Flyback diode configurations

Looking around, it seems there are a number of ways to place a flyback diode across an inductive device (such as a solenoid in this case) to protect against the voltage spike at switch off - namely;

  • conduct the spike to ground
  • recirculate the spike to the positive of the device
  • or, either of the above with a resistor in series with the diode to dissipate more of the spikes' energy as heat

I've heard it's awful practice to conduct the spike to ground, but it would also appear recirculating it back to V+ isn't great, as that slows down closing of the solenoid.

I'm not looking for a "well, it works dosen't it?" type solution. I'm looking to build a robust, effective setup.

So, questions -

a) what are the advantages / disadvantages, other than those I've named - of each setup?
b) are there any other configurations other than those above?
c) ...it would appear, I'm also at a loss as to how to actually select the appropriate diode. With relays in the past, I've just used a 1N4007 - but I'm sure there must be better diodes for the job out there.

I'm switching typically 0.75 - 1.5A through the solenoids on a 12V power supply.

For driving a current in that range, I would use a 1N5407
Peak forward surge current
8.3 ms single half sine-wave 200amps
superimposed on rated load

I would place this across the coil as close to the terminals as possible!

I've heard it's awful practice to conduct the spike to ground, but it would also appear recirculating it back to V+ isn't great, as that slows down closing of the solenoid.

I'm not looking for a "well, it works dosen't it?" type solution. I'm looking to build a robust, effective setup.

You might look up actual data re "slows down closing of the solenoid". How slow is it?
2-msec? 2-sec? As far as effective goes, probably 99% of the time [guesstimate], people
just put the diode in reverse across the coil, so that should say something.

I would place this across the coil as close to the terminals as possible!

Cheers for the device recommendation! I can't get close to the solenoid terminals unfortunately, these'll be a couple meters away.

You might look up actual data re "slows down closing of the solenoid". How slow is it?
2-msec? 2-sec? As far as effective goes, probably 99% of the time [guesstimate], people
just put the diode in reverse across the coil, so that should say something.

1ms extra opening would be far, far to long in my application - that's why I'm concerned. I'm in the 1% unfortunately. I can't put the device across the coil (I should have mentioned that before!) but I can feed it back into the power supply that feeds the board before the regulators etc, as that ultimately is the same supply that feeds the external solenoids.

I'm not sure if that, or conducting it to ground would be better...

jtw11:

I would place this across the coil as close to the terminals as possible!

Cheers for the device recommendation! I can't get close to the solenoid terminals unfortunately, these'll be a couple meters away.

You might look up actual data re "slows down closing of the solenoid". How slow is it?
2-msec? 2-sec? As far as effective goes, probably 99% of the time [guesstimate], people
just put the diode in reverse across the coil, so that should say something.

1ms extra opening would be far, far to long in my application - that's why I'm concerned. I'm in the 1% unfortunately. I can't put the device across the coil (I should have mentioned that before!) but I can feed it back into the power supply that feeds the board before the regulators etc, as that ultimately is the same supply that feeds the external solenoids.

I'm not sure if that, or conducting it to ground would be better...

Well, I think you're gonna be in some trouble here. A couple of meter long wires to squelch
an inductive spike is probably almost as bad as using nothing, as far as EMI generation goes,
at least. Think - big loop antenna, high voltage up to 400V, large currents for short time
periods. You get what you pay for. Worse for bigger solenoids.

Far, far too long! Check the movement speed of your solenoid. I suspect it's MANY msec
longer than 1-msec. Mechanical devices tend to have a bit more inertia than electrons.
Worse for bigger solenoids.

Far, far too long! Check the movement speed of your solenoid. I suspect it's MANY msec
longer than 1-msec. Mechanical devices tend to have a bit more inertia than electrons.
Worse for bigger solenoids.

My solenoid in this case is a high impedance fuel injector, and of course - with typical injection pulsewidths ranging from 1 to 15ms, the mechanical movement time of the injector is very small indeed. That's the very reason I cannot avoid the long wire runs, but then again - it works successfully in everybody's vehicle on this forum.

Of course, if I could - I wouldn't have the long wire runs, but I simply cannot avoid it.

So really, my choice is - do I conduct the spike to battery voltage, or do I conduct to ground.

Well, in that case, I guess you need to find out what everyone else on this forum is using.
Or else find out what the fuel injector people recommend.

Use a diode across the coil at the PCB.. I know it sounds rather contrary considering the potential and real EMI but that path is a little more certain than the battery as that has a long way to the battery through the ignition switch, A lot more wire to radiate. The other consideration is that anything that is designed to work in that environment is designed to work in a very adverse environment. I would worry more about my equipment being affected by My EMI than I would about radiating stuff that might be a cause for trouble for others. Remember that what you are trying to do is emulate the driver so..
It wouldn't be out of place to find both a schematic and an OEM solenoid controller or ICU and see how it was done by "The Big Boys".

Bob

http://automotivetestsolutions.com/images/escope/fuelingectortutorial/FuelInjectorWaveforms.htm
Use a zener diode

This is not basic 101 there a lot that has to happen here to do this right.

This induced voltage is called the fly back voltage. The flyback voltage is then clipped (Figure 2 Part G). This voltage level changes between manufactures and systems. The flyback voltage is adjusted for the injector design being used in the circuit. The flyback voltage is set for the electromagnetic coupling of the injector and the mechanical spring rate. The magnetic field around the injector winding is stored energy which is used to control the speed that the pintle is closing with. If the pintle is allowed to close too fast the pintle and seat will become pounded out and will begin to leak fuel. With a fast closing rate the pintle can also bounce causing extra fuel to be delivered to the engine. This extra fuel cannot be controlled accurately so the engineer must adjust the energy held within the flyback voltage to accurately control the closing rate of the injector.

http://automotivetestsolutions.com/images/escope/fuelingectortutorial/FuelInjectorWaveforms.htm

Definitely helps to do a little legwork on one's problem aforehand. There are some good tutorials there,

http://automotivetestsolutions.com/tew.htm

Yes there some really good stuff on that site how the zener set's the time it take's to close the injector.

This is done by using a zener diode across the PCM driver (transistor or MOSFET). See Figure 6.
As the magnetic field falls back into the injector winding the energy is allowed to loop through the circuit. This allows the current to diminish at a set rate. The lower the voltage is set by the zener diode, the more energy is allowed to loop through the circuit. If a diode were used rather than a zener diode, this would let the most energy allowed loop through the circuit.
A diode will allow the stored energy to loop until it reaches source voltage, in this case 12 volts. This would allow the injector the longest period of time to close. The higher the zener diode voltage, the shorter the period of time allowed for the injector to close. This is due to the energy looping through the circuit being cut off early by the voltage rating of the zener diode.
If a 65 volt zener diode is used, the energy that is looping through the circuit is stopped at 65 volts which is 53 volts sooner than a diode, that allows the energy to continue to loop through the circuit until it reaches 12 volts. So the energy from a higher rated zener diode will shut off the energy looping through the circuit sooner which allows a faster pintle closing rate. Likewise, the energy from a lower rated zener diode will allow the energy to loop longer which will cause a slower pintle closing rate. This rate is set by the zener diode voltage which is matched to the injector design.
The delay in closing voltage can be seen in Figure 2 Part H on page 27. Once the Pintle starts to fall through the magnetic

Also, there is so much electrical noise in an automotive environment that a bunch of EMI generated by injectors
is probably miniscule compared to everything else, especially spark plugs and ignition coils and alternators, as
well as starter motors, air compressors, power steering pumps, on and on.

Therefore, the manufacturers will have gone to great lengths to add protection cktry on their ECUs, on both the
I/O pins, and the power and ground pins. I imagine an Arduino board would last about 3-msec in there.

oric_dan(333):
Also, there is so much electrical noise in an automotive environment that a bunch of EMI generated by injectors
is probably miniscule compared to everything else, especially spark plugs and ignition coils and alternators, as
well as starter motors, air compressors, power steering pumps, on and on.

Therefore, the manufacturers will have gone to great lengths to add protection cktry on their ECUs, on both the
I/O pins, and the power and ground pins. I imagine an Arduino board would last about 3-msec in there.

Couldn't agree more. I'm not using an Arduino board, I'm in the process of designing my own PCB, and just happen to be using an AVR chip and have stuck with this forum as a number of people have been incredibly helpful.

The Zener info is great, thanks. Conducting to ground is what Texas Instruments recommend in a couple of application notes, using a ~33V Zener. However, that was an old app note. I'll test a number of different setups, scope it out and see what I get. It would seem to, if a Zener failed - there's a direct path to ground. Using a higher voltage Zener and recirculating sounds like a viable option.

On the subject of noise again however, my application won't be locating the board in an engine bay - and have been spending considerable time protecting parts of the design from noise.

Thanks for all the input! When I've got results, I'll post them up here with some scope shots etc and let you all know what I've got.