P channel mosfet body diode question.

[Sounduser]

Hi, I want to switch between two power supplies on the high side. (Battery and PSU with common ground)

When the battery is too low I want my Arduino to switch to the PSU.

I found a diagram using an NPN transistor and two resistors to switch the gate of a P Channel mosfet on and off.

Now if I connect the two mosfets together on the load side, will the body diode allow current to flow backwards from the battery to the psu and the other way around?? I dont want to use diodes in the circuit because of the voltage drop.

Thanks.

[suicidaleggroll]

Yes
You can get around this by using two P-ch mosfets back to back.  It will double your Rds(on), but if you use devices with a reasonably low Rds(on) it shouldn't be a big issue, and will still be significantly better than a diode.

[Sounduser]

Okay thanks.
So power + D S - S D load Gnd?

[JohnnyBerg]

Have a look at the schematic of the Arduino UNO, it has a circuit that switches power between the barrel jack and USB.

AFAIK It uses 1 P channel mosfet and a LM358 as a comparator.

[suicidaleggroll]

Okay thanks.
So power + D S - S D load Gnd?

Yes

Here's an example:

From this page:
http://www.electronic-products-design.com/geek-area/electronics/mosfets/using-mosfets-as-general-switches

When the control is high, current can flow in either direction through the switch.  When the control is low, current can not flow in either direction.

[Sounduser]

Thanks. I'll have a play.

[Sounduser]

What kind of voltage drop should I expect? I quickly bread boarded the circuit and had a 0.5 volt drop? 

[Dave]

What was the current flowing through?
What was the supply voltage?
Which FETs did you use?

[HackedFridgeMagnet]

Okay thanks.
So power + D S - S D load Gnd?

Yes

Here's an example:

From this page:
http://www.electronic-products-design.com/geek-area/electronics/mosfets/using-mosfets-as-general-switches

When the control is high, current can flow in either direction through the switch.  When the control is low, current can not flow in either direction.

but the turn off will be slow, will that make the fet overheat?
you have to make sure this arrangement is protected from a shorted load, a lot of the power will be absorbed by the not fully turned on FETs, but I guess you would have had to protect against this anyway.

[mikerj]

but the turn off will be slow, will that make the fet overheat?
you have to make sure this arrangement is protected from a shorted load, a lot of the power will be absorbed by the not fully turned on FETs, but I guess you would have had to protect against this anyway.

Switching speed is usually only a concern where you are turning the FET on and off very quickly, e.g. PWM, or where the expected currents are very large.  In this case the power dissipation is going to be negligible for single switching events.

Of rather more concern in that circuit is the absolute maximum Vgs is specified as 20v, so those MOSFETs are just waiting to go pop with a 20v input.  Clamping Vgs with e.g. a Zener would fix this.

[Sounduser]

IRF9510 and between 12 and 14.4 volts.

Cheers

Oh and the draw is around 650ma.

[suicidaleggroll]

Yikes - that FET has an Rds(on) of 1.2 ohms.  Find one of the many, many options with a more reasonable Rds(on) and that voltage drop will all but disappear.

For example, I'm using the IPP80P03P4L in one of my systems.  Vgs max of -16 (if your system runs at 14.4, you may want to use a voltage divider or zener as mikerj suggested to keep away from that limit), and an Rds(on) of .0037 ohms at Vgs=-10.

[Sounduser]

I admit they're the first ones I saw on ebay.

[Sounduser]

Could anyone recommend a good low rds on fet for me? Not worried about package too much. But I'll want about 20 so don't want to pay too much  for each one.

Something around 20 volts and 5 amps.

Thanks.

[suicidaleggroll]

digikey and mouser both have good search functions that allow you to specify your desired max Vgs, Vds, Rds(on), package type, etc.  Filter it to "in stock", sort by price, and pick your favorite one.

[HackedFridgeMagnet]

Did a quick sim and the peak power was 18W on turn off and this power averaged for maybe around a 10ms turn off time.
Sure it's probably not an issue in this case, but I wouldn't call it negligible.

[suicidaleggroll]

470k is a bit much for the gate pull up, I'd change that to maybe 50k.  Should help accelerate the shut off without wasting significantly more power.