How to spec a flyback diode?

[Phoxtane]

A project I am working on has a couple of DC fans involved. I know I need a flyback diode to help protect the rest of the circuitry, but I'm not sure how to determine the diode specs so that it does an adequate job.

In my case, I will have two 5V fans in parallel, not controlled independently. As such, I know the diode should be able to withstand at least 5V in reverse bias (e.g. when the motors are turned on) and withstand at least twice the current draw if a single fan in forward bias (e.g. immediately after fans are turned off).

However, I know that when the motors are turned off the kickback voltage generated can be many times greater than the supply voltage of the motors! In addition, I would like some safety factor in the diode specs so I'm not running them at the ragged edge of its capabilities all the time.

So, for a flyback diode being used with DC motors of a given voltage and current draw, how do I determine the necessary reverse bias voltage rating, and the necessary forward bias voltage rating and current rating? Are there other specifications that I am missing?

[ovnr]

Well, if they're normal fans - think computer fans - you don't even need a diode.

If they're actually old-school brushed motors, you don't need much. Some rules of thumb:
Current rating: At least 1/10th of the normal motor draw. Voltage rating: At least 1.5x the working voltage. Forward voltage: Doesn't matter much.

[Phoxtane]

They're not exactly computer fans (or at least being sold as computer fans), but I'm fairly certain they're brushless DC motors - which means then I don't actually need the flyback diode, as you suggested. Thanks!

[Siwastaja]

If they're actually old-school brushed motors, you don't need much. Some rules of thumb:
Current rating: At least 1/10th of the normal motor draw.

This is totally wrong. You need at least the normal worst case (locked rotor, or whatever limited by current limiter circuit, if any) motor current draw, simply because the full motor current goes through the diode. Depending on the load, it can be in a continuous mode with relatively small current ripple, which would mean that the current is basically the same for both MOSFET and the diode. In this case,  you also need to watch out diode dissipation, especially at low speeds (since the duty cycle for diode is now high), but this is often not a problem with fans since they require very little torque (hence current) at low speeds.

A computer fan is indeed easy, however. But even with freewheeling internally taken care of, you might still want to have the diode at least to bypass magnetic energy stored in wiring inductance, and for the case someone connects another type of fan or motor. But with short wires and a normal BLDC fan and no way to use it for other types of motors, don't bother with the diode.

[Phoxtane]

I actually will have quite long wires for the fans. I guess that means I do need the fly back diode after all.

[Neilm]

Read the datasheet. There are some fans out there that have protection circuits built in, so if the fan stalls it only draws bursts of current as it trys to start the fan

[ovnr]

If they're actually old-school brushed motors, you don't need much. Some rules of thumb:
Current rating: At least 1/10th of the normal motor draw.

This is totally wrong. You need at least the normal worst case (locked rotor, or whatever limited by current limiter circuit, if any) motor current draw, simply because the full motor current goes through the diode. Depending on the load, it can be in a continuous mode with relatively small current ripple, which would mean that the current is basically the same for both MOSFET and the diode. In this case,  you also need to watch out diode dissipation, especially at low speeds (since the duty cycle for diode is now high), but this is often not a problem with fans since they require very little torque (hence current) at low speeds.

For a flyback diode on a brushed motor, the diode only sees the inductive spikes above the bus voltage. In the locked-rotor case, your switching device may see excessively high current, but the diode will just sit there doing nothing.

When running normally, there's going to be a small amount of commutation-related noise, but nothing major.


Of course, if you PWM the thing, you're in a somewhat different area, but OP never mentioned that use case. Nor did he mention fan connections exposed to the user.
As for "long" wires - say 50 cm: Too low energy to actually be a big deal, plus you have a cap in the fan that'll take most of the sting out of it.


To OP: Is your application cost or size-restricted? If no, adding a diode is cheap insurance, and it won't hurt anything. It'll be fine without it, but overengineering it is always good in my book.

[Phoxtane]

Of course, if you PWM the thing, you're in a somewhat different area, but OP never mentioned that use case. Nor did he mention fan connections exposed to the user.
As for "long" wires - say 50 cm: Too low energy to actually be a big deal, plus you have a cap in the fan that'll take most of the sting out of it.

To OP: Is your application cost or size-restricted? If no, adding a diode is cheap insurance, and it won't hurt anything. It'll be fine without it, but overengineering it is always good in my book.

Not exactly cost OR size-restricted. I already will need one on the RESET for the microcontroller (following Atmel's application notes), so I figure whatever diode I use for the fans here will be suitable for that as well.

I'm planning on using one of the fans from this family of devices: http://www.qualtekusa.com/Catalog/Fans/pdf_files/FAD1-06020.pdf

I'm still lost on how to determine the specs of my diode due to the conflicting information above - and, the datasheet I linked in says these fans are 'impedance protected' - which makes me unsure if I even need a diode in the first place!

[T3sl4co1l]

This is totally wrong. You need at least the normal worst case (locked rotor, or whatever limited by current limiter circuit, if any) motor current draw, simply because the full motor current goes through the diode.

Well.  This is totally wrong, because the diode is pulsed, not continuous.

Except for the very smallest of diodes, the pulsed rating is always very favorable.  Example:
https://www.onsemi.com/pub/Collateral/MMSD4148T1-D.PDF
200mA average, 1A surge (< 1 s).  It's unlikely that flyback would last that long, so you can use the thermal profile (if provided) to determine exactly how much is tolerable for other pulse width and duty cycle.

In short, the diode does not need to be any larger than the switch delivering that current in the first place.  Whenever I see 2N3904 driving a load with a 1N4001 across it, I'm amused.  1N4148 is more than sufficient there.  Now, 1N4148 would be inappropriate for a TIP120 or IRF540 switch, say, where a 1N400x, or even better still, UF400x, would be appropriate.

A zener/TVS can also be used, which has the advantage of faster discharge time (faster relay and solenoid operation) and returning the current to the switch's common terminal (GND), protecting the switch.

Note that the flyback diode should be similarly returned to common, NOT placed directly across the load.  This is unimportant for short lead lengths, but an exaggerated case -- say, many meters of cable between switch and load -- introduces stray inductance, which is the very thing you're trying to deal with in the first place.  Note that, in this case, "common" means AC ground -- the exact "common" you would use is +V, bypassed to GND (or switch common) with a large enough capacitor, at the switch.  Thus, when the switch turns off, the current flows from the switch, to the diode, through the capacitor, back to the same common.  This loop being physically small ensures minimal stray inductance, and thus overshoot voltage at the switch.  This principle extends to proper switching converters, where it's done at scale, and overshoot is a necessary consideration.

Tim

[ovnr]

Not exactly cost OR size-restricted. I already will need one on the RESET for the microcontroller (following Atmel's application notes), so I figure whatever diode I use for the fans here will be suitable for that as well.

I'm planning on using one of the fans from this family of devices: http://www.qualtekusa.com/Catalog/Fans/pdf_files/FAD1-06020.pdf

I'm still lost on how to determine the specs of my diode due to the conflicting information above - and, the datasheet I linked in says these fans are 'impedance protected' - which makes me unsure if I even need a diode in the first place!

"Impedance protection" probably means "the winding resistance is high enough that it won't catch on fire no matter what"...

Anyway, no, you don't need the diode for that fan. Feel free to add one; at these power levels, any cheap and cheerful diode will do the job.

It's like adding a overpressure vent to a pneumatic system; if your compressor is limited to 10 bar and the smallest possible vent is 11 bar, adding it doesn't hurt anything. It's just not going to do much. But sure, if ~something~ happens, it'll be useful.

[danadak]

Basically safe area considerations govern diode choice.

This paper sheds some light, as well as a number of references at the end of
the paper.


http://www.5scomponents.com/pdf/copyright_ieee_01mr.pdf


Regards, Dana.

[Siwastaja]

While I don't want to argue about semantics, I have seen the term "flyback diode" used synonymously to "freewheeling diode", i.e., a diode directly across the load, in regards to motor control so freaking often that I didn't even think about the possibility that it could mean anything else in this context.

All the discussion is 100% meaningless without a schematic provided to confirm what is the intended position of the diode. What I said naturally only applies to the diode across the load, which, together with the MOSFET, completes the half bridge, providing the current path in the motor windings, which, in case of a brushed motor in continuous mode, is typically the same as the MOSFET current, and in some cases (running low RPMs) at duty cycle way over 50%.