High Side MOSFET Check

[joeyjoejoe]

Can someone help me with a high side mosfet switch?

The MCU is 3.3V logic. +BATT is a single cell LiPo, so 4.2V down to, say, 3.6 or so.



Can I use a MOSFET such as the BSS308PE to avoid a high-side driver transistor or MOSFET? It has a Vgs of 2V.



So, when pulled/driven high at 3.3V, worst case with a fully charged LiPo, we have 0.9V Vgs, so the MOSFET will remain off. When pulled low, we have (worst case) 3.6V which is plenty to drive the 30mA load.

[Audioguru]

The datasheet for your Mosfet shows that its gate-source voltage must be at least 4.5V for it to properly turn on. But if you buy many of them and test them all, you might get lucky and find one that is extra-sensitive since some have minimum spec's and some have maximum spec's.
You never said what is the load current.   

[joeyjoejoe]

Thanks Audioguru. I totally did a ninjaedit while you were responding and I apologize. Nobody had read it yet and it wouldn't let me delete the post and make a new one.

[bson]

Your P ch is upside down; the source needs to be connected to the supply and the drain to the load. The body diode faces from drain (A) to source (K).

The main problem with your arrangement is Vgs will vary with battery voltage, which will make it difficult to completely shut it off when the battery is fully charged, and alternatively fully open DS when it's approaching discharge.  What is the goal here?  There may be other ways to accomplish what you're looking to do more reliably and effectively...

[joeyjoejoe]

The goal is to turn off the power supply to the eInk display when it's not in use. The display itself is in module form, and I think it's best that way as I can't get the panel and parts to drive it cheaper anyways. Part of that is it's own 3V3 LDO regulator.

The quiescent current for that regulator is about 150uA. Not a ton, but given that the MCU in deep sleep is 75uA, it's adding 60% to the idle current, and it will be sitting at idle for 99.9% of the time.

Hence, shut it off completely while the MCU is sleeping.

[Audioguru]

The graphs on a datasheet show only a "typical device" that you cannot buy. You need the circuit designed to work properly with devices that have minimum or maximum spec's, or buy many and test them all and maybe you will get lucky to find one that works in your circuit. The new BSS308PE Mosfet has a Vgs(th) of 1V to 2V where it is barely turned on and has a Vgs of 4.5V where all of them conduct fairly and have a Vgs of 10V where all of then turn on very well. 

[joeyjoejoe]

So best to stick to having something drive the gate? Transistor or MOSFET? Any pros/cons to either in a simple application such as this?

[joeyjoejoe]

I'm off for the evening, food for thought and critique. Thanks for the continued help!



DMG2301L - https://www.diodes.com/assets/Datasheets/DMG2301L.pdf
MMBT3904 - https://www.diodes.com/assets/Datasheets/ds30036.pdf

[bson]

Since you have two different voltages it helps to have a complementary pair, like a Diodes Inc DMC2053UVT. 

https://www.diodes.com/assets/Datasheets/DMC2053UVT.pdf

For example:



I've attached the KiCad symbol used above.

Edit: oh, and the exposed battery terminal (P ch source S2, pin 2) must be ESD protected with a TVS diode!  Use a dual TVS diode and add one to S1 also, though not strictly necessary.  And maybe D2 is the load is touchable/pluggable.

[joeyjoejoe]

That looks pretty sweet and the price is right! I assume I should put a pull down on G1 for when the MCU is sleeping to be on the safe side?

[joeyjoejoe]

Can I use something like 100k for R1 to keep the current negligible when enabled (and reduce BOM count as I'm using 100k's in a few places for pullups)

[langwadt]

simple easy safe solution is using an USB supply switch like e.g. http://www.ti.com/lit/ds/symlink/tps25221.pdf

[MarkF]

Here is a circuit I found some time back for a solar panel charge controller doing a similar thing

   

[joeyjoejoe]

@bson, thanks again. I've contributed the DMC2053UVT to the official KiCad repository