recirculating/flyback diodes

What's the best type of diode to use for freewheel diodes on stepper motors driven by PWM?

If you do NOT use a recirculating diode when driving an inductive load like a motor coil, then the equations say the coil current will go to zero instantly when the switch is turned off, which means the voltage would go to infinity. With a mechanical switch, this often causes a spark. But I'm not sure what actually happens with a transistor. Either the transistor breaks down under the high voltage spike, or the field is dissipated some other way.

Of course it's common to shunt the motor coil with a diode. From what I understand, if you use an "ideal" diode for a flyback diode, the current in the motor coil will decay slowly after the driving voltage is switched off, because the only thing that causes the current to decay is the resistance of the motor coil. If the motor coil was superconducting, current would flow forever. There would be no voltage across the transistor, because an ideal diode doesn't drop any voltage.

If you use a typical silicon diode, the voltage drop across the diode will be about 1.2 volts. The diode itself will dissipate some power when the driving voltage is removed, although I don't know how you could calculate how much since it's hard to estimate the voltage spike. There will be 1.2V max across the transistor.

If you use a Schottky diode, the voltage drop across the diode will be about .3V. This means the diode itself MAY dissipate less power, and that the current will circulate longer after turning off the transistor. But is this a good thing?

Is slowly-decaying current a GOOD thing? We can see that if the current decayed very slowly or not at all (as in the case of a superconducting coil and ideal diode), then the motor wouldn't step at all. The current MUST decay completely in less time than the shortest step period anticipated for the motor. Correct?

In the case of coils which are current-limited by PWM, is it important for the current to decay in less time than the PWM period? I think not, but I'm just checking. I'm pretty sure it's ok if the coil current stays more or less constant across PWM cycles, as long as it decays fast enough to react to the actual stepping.

I have some 1N4005s, should I just use these and stop thinking about it so much?

Any diode that handles the current pulses will probably do fine - I take it this is a unipolar configuration so you'll have 4 diodes, one for each output. With bipolar the diodes have to go across the H-bridge not the windings directly. MOSFET H-bridges don't need diodes because they are built-in to every MOSFET

MarkT, Are they "built in" or is the commutating diode a result of the design - it effects a diode? In other words, can I make bank on that MOSFET-diode's making any additional diode totally unnecessary?

The use of flyback (or other terms) diodes is an interesting subject. These two links talk about relays, but as you can see, using a simple diode can increase the amount of time a coil takes to 'release it's hold'.

http://relays.tycoelectronics.com/appnotes/app_pdfs/13c3264.pdf and http://relays.tycoelectronics.com/appnotes/app_pdfs/13c3311.pdf

There was another very good one, that I didn't save (wish I had) showing the various waveforms. it appears that inserting a Zener diode (about the same voltage as the energising supply) in series with the diode, will stop the back emf, and not result in am increase in the effective Off time.

Until now I was under the assumption that the diode was it, but maybe not so. For a stepper, this may be important to the top speed.

Mark

runaway_pancake: MarkT, Are they "built in" or is the commutating diode a result of the design - it effects a diode? In other words, can I make bank on that MOSFET-diode's making any additional diode totally unnecessary?

Both, they are required to prevent latch-up or something and they are always rated in the datasheet.