Switching P MOSFET slowly.

I need to switch some P channel MOSFETs for a minimum on time of 0.5 seconds. These need to switch 12V using a (5V) pic pin. I'm wondering if a 220R resistor between the gate and 12V will hold the MOSFET off and a transistor (NPN) controlled by the pic pin could ground the gate to switch it on. I've not chosen a MOSFET yet and the (inductive) load is ~1.25A . What do people feel the minimum continuous rating for the MOSFET should be - I'm assuming switching losses won't really factor in to the calculations - or will they.
Or, has someone a better suggestion?

Thanks all,

Mike.
P.S. I considered grounding the gate with just the pic pin but the protection diodes will hold the MOSFET (partially?) on - gate at 5.7V.

I often use this circuit for high side switching of 12 and 24 Volt loads.



I choose mosfets more by their RDSON than just their current rating. Low milliOhms = low power dissipation. And usually translates to an Amp rating generously higher than necessary.

My NPN choice is usually a pre-biased BJT. The MUN5211 has two parts in a tiny SOT-363 package for very little money. Other configurations and resistor values and connections are available from multiple vendors.
https://www.digikey.com/en/products/detail/onsemi/MUN5211DW1T1G/1749059

Choose values for R1 and R2 based oh what on/off timing you need. Choose the ratio to fully turn the P-mos device on.

You will need to provide protection to the mosfet from the inductive spike at turn off.

When I started looking I found the SI2305CI which, with RDSon=46mΩ Vds=20 and Ids=5 and 5c seems ideal.
I'm assuming I'll need a diode additional to the parasitic diode.

Mike.

Pommie said:

When I started looking I found the SI2305CI which, with RDSon=46mΩ Vds=20 and Ids=5 and 5c seems ideal.
I'm assuming I'll need a diode additional to the parasitic diode.

Mike.

Click to expand...

Yes, the parasitic diode is just a by-product of the manufacturing process, and not for actual use.

Have you considered SM dual-FET's?, I use ones like the DMC3016LSD (it varies depending what happens to be in stock at the time of order), they include an N-gate and a P-gate, are easily configured for high side switching, and are low RDS and stupidly high current for their tiny size. Simply replace Q1 above with the N-Gate half, and alter the resistors.

The parasitic diode often used in motor control, low speed applications, because its
reverse recovery time was poor. But newer processes and device architectures improve
that such that it can be used as general purpose clamps in L load switching.

Consult data sheet.



Google "fast body diode mosfet", quite a lot of info on the web.


Regards, Dana.

Regarding the circuit in #2, three things:

1. Because a GPIO pin can both source and sink current, you do not need R5 to assure the complete and rapid turn-off of Q1.

2. If you replace Q1 with a 2N7000 small MOSFET, you can eliminate R4. A small MOSFET in this application does not need a current limiting or EMI suppression resistor at its gate.

3. R1 and R2 form a voltage divider across M1, reducing the gate drive voltage. In a 12 V system, Vgs is reduced to 8 V. With a FET rated for 20 volts Vgs, I would eliminate R2 and drive the M1 gate directly. With the increased Vgs, you now are not limited to using a logic-level MOSFET, as most "normal" power MOSFETs are rated for full enhancement (minimum Rdson) with Vgs = 10 V.

Regarding post #1, 220 ohms is a very low value for a gate turn-off resistor. You are correct that one is needed, but something in the 1 K to 10 K range is much more common. As the resistor value gets smaller, the collector or drain current and the power dissipation in the driver transistor increases. At 220 ohms, you are unnecessarily sinking over 50 mA.

For minimum ratings, a very common industry rule of thumb is 2x.

12 V circuit = 25 V (or higher) voltage rating for semiconductors, capacitors, etc.
1.25 A = 3 A or higher current rating for semiconductors, capacitors, connector pins, etc.
1 W power dissipation (resistors, FETs, whatever) = 2 W or higher components.
etc.

Also, in a p-channel MOSFET the body diode "points" toward the source. You do not need an external diode to prevent unwanted conduction through the body diode unless the load device or circuit has significant energy-storage capability (such as a large input capacitor) and the 12 V source can go low enough to cause enough reverse current to matter. This is a relatively rare condition. and should not come up when simply powering an inductor.

ak

Hi,

Inductive loads are a little more tricky than purely resistive loads.
Just keep in mind that the inductor turn-off back EMF voltage spike is of the opposite polarity to the applied voltage when it is energized.

When you switch an inductor on with an N channel device, you are effectively creating a boost converter. When the N channel device turns off, the voltage across the inductor can shoot way way up, thus damaging the transistor.
When you switch an inductor on with a P channel device, there is not any difference except in the polarity of the voltage of the inductor, which can shoot way way down. That means it could be an extremely negative voltage, such as -200 volts, for a short time while the transistor burns out, even if the system voltage is only 12vdc.

To remedy this in the N channel case, you can often get away with a diode across the inductor with the cathode connected to +Vcc. Sometimes you can't however, and you need a snubber as well. The snubber is connected in close to the transistor so it starts to absorb the spike energy right away.
For the P channel case, the diode is still across the inductor, except now the anode connects to ground or the most negative -Vcc in the system. A snubber may still be required.

Sometimes more than one diode is used in series because with just one diode the frequency is limited more than with say three diodes in series. That's because the inductor energy dissipates faster with a higher voltage across it while it discharges. Thus, three diodes are better than one, if you have to use a higher frequency for the on/off cycles of the transistor. In some lower power circuits you may be able to use a zener diode, although there are other ways too.

This is true for a bipolar transistor as well.

"Switching P MOSFET slowly"

Is "slowly" a requirement or a problem with the usual clamp diodes overdamping the shutoff time. (T=L/R)

"the minimum continuous rating for the MOSFET should be?"
There is no Imin except perhaps leakage in the off-state.
The Imax is rated with 1 sq.in. of 2 oz Cu 2-sided copper heatsink at some maximum junction temperature at 25'C ambient. It is a wise idea to derate temperature rise 50% which implies I max continuous and heatsink specs.

e.g. Vgs(th) =-0.4 to -1.0V is rated at 250 uA which means with Vgs=Vds = 1.0V the Rds=1V/250uA = 4 kohm maximum @ 250 uA as the turn-on "threshold".

This is what I've ended up with,

I think I've acted on everyone's advice above.
The Solenoid is 12V and connected to Ground.
Datasheet for the mosfet attached.
Can anyone see any problems with this?
All suggestions greatly appreciated.

Thanks,

Mike.

Looks good from here. My first thought was that R1 is unnecessarily low, but the FET has a large enough gate capacitance to justify it. The time to 95% charge or discharge is around 5 us, plenty fast enough to minimize device heating.

ak

Thanks for the input.
The solenoid will be switched on for either 0.5S, 1S or 2S (think tyre inflator) so switching times shouldn't be a problem (Hopefully).

Mike.

The max Gate-Source voltage of your mosfet is 12 Volts, which is what it'll see in you're circuit. But if the 12 Volts is from a car with the engine running, it'll see more than 14 Volts. Plus whatever transient's there might be.

I'd suggest choosing a mosfet with a higher Vgs. 20 Volt Vgs parts are common.

If you want to stay with that mosfet, then you'll need to add a resistor between the collector of Q1 and the gate of Q2.

Below is an enhanced version of the chart on page 4 of your datasheet.
A 2:1 ratio will give you ~0.038 Ohms. You can use this chart to see what effect other ratios will have.

I hadn't noticed the max Vgs on that MOSFET so have switched to the AP40P05 which has Vgs of ±20V.

Thanks,

Mike.

Pommie said:

I hadn't noticed the max Vgs on that MOSFET so have switched to the AP40P05 which has Vgs of ±20V.

Click to expand...

Alternately, you could use two resistor in series for the gate.
For example 1k between the supply and the gate, in series with 500 ohms between the gate and Q1's collector, will give a maximum Vgs of 10V with the supply at 15V.

Is this a vehicle application ? If so load dump in cars/trucks :






Regards, Dana.

danadak said:

If so load dump in cars/trucks :

Click to expand...

That can certainly cause a large voltage transient, but how often does a vehicle have the battery connection opened when the engine is running?
And how often does that occur in the average automobile.

crutschow said:

That can certainly cause a large voltage transient, but how often does a vehicle have the battery connection opened when the engine is running?

Click to expand...

Usually load dump is the electric clutch for the ACU.

Tony Stewart said:

Usually load dump is the electric clutch for the ACU.

Click to expand...

What's ACU?

That's not the automotive load dump that danadak or myself was referring to.

crutschow said:

That can certainly cause a large voltage transient, but how often does a vehicle have the battery connection opened when the engine is running?
And how often does that occur in the average automobile.

Click to expand...

- this may not be an automotive application, but other dumped loads include air conditioner units, engine starters in cold weather and loose battery connectors being the original condition for the load dump test.

I like the fact that battery clamps are now more reliable replacing lead with plated steel allowing greater compression.

I'd use a 2N7002 instead as someone already mentioned, less current draw from the MCU, swap R2 to 100K pull-down from gate and it's a drop in replacement.

For max 2Hz and the current load stated 1.8k is very over-kill, a 15k will be amble.

If it's for a vehicle AFAIR the norm is 36v although I assume that's also to support 24V vehicles like trucks and buses.