[SOLVED] A question about MOSFETS and flyback Diodes

Hi,

Just wondering about the real need of Flyback diodes when driving inductive loads (as a DC motor, in a typical bi-directional H bridge). As far as (I thought) I understood about this, it is convenient to place in anti-parallel freewheeling diode across a MOSFET to clamp the otherwise high voltage spikes that would result at MOSFET turn off. For this I know a fast switching diode is a must if you are using PWM to control the effective voltage "seen" by the motor. An example of such a DIODE might be this one **broken link removed**.

But, taking a closer look at the specs of that diode for the reverse recovery time (as I understand it, the time it needs to start conducting under the specified conditions and so, the time the voltage would rise once the MOSFET is off until it is effectively clamped by the diode) and comparing it with the reverse recovery time of the intrinsic diode of this example of MOSFET: http://www.irf.com/product-info/datasheets/data/irf3707.pdf it seems there is not really much difference, is it?. I mean, ignoring the fact that the MOSFET will continue to dissipate heat at turn off through the body diode, would it be able to safely survive the voltage spikes?. Or am I missing something here?.

The reason I ask this is because looking through a post about electric bikes controllers (take a look at the pictures of the controller **broken link removed**) I realized those are really not using any freewheeling diodes! (at least none I can see).

Kind Regards

Reverse recovery is the turnoff time. Worst case you would
have to remove Qrr in that Trr which could give you a rough
idea of current, voltage, Joule energy to be tolerated. It is
likely a trivial charge-slug, that diode charge, as far as the
FET is concerned.

Turnon time is much less, always. Except for a Schottky which
has no minority carrier charge storage to speak of.

The diode shunts the larger problem, which is the stored
inductor current. Push that 20A down an "off" FET at its
rated BVdss, and you're talking some serious heat spike
and probably a bad smell in the lab.

Some small portion of power FETs are avalanche rated,
at all. A diode in reverse breakdown tends to not conduct
as uniformly as one in forward mode, there are usually
hot spots (i.e. first to fail) as well as the raw dissipation
being 100X or more, that of the forward diode.

You would prefer to return or recycle current losslessly,
or nearly so.

Just because something appears on the Internet, doesn't
mean it is a good, or even complete, idea.

dick_freebird,

OP mentioned that this is a bi-directional H bridge, i.e., it's a full bridge. Therefore, the inductive energy will always be clamped by a forward-biased diode, not a reverse-biased diode in avalanche. His question seems to be whether placing a discrete flyback diode in parallel with the MOSFETs is necessary.

gonespa,

MOSFETs are generally not designed with much thought invested into making their body diodes good. Therefore, the body diode will probably exhibit greater on-resistance and perhaps more Qrr (for a given voltage standoff, current rating, Vf, etc). So if the body diode is called on to conduct the entire flyback current rather than a specifically engineered diode, the power dissipation will probably be greater.

However, it is not difficult to turn the MOSFET on during the period its body diode would be conducting the flyback current. The body diode would then only conduct during the dead time (the short time during which both Top and Bottom MOSFETs are off, programmed into the controller to avoid shoot-through current). Then, the overall efficiency would be maintained or improved, and you wouldn't need to buy additional flyback diodes.

ZekeR said:

dick_freebird,

OP mentioned that this is a bi-directional H bridge, i.e., it's a full bridge. Therefore, the inductive energy will always be clamped by a forward-biased diode, not a reverse-biased diode in avalanche. His question seems to be whether placing a discrete flyback diode in parallel with the MOSFETs is necessary.

gonespa,

MOSFETs are generally not designed with much thought invested into making their body diodes good. Therefore, the body diode will probably exhibit greater on-resistance and perhaps more Qrr (for a given voltage standoff, current rating, Vf, etc). So if the body diode is called on to conduct the entire flyback current rather than a specifically engineered diode, the power dissipation will probably be greater.

However, it is not difficult to turn the MOSFET on during the period its body diode would be conducting the flyback current. The body diode would then only conduct during the dead time (the short time during which both Top and Bottom MOSFETs are off, programmed into the controller to avoid shoot-through current). Then, the overall efficiency would be maintained or improved, and you wouldn't need to buy additional flyback diodes.

Click to expand...

I see. I assume those controler examples are using active/synchronous rectification so they don´t really need the power diodes.

Kind Regards.

MOSFETs are generally not designed with much thought invested into making their body diodes good. Therefore, the body diode will probably exhibit greater on-resistance and perhaps more Qrr (for a given voltage standoff, current rating, Vf, etc). So if the body diode is called on to conduct the entire flyback current rather than a specifically engineered diode, the power dissipation will probably be greater.

Click to expand...

I agree in so far that many MOSFETs aren't designed with fast body diodes. Saying this doesn't matter in H-bridge design would ignore some serious problems however. Unfortunately you can't easily avoid a temporary diode current in hard switching H bridge design. It causes considerable commutation losses, restricts the allowed dV/dt and can even cause MOSFET breakdown by triggering a parasitic BJT. Major manufacturers are offering MOSFET series with fast body diode for this reason, e.g. Infineon CoolMOS C6 CFD.

However, it is not difficult to turn the MOSFET on during the period its body diode would be conducting the flyback current. The body diode would then only conduct during the dead time (the short time during which both Top and Bottom MOSFETs are off, programmed into the controller to avoid shoot-through current). Then, the overall efficiency would be maintained or improved, and you wouldn't need to buy additional flyback diodes.

Click to expand...

According to the point mentioned above, you'll strictly want to avoid (large) diode current during MOSFET turn-off, at least in a fast and particularly high voltage H-bride. This means, the dead time has to be adjusted carefully at the ege of cross-conduction. Fortunately, MOSFETs will turn-on gradually by the effect of source inductance.

Apart from this point, adding external "flyback diodes" will rarely help to improve H-bridge performance. It's about useless for dynamic currents and not needed for static currents, because synchronous switches outcompete it.