <ahumanbeing2000@>
> > > The polarity of the body diode cannot discharge the coil
> (inductive/magnetic) current if one MOSFET is used as a switch. The only
> way by which it can be discharged in this case is in the MOSFET avalanche
> region. This may or may not destroy the MOSFET. It depends on the amount of
> energy discharged and its rate to happen.
> > > The best method to discharge the coil magnetic energy depends on the
> application. The easiest way is to use an external diode in reverse and
> parallel to the coils (lowest drain-source voltage, about Vcc + Vdiode, and
> slowest turn off, L/R time constant is largest).
> > >
> > > Kerim
> > >
> > Here is another factory document that states that the body diode (a
> zener in this case) can handle the freewheeling. In fact the body diode is
> rated at 36A and Vf of 1.3V. The MOSFET is unlikely to be damaged by just
> 1.3V of reverse voltage.
> > The IRL540N itself is rated at 100V 36A - kind of overkill for a 2A
> > See Section 1 here:
> > http://www.irf.com/technical-info/appnotes/an-936.pdf
> > Datasheet:
> > http://www.redrok.com/MOSFET_IRL540N_100V_36A_44mO_Vth2.0_TO-220.pdf
> > See "Source-Drain Ratings And Characteristics"
> > Richard
> You are right Richard. Naturally, if the "Single Pulse Avalanche Energy"
> of the MOSFET is much larger than of the coil (1/2)*L*(I^2) there is no
> need for an external diode. In this case, it is even better if an external
> diode is not added because the turn off time would be shortest which may be
> desirable in some applications. On the other hand, this discharge will
> raise the junction temperature, therefore, there is a maximum rate that it
> could be repeated (this is a topic by itself). But in the application we
> discuss here, the rate is likely much slower.
[Non-text portions of this message have been removed]