High Current MOSEFT Circuit Resistors

[Silntknight]

If I am connecting a 12V circuit across a 60V/30A MOSFET to a high impedance (10-20 Ohm) fuel injector, do I need to connect a resistor between the +12V source and the MOSFET? The fuel injector is designed to run on straight 12V and high impedance fuel injectors aren't supposed to need additional resistors. See below:

+12V or without the resistor?
^
|
|
|
\
/
\
/
-------> Fuel Injector
|
|
|
MOSFET

 

[dknguyen]

The circuit you have, as drawn, is sending a low current, low power, voltage signal. That bit is important- a SIGNAL which means to transmit information (like data, commands) but not power or current. As a signal it implies low power since your goal is to transmit information rather than power and to conserve power you make it as low power as possible.

And it's a voltage signal, so it's not even a current signal that uses small varying levels of current to send information, but a voltage signal which means as close to zero current as possible (to conserve power). In that case you use much larger resistors (kohms) and much smaller transistors since your goal is to transmit a signal, not power so you want to save power.

So the circuit as you have drawn is only correct if you are trying to send a voltage signal to your fuel injector, possibly to command it turn on or off, but not to power it or connect and disconnect it from the power source.But like I said earlier, in that case you use large resistors and small low current MOSFETs and your MOSFET is enormous which leads me to think you are just trying to shut power on or off to the fuel injector.

To actually pass current and power through the injector, there actually is no resistor, but the circuit is wrong. Removing the resistor in your circuit just makes a short-circuit whenever the transistor turns on which bypasses the injector, providing it no power or current, and blows up your battery and MOSFET at the same time.

The proper circuit is this:

+12V----->Fuel Injector----->(Drain)NMOS(Source)------>GND

I replaced your MOSFET with an NMOS, which is the most common type of MOSFET and the easiest to work with in the position shown. The alternative is:

+12V----->(Source)PMOS(Drain)---->Fuel Injector----->GND

Technically you can use either a PMOS or NMOS in any of those two positions (ie. an NMOS closest to +12V or a PMOS closest to GND), but it is not as natural to generate the required gate signal for the MOSFET (the third pin which controls whether the transistor is on or off). For your reference, here are those two circuits:

+12V----->Fuel Injector----->(Source)PMOS(Drain)------>GND
+12V----->(Drain)NMOS(Source)---->Fuel Injector----->GND

 

[Jaguarjoe]

A 10-20 ohm injector will draw about 1 to 1.5 amps. No resistor is required for this. Use an N channel MOSFET switching the ground side of the injector. Be sure your MOSFET has a high enough Vds to survive the back emf pulse from the injector. Be sure to use a good gate driver scheme to overcome gate capacitance which will allow fast MOFET switching.

 

[dknguyen]

A 10-20 ohm injector will draw about 1 to 1.5 amps. No resistor is required for this. Use an N channel MOSFET switching the ground side of the injector. Be sure your MOSFET has a high enough Vds to survive the back emf pulse from the injector. Be sure to use a good gate driver scheme to overcome gate capacitance which will allow fast MOFET switching.

Which is this from my post:
+12V----->Fuel Injector----->(Drain)NMOS(Source)------>GND

 

[Silntknight]

Ok. Jaguarjoe, first, could I use a high voltage diode to prevent backflow EMF? Second, do you mean the minimum opening time? I know it takes about 1ms for the coil to charge and the injector to open. I have that corrected for in my program.

 

[Silntknight]

Or should I just use a 500V NMOS?

 

[dknguyen]

If you're only switching the injector on and off and low frequency (maybe once every 30 seconds or every few minutes, a human time scale) then you don't need to worry too much about heating due to switching too slowly (normally in the low ms or much faster unassisted by a gate driver). The time spent switching is a tiny fraction of the operating time.

But if you're switching at high frequency thousands of times per second for PWM to run at variable duty cycles, then that's a different story.

The back EMF is a result of the abruptly interrupting current flow through the motor inductance (any inductance really) whenever you switch it off. A higher voltage NMOS would certainly be more resistant to damage, but higher voltage NMOS have to trade off certain performance characteristics for that high voltage. Primarily they aren't as efficient (higher resistance) which means more heat which may or may not be a problem.

Just use a fast diode in anti-parallel with the injector (parallel in the direction that doesn't short out the injector allowing battery current to bypass the injector and travel through the diode instead). It doesn't need to be high voltage either. It mainly just has to be fast. It doesn't even have to have a very high continuous current rating either either since it's use is not momentary rather than continuous which means that equal to yoru operating current is plenty. Obviously, it has to be able to withstand and block your battery voltage plus some for safety (maybe 15V, 20V, or 30V).

 

[ronv]

Injector driver

One of the circuits below should work. There are 3. One with no clamp or snubber. This one turns off quick, but as you can see by the high voltage it is the avalanche of the fet that limits the rise of the voltage.
Next is a diode clamp. This is nice but if you look at the current it continues to flow for a while after the fet is turned off. If you can compensate for the additional on time this would be best.
3rd is a snubber. It's kind of a compromise between the 2 and would need to be "tuned" a bit to match the injector.
What are you driving it with?

 

[Jaguarjoe]

The zener diode in picture #2 is not correct. As it stands, it is acting as a regular diode across the coil like you would do with a relay coil. 0.7v drop.
It should be across the MOSFET with a rating less than the Vds of that MOSFET, like maybe 40 volts or so.
You want to eat the spike but not slow down the injector (assuming this is an IC engine application).

 

[dknguyen]

Two ways to do it I suppose.

It's fine the way it is across the inductance except it should be a schottky diode, not a zener. When the current is interrupted through the inductor it allows a current path around the inductor and clamps the BEMF voltage spike to the forward voltage drop of the diode. The result is that the energy in the BEMF can dissipate as a low voltage, high current surge rather than a low current, high voltage spike.

Your "zener in parallel with the MOSFET" works...but I'm not sure why you would want to use a zener anti-parallel with the MOSFET in this case since it has a higher clamp voltage than just placing a schottky diode anti-parallel with the MOSFET.

EDIT: Oh you, I see. You're trying to not slow down the shut off speed of the injector and that requires a sacrificing some low voltage clamping in order to get speed. Is that critical in a fuel injector? Doesn't seem to be a problem in many cases. I don't know much about engines so you'll have to tell me if a slower shut off time is a problem or not.

 

[Jaguarjoe]

It's correct the way it is across the inductance. When the current is interrupted through the inductor it allows a current path around the inductor and clamps the BEMF voltage spike to the forward voltage drop of the diode. The result is that the energy in the BEMF can dissipate as a low voltage, high current surge rather than a low current, high voltage spike.

You're thinking of an H-bridge

No H-bridge thoughts here. The problem with using the zener as presented, or a Schottkey is that it will slow down the closing of the injector. A properly placed zener of correct value will provide for pretty quick closing and pretty good back emf clipping at the same time. Installed as shown, the coil back emf will be 0.7v and Vz will never enter the picture. Installed the other way around, the 0.7 Vf of the zener will allow 12v to bypass the coil and go to ground through the MOSFET releasing smoke.
The rise time of the injector coil and the fall time of the coil have a large bearing on injector on time. Not much can be done about the rise time, but the fall time is controllable.

 

[Silntknight]

No H-bridge thoughts here. The problem with using the zener as presented, or a Schottkey is that it will slow down the closing of the injector. A properly placed zener of correct value will provide for pretty quick closing and pretty good back emf clipping at the same time. Installed as shown, the coil back emf will be 0.7v and Vz will never enter the picture. Installed the other way around, the 0.7 Vf of the zener will allow 12v to bypass the coil and go to ground through the MOSFET releasing smoke.
The rise time of the injector coil and the fall time of the coil have a large bearing on injector on time. Not much can be done about the rise time, but the fall time is controllable.

Why can't I post a normal reply?

 

[Silntknight]

Why can't I post a normal reply?

I have to post with a reply or else it doesn't let me post at all...
This is the seventh time I typed this because the forum isn't displaying my post... I am using this circuit currently (RMCybernetics - DIY Homemade Ignition Coil Driver). The NMOS would switch for 1ms (on-off) then rest for at least 8.3ms. Do I need both a snubber and a clamp? I like the Schottky solution best (for space saving) but the capacitor/resistor for switching time. If my NMOS is rated for "2kV of ESD protection," do I need either? Finally, if I used a capacitor/resistor snubber, would I use a capacitor with a lower voltage than the NMOS? What about the resistor wattage rating?

 

[Silntknight]

Also, if I have 4 of these in a row, can I use the same capacitor/resistor snubber or the same Schottky diode for all of them, or do I need 4 separate ones? Only two would ever be activated at once.

 

[Silntknight]

Sorry, I have five in a row. 1 for a spark plug, 4 for fuel injectors (2 activate simultaneously).

 

[dknguyen]

JaguarJoe:I think I spent too much time editing my post and you replied while I was still editing. I realized what you were going for (ie. not an H-bridge). Might be best to just clear the mention of the H-bridge response so people don't get confused by my irrelevant comment.

SilentKnight:
When you say "in a row" what do you mean exactly? Do you mean multiple MOSFETs each controlling a single device or in some cases a MOSFET switching a group of devices? If the latter is the case, are the devices in the group connected in series or parallel? Either way...one snubber/diode is good enough for each GROUP of devices (one snubber/diode for each MOSFET whether it goes across the MOSFET itself or the devices the MOSFET switches). The devices in each group should also be connected in parallel unless there are special requirements. Connecting them in series causes each device in the group to see only a fraction of the 12V from the supply- not what you want if each device runs off 12V.

Either way, you only need one snubber across all of them. But yeah, RC snubbers are fast (no switching time they are basically always on) but smooth out and slow the spike down rather than providing a hard ceiling on the voltage clamp. Diodes on the other hand clamp the voltage to a set level but take time to turn on. Best case is to use both. It's really up to you.

For most applications just a diode is a good choice though because schottky diodes are very fast already and usually fast enough plus they are safe in that they clamp the voltage to a set level which makes them very easy to select.

To properly size the components of a snubber can get really complicated because you need a lot of information about the switching time, inductance, resonsances and all that mostly unmeasurable stuff. Too small and the voltage spiked isn't slowed down enough and can reach damaging levels. Too large and the components get expensive and lots of excessive heating and energy waste happens in the snuber components on every switching. There are a some PDFs on google that describe the process in somewhat understandable of complexity. But even then a lot of it is to get a starting point for the values and then you test it and view things on an oscilloscope and adjust the values until you get what you want.

Pairing the two up makes it easier since the snubber can be small with looser requirements since it just has to hold things at bay until the diode kicks in.

 

[Jaguarjoe]

How fast is this engine supposed to run? This will have a big bearing on how the spikes will be suppressed.

Using a 3055 to drive a coil is not a robust solution. This is a better choice- http://www.datasheetcatalog.org/datasheet/motorola/MJ10012.pdf If you NEVER run with the coil secondary unloaded (no spark plugs) the back emf pulse will only be 250-300 volts, unloaded it will be at least twice that and will require zener clipping across the transistor. Snubbing across the coil will adversely affect the spark.

Your coil driver circuit has no current limiting. a 1 ohm coil will eat 10 amps or so. A ballast resistor will take that to a reasonable level- like 5 amps.

 

[Silntknight]

Sorry, in a row literally meant lined up next to each other. That's how they are on my PCB design right now. There are 5 MOSFETS, each connected to an independent device, each connected to a different microcontroller pin. 4 of the MOSFETS are each connected to 1 fuel injector (1 is actually an injector, the rest are being used as fast acting solenoids). The fifth is controlling a spark plug. I thought that all could be treated in the same way because each one is basically a NMOS controlling an inductor coil.

Back to the heart of the matter, because it is one device connected to one NMOS, I should have a total of 5 clampers/snubbers or just clampers. Because I know the requirements for the spark plug snubber, I'll probably keep that one along with the clamper. It might be best to have both because of the fact that an ignition coil can reach 60kV, and regularly does. Do you think that both are necessary? Speed is very important to me. If you think both are necessary, I could use the same components for both, because the ignition coil snubber/clamper has to slow/clamp a potentially higher voltage (no pun intended). The parts are also pretty cheap.

If this clarifies anything, I was basically asking if I could use 1 snubber/clamper for all 5 NMOS circuits and connect it in parallel. Then I realized I couldn't because of the way it's connected.

Also, what do you think about this. I currently have two NMOS controlling two devices (both fuel injectors) but I want them to be synchronized. Instead of using my microcontroller to synchronize two outputs, could I connect it like this: +12V--->[+](fuel injector 1)[-]--->[+](fuel injector 2)[-]--->[drain]NMOS[source]--->Gnd
My only concern is voltage drop across fuel injector 1 and the trace width requirement for handling . I know each will require less than 1 amp, so a total (maximum) of two amps will be flowing. Using an online calculator, I think I should be fine with 1.02mm traces (I know, I have a lot of overhead here), but will voltage drop be a concern. The injector's resistance is between 12 and 20 ohms.

 

[dknguyen]

Back to the heart of the matter, because it is one device connected to one NMOS, I should have a total of 5 clampers/snubbers or just clampers. Because I know the requirements for the spark plug snubber, I'll probably keep that one along with the clamper. It might be best to have both because of the fact that an ignition coil can reach 60kV, and regularly does. Do you think that both are necessary? Speed is very important to me. If you think both are necessary, I could use the same components for both, because the ignition coil snubber/clamper has to slow/clamp a potentially higher voltage (no pun intended). The parts are also pretty cheap.

If you feel it's necessary go for it. Just make sure to use non-inductive, surge-resistant resistors and capacitor types. Good common candidates are NON-spiral cut carbon composition resistors and ceramic capacitors. There are of course, more better and much more (and I do mean much more $5-$25 a pop) expensive types like bulk ceramic resistors and polypropylene capacitors specifically designed for the job. And minimizing trace lengths and loop areas for fast acting circuits like this (both snubbers and flyback diodes) is critical if they are to do their job.

https://www.electro-tech-online.com/custompdfs/2010/11/design.pdf
The thing to carry away from this PDF is the power dissipation required by the resistor based on the size of a capacitor on page 4. The rest is good if you understand it. Larger capacitor and smaller resistor damps more but at the cost of component size, heat dissipation requirements and expense (this last one is especially noticeably because of the cost of the high quality snubber resistor and capacitors required). A good diode should take up your slack though which is why it's easier if both are used than just a snubber.

If this clarifies anything, I was basically asking if I could use 1 snubber/clamper for all 5 NMOS circuits and connect it in parallel. Then I realized I couldn't because of the way it's connected.

Sounds like you sorted that out on your own.

Also, what do you think about this. I currently have two NMOS controlling two devices (both fuel injectors) but I want them to be synchronized. Instead of using my microcontroller to synchronize two outputs, could I connect it like this: +12V--->[+](fuel injector 1)[-]--->[+](fuel injector 2)[-]--->[drain]NMOS[source]--->Gnd

This is exactly what I was talking about when I said that multiple devices under the control of one MOSFET should be conneced in parallel with each other, and not series. In series, each would be running on about 6V, half of the 12V supply (assuming identical injectors). The full12V would be split between the two injectors based on their impedances relative to each other. In parallel, each would be operating like normal as if it were on it's own with the MOSFET handling the combined current of both.

           /--->[+](fuel injector 1)[-]---\
+12V------                                  ----->[drain]NMOS[source]--->Gnd
           \--->[+](fuel injector 2)[-]---/

 

[Jaguarjoe]

.




Also, what do you think about this. I currently have two NMOS controlling two devices (both fuel injectors) but I want them to be synchronized. Instead of using my microcontroller to synchronize two outputs, could I connect it like this: +12V--->[+](fuel injector 1)[-]--->[+](fuel injector 2)[-]--->[drain]NMOS[source]--->Gnd
My only concern is voltage drop across fuel injector 1 and the trace width requirement for handling . I know each will require less than 1 amp, so a total (maximum) of two amps will be flowing. Using an online calculator, I think I should be fine with 1.02mm traces (I know, I have a lot of overhead here), but will voltage drop be a concern. The injector's resistance is between 12 and 20 ohms.

Two injectors in series will consume 1 amp total. However, each one will have 1/2 the available voltage across them. So that is 6v and 6v, not enough to turn the injectors on unless you use a 24v supply. If you put the injectors in parallel they will have full the 12v across them, but at twice the current.

I didn't follow the beginning of your post very well. If you use a 3055 for a coil driver, it won't last long. Applying 250-300 volt spikes to a 60 volt transistor is stupid. Use the MJ10012 device and put 350 volts worth of zeners across it. It will last forever. Do not put anything across the coil. For injector and solenoid drivers pick any MOSFET rated at 100v/ 5 amps or better. Put a 47v 5 watt zener like a 1N5368 across each MOSFET. Rock and roll!