Switching water pump circuit power with Infineon IRF3415PBF MOSFET

[walterc]

This application is for an autonomous agricultural field planter. I need to turn the power on to a 24VDC water pump on the planter with a BeagleBone Black so long as the water pressure, monitored over I2C by the BeagleBone Black is within range. To do this, I use an optoisolator to switch a MOSFET, the IRF3415PBF, on to turn the power on to the pump motor. The pump has a pressure switch that controls whether the pump runs or not to keep pressure in the tanks between 80 and 100 psi. I have a flyback diode across the motor leads close to the motor.

The MOSFET turns on, energizing and running the pump. However, the MOSFET gets extremely hot. It has an Rds of 42 mOhms, and the steady-state running current of the pump is ~6 amps so that I would expect heating of around 1.5 Watts. With a good heat sink, which I have on the device, I would not expect the kind of heating I’m seeing. The Vgs is 9.5V. The max for the device is +/- 20V.

Any ideas what might be going on? I’m not very experienced with this type of circuit, but it seems really simple.

[APDahlen]

Hello @walterc,

What is the MOSFET V_DS while the pump is running?

Verify V_GS is steady at 10 to 15 VDC.

Consider shifting to TO-247 package.

—Aaron

[walterc]

I’ll get those values. But our V_GS is 9.4V provided via a resistor divider from the 24VDC supply.

The IRF3415 is not available in the TO-247 package.

What would cause this kind of heating in a MOSFET when the load isn’t running and the MOSFET isn’t being switched on and off?

[walterc]

Vds when the pump is running is ~4.1V. When the pump is not it is 24.4V.

Interestingly, even though the max Vgs is 20V, the prototype 1 board is wired to use 24V for the gate voltage which is switched through the optoisolator. This runs but it gets really hot too and doesn’t last because of the gate voltage exceeding the max (My mistake).

On hacked a second board, prototype 2, to use a voltage divider to get 9.4V from the 24VDC supply to provide the gate voltage. I connected this one to measure Vds too but forgot to install the heat sink! I was able to measure the Vds on it before it fused the MOSFET from overheating. Vds on this one also measured ~4.1V.

Thoughts?

[rick_1976]

@walterc

Vds when the pump is running is ~4.1V.

Then the FET’s burning close to 25 watts with a 6A load current.

A device with 42mΩ on resistance shouldn’t show that much Vds until the load current approaches 100A. Most likely explanation for excess Vds is insufficient drive, for one reason or another.

[David_1528]

Hi walterc,

It might help if you could post a sketch of your schematic.

Agreeing with @rick_1976, if the gate of the IRF3415PBF were turned on properly (fully enhanced), you should be seeing a Vds in the range of 0.25V with a 6A load (0.42Ω x 6.0A = 0.252V).

The divider circuit seems a little questionable. Seeing the schematic might clear that up. Also, just to be clear, an N-Channel MOSFET is to be connected below the load (i.e. a low-side switch, between your pump and GND) unless you use a special driver designed to drive a high-side N-Ch MOSFET. Otherwise, a P-Channel MOSFET can be used on the high side of a load.

[heke]

Hi @walterc ,

You could also consider using a solid-state relay. Simple to use.

CPC1709J | DigiKey Electronics

Cheers, heke

[APDahlen]

Hello @walterc,

Thinking about this over the weekend:

Silly question, by chance did you place the motor between the MOSFET source and ground?

If yes, move it so that it sits between the positive rail and the drain.

We have all been there…

Sincerely,

Aaron

[walterc]

@heke I am definitely considering doing this. The one you posted can’t handle this pump’s motor current (~6Amp steady state, ~12A starting) and it will be on an autonomous agricultural field planter where summer temps in the enclosure, with no active cooling although shaded, could reach 150-160F. So I am looking at this SSR - Crydom CMX60D20. It can handle 20 A but derating it per the spec sheet it should operate fine for us I think. I don’t know how to add a heat sink to that package though to provide some extra cooling insurance. Any thoughts?

[walterc]

@APDahlen Oh, YES!

Here is the original schematic showing the incorrect placement of the load, the Vgs that’s too high (sourced from the +24VDC battery, max values are +/-20V) and I also suspect the 10K resistor onthe gate to ground is wrong. (Is it?)

[APDahlen]

Hello @walterc,

There it is! No worries, we have all been there.

I’ve written about the MOSFET here.

Guide to Selecting and Controlling a MOSFET for 3.3 VDC Logic Applications

Guide to Selecting and Controlling a MOSFET for 3.3 VDC Logic Applications It is common knowledge that MOSFETs provide an easy-to-use solution for controlling digital loads. In theory, a microcontroller can turn on a MOSFET by directly applying voltage to the MOSFET’s gate terminal. In practice, things are not so simple, especially if we use 3.3 VDC logic. The MOSFET is a voltage-controlled device where the gate voltage determines the conductivity. Classic MOSFETs such as the IRF520 require a …

Please let me know if you have questions.

Sincerely,

Aaron

[walterc]

@David_1528 I agree on the divider. It was a quick way to lower the Vgs. I posted the original schematic below and you’ll see I had made the mistake of using the +24V for Vgs when the max rating for Vgs is +/-20V. I just missed that. I had some 9V LDO linear regulators but they must be bad because they don’t hold 9V so I just tried the voltage divider. All our bench supplies were tied up with other work! It’s definitely not ideal and I won’t use that for a production board.

I think we’re on to the problem with the load being in the wrong place. And, it explains why the original perfboard proof of concept has been running so well for so long. I can’t confirm but I suspect I originally had the load in the right place. That board has never failed.

[walterc]

@APDahlen This is awesome and the references are terrific. The DK forum is always great so thank you very much!

I do have some more questions and as soon as I can get a revised schematic together, would you mind reviewing it to make sure I did things right this time?

I’m using the Infineon IRF3415PbF. I plan to add a 10V LDO regulator fed from the battery, with caps on input and output, and use that 10V for the gate voltage but switch it to the gate with through the optoisolator. Essentially, isolating it from the BeagleBone Blacks’ 3.3V GPIO and enabling a higher gate voltage. The spec sheet for the MOSFET doesn’t have a curve for 3.3V gate voltage and I almost always isolate the Beagle optically from power devices like motors and solenoids.

1. It looks like from the Typical Output Characteristics curves for the IRF341PbF a 10Vgs will work well and keep Rds low and keep power dissipation on a steady state 6A load good. Thoughts?
2. Is a 1K resistor to the gate good?
3. Is a 10K resistor from gate to ground necessary, advantageous? (I’m showing my experience with Darlington transistors as switches here)

[walterc]

@APDahlen one more question… 5. What current should I plan for on the gate? There is a 10V LDO regulator that can supply 500ma and is SMD. I think 500ma should be more than enough but can you confirm?

[David_1528]

Hi walterc,

Yeah, I was kind of wondering about that, as it sounded like your FET was not fully enhanced.

That happens when the N-Ch MOSFET is placed above the load because when it turns on, the Source voltage gets pulled up to very close to the supply voltage, and the Base must be several volts higher than the Source to fully turn on the FET, which means it would need to be above your supply voltage to fully enhance the FET. So it does this dance where it starts to turn on, but can’t fully because the Base voltage can’t go above the supply voltage. Getting that MOSFET to the low side should take care of the problem.

Regarding your alternative idea of going with an SSR, there are a few reasons why it still may be worth considering. The main reasons I can think of immediately are that it simplifies layout, reduces component count, and eliminates a few points of failure. With an SSR, you may not need to use an optocoupler from the BeagleBone, since the SSR isolates the input from the output already. Additionally, you won’t need to worry about doing the level shifting to properly drive the base of a MOSFET since you won’t be using a MOSFET. And, the SSR can be connected on either the high side or the low side of the load and still work.

The downside is primarily component cost. SSRs are typically much more expensive than a MOSFET solution. Also, most SSRs require a bit higher input voltage than a BeagleBone I/O pin can supply (especially over full temperature range and load current range), so you may require a level shifter or current buffer between the I/O pin and the input to the SSR to get it to work reliably. This may mean adding that optocoupler back in to do this job.

[walterc]

@APDahlen @David_11869 @heke @rick_1976 Before I dash off here I want to say a BIG THANK YOU for your help with this. I think we’ve identified the problem. I may yet go with an off the shelf SSR but they are pricey and I think I just about have this design whipped now. Here’s the revised schematic based on how I understood all your comments. Let me know your thoughts.

I plan to put one of these together on a perf board today and try it. (Side note: Is there a breadboard that can handle these high currents? I know mine can’t)

[APDahlen]

Helo @walterc

1. MOSFET resistance is the key to calculate power (P = I2R). CAUTION, the MOSFET resistance is given for a particular V_GS.

2. 1 KΩ is a good starting point for a simple on / off system. This would require more attention if PWM was used.

3. Research MOSFET application. The purpose is to hold MOSFET in the off conditions and to dissipate gate capacitance. Some people will use a diode clamp for static electricity mitigation.

4. Almost zero gate current when the MOSFET is on. However, there is considerable impulse current to charge the gate. The 500 mA LDO is more than sufficient.

Happy soldering,

Aaron

[David_1528]

Hi walterc,

Regarding the MOSFET, the max output current is related to the gate drive, as shown in Figure 2 of the datasheet:

Looking at the graph, it limits out at about 30A at 4.5V, 42A at 5V, etc. The higher the Vgs (to a point) the lower the Rds-on, which is the limiting factor in how much current it can carry. Basically, getting it up to it’s full capability means applying 8V or more.

I would note that the IRF3415 is a relatively old MOSFET, and that some of the newer ones will have improved characteristics. Since your application does not require fast switching (it’s not a PWM application), focusing on parts with a very low Rds-on will minimize heat production.

Parts like the IPP026N10NF2SAKMA1 or IPP019N08NF2SAKMA1 are decent compromises of Rds-on, voltage rating, availability, and pricing.

The 7800 series of regulators should work fine for generating your gate voltage. Some other LDO type regulators have a minimum load current requirement which might be unsuitable for this application, but the 7800 series does not have this issue.

@walterc

Is a 10K resistor from gate to ground necessary, advantageous? (I’m showing my experience with Darlington transistors as switches here)

That pull-down resistor is fairly important because it ensures that the MOSFET gate is pulled low when the input is off. Otherwise internal capacitance in the MOSFET may maintain a voltage on the gate even when the optoisolator is off. This could keep the MOSFET partially turned on, which is obviously undesirable.

Yes, 500mA is more than enough. Without the requirement for high-speed switching (you’re not doing PWM), even a few tens of milliamps would likely be sufficient. The only time the gate draws significant current is the first few microseconds as the input capacitance of the FET gets charged. After that it draws essentially zero current.

@walterc

I plan to put one of these together on a perf board today and try it. (Side note: Is there a breadboard that can handle these high currents? I know mine can’t)

There really aren’t any “high current” breadboards, as far as I am aware of. If you get your MOSFET properly turned on, it should not be a significant source of heat, so that shouldn’t be an issue. Regarding the current, 6A steady-state isn’t extremely high, so that may not be a real problem. To minimise that risk, I’d either plug multiple wires from the motor to the nearest possible holes to the MOSFET Drain and place several wires from the ground bus to adjacent holes of the Source. Better yet, use external jumpers and clip directly to the legs of the MOSFET.

[walterc]

@David_1528 Great feedback! Thanks. I have some perfboards and I’ve started building the proof of concept on it. I will look at the alternative MOSFETs.
I’m planning to use 10V for Vgs so the MOSFET should driven on very well.

[walterc]

@David_1528 @heke @APDahlen @rick_1976 SUCCESS!!!

I built the revised circuit on a perfboard, hooked it up, and it ran the pump for 3 60-second cycles with a 10-second break in between, and the MOSFET didn’t even get warm.

Now to update the PCB layout, order it and some more parts and we’ll be ready for a bigger test.

@David_1528 The alternates that mentioned appear to be good choices so I think I’ll switch to the IPP026N10NF2SAKMA1. It’s a bit more expensive but is pin-to-pin compatible with the other MOSFETs in this package so I’ll be able to adjust later if necessary.

Thank you all for your willingness to help a fellow traveler on this journey! The DigiKey forum is the best! No pretense, no criticism, and just a desire to help out! Such a blessing!.

WC