MOSFET voltage drop when turning on

[jacau]

I got around to building a circuit with a  P-Channel MOSFET to act as a voltage switch (controlled via MCU GPIO) that I researched about 6 months ago.

I recall the importance of a limit resistor on the gate to stop FET capacitance from damaging the MCU pin but I still have a large 0.5-0.8 volt drop when the GPIO is driven low and the FET turns on.

There was a 1000uf cap on the drain (nothing else connected at this stage apart from a 100K drain resistor) but when I took that out of the circuit, the drop was still there.

Edit: Those resistor values was just what I had during testing and the circuit worked - but my test battery was 3.5volts so wasn't dropping low enough to reset the MCU .

I have included my schematic below, have I misunderstood how to use a FET or did I choose the wrong part? (Nexperia BSP250,135  https://au.mouser.com/datasheet/2/916/BSP250-1594323.pdf)

Any suggestions?
Thanks

[ Specified attachment is not available ]

[wasedadoc]

[ Specified attachment is not available ]Do you understand this parameter?

[Zero999]

Insufficient gate drive. The BSP250 requires 4.5V to give a low on resistance.

A 1000µF load is also a lot of capacitance for s small MOSFET.

[Terry Bites]

See the charts for Drain-source on-state resistance as a function of gate-source voltage typical values. assets.nexperia.com/documents/data-sheet/BSP250.pdf
As you can see, a VGs of -3.3V is marginal, you cant rely on it turning on properly. Note that a 100k load resistor isn't going to give you representative results. You need simulate your load with a resistor that draws the expected load current.

Look for devcies that meet your Rds @ Vgs needs eg. https://www.vishay.com/en/mosfets/p-channel

At 3.3V its not easy to find a mosfet that'll meet your spec and be a commonly available/ economical part. You might be better off with a made for purpose load switch eg TPS22967.

[wasedadoc]

Points that many overlook is that typical data sheet tables, including the one for the MOSFET in this topic, specify the Vgs threshold under the condition that Vds = Vgs and with quite small drain current.  (1 mA in this case.)

[Simon]

If you are not looking at graphs on the datasheet then you are looking at lies. The headline specs are utter bullshit, the electrical specs are half lies, the graphs at the bottom tell you the truth but this requires you to understand how the part works and to build up a model in your head of how it works based on the information in the graphs.

[Zero999]

If you are not looking at graphs on the datasheet then you are looking at lies. The headline specs are utter bullshit, the electrical specs are half lies, the graphs at the bottom tell you the truth but this requires you to understand how the part works and to build up a model in your head of how it works based on the information in the graphs.

Even the graphs should be taken with a pinch of salt, because they often only show typical characteristics, not the worst case and are under a certain set of conditions. Electrical specifications such as maximum on resistance are more reliable.

[Doctorandus_P]

If you're learning about mosfets (or any other part of electronics), then start with building some simple things on a breadboard. Take your part, put some voltages on it and see how it behaves. Read the datasheet and make correlations between your own measurement and what you read in the datasheet.

[Simon]

If you are not looking at graphs on the datasheet then you are looking at lies. The headline specs are utter bullshit, the electrical specs are half lies, the graphs at the bottom tell you the truth but this requires you to understand how the part works and to build up a model in your head of how it works based on the information in the graphs.

Even the graphs should be taken with a pinch of salt, because they often only show typical characteristics, not the worst case and are under a certain set of conditions. Electrical specifications such as maximum on resistance are more reliable.

Yes which is why I say try to build a model of how it will work as the graphs only give one thing versus another, there are multiple influences at the same time. I ended up writing a calculator in a spreadsheet to calculate the static current I can pass based on my working ambient, the maximum the mosfet can take internally and the thermal resistance of the cooling solution and I still regard the result of the combined calculation as just a guide but it does immediately show that the headline current handling figures are totally bollocks, at best you may get a quarter or less sometimes.

[gamalot]

did I choose the wrong part?

Any suggestions?

- did I choose the wrong part?

Yes, a logic level MOSFET would be more suitable for your needs.

- Any suggestions?

No,  at least you have to let people know what the operating current of the load is.

[Simon]

Any MOSFET really likes 10V to fully close it, even at 5V you won't get the lowest channel resistance. Finding one that will work at 3.3V means finding one that was made to work at these voltages, I would carefully look at the datasheet because even labelling it as "logic level" does not properly qualify it. Logic level is a loose term to describe MOSFETS that are "useable" at 3.3V. Back in the day I had to show a subcontract design engineer how to fix the problem we had, swap a IRF540 for an IRL540, I don't know if given the similarity in name they are made as the same part and then 'binned" based on performance but the IRL had a threshold of 2-4V, this still meant that they were talking about 100µA at 2-4V and really both parts still wanted 10V to fully close.

[jacau]

I see I need to do a lot more learning on FETS and all the specs in the datasheet.

Just for completeness to my question:

I built a lithium powered remote sensor with a NBIOT module. My original design used one of the common 4 pin 3.3V boost modules so I could use a 3V lithium battery to power MCU and then ensure enough power to turn on boost and send data via NBIOT.

Worked well but I could get my quiescent current under 100 uA just doing away with the boost module and finding an NBIOT module that down to 2.1V (SIM7020). Had 2000uA of caps and lots of leakage there but that's another lesson learned.

I had two instances where my cheap NBIOT module from aliexpress ending up leaking current from some SMD caps so I thought using a FET as a switch would give me even lower consumption AND help protect from such a thing happening.

So I read up on FETS and built a little PCB with the same footprint as the boost modules so I could retrofit. I liked the BSP250 because it had low leakage when off but as you can all see, I really have NFI. So when I soldered it up and mated the system to a semi-discharged LI-ion at 3.5V and it just worked, I just made the mistake of assuming it was that easy and all sorted. My plan was 3.6V Lipo4 which is why the battery change from my original 3V lithiums.



Time to do the thing I should have done in the first place and get a better understanding.

Thanks for the replies.

[gamalot]

I searched and found that the only PMOS in SOT223 package I have is ZXMP7A17G, which is not a logic level MOSFET, but the on-resistance at low gate voltage is lower than that of BSP250.

Tested it on a breadboard, the power supply voltage is 3.3V. When a current of 320mA is passed through a 10ohm load resistor, the voltage drop across the ZXMP7A17G is about 85mV. When the gate is pulled high with the 10K resistor (with the 910ohm pull-down resistor disconnected), the leakage current is about 23nA. I guess the result is good enough for the OP's needs.

https://www.diodes.com/assets/Datasheets/ZXMP7A17G.pdf

[Simon]

Any mosfet should have low leakage when off, the resistance goes up to levels hard to measure with even a good multimeter.

[jacau]

I know I'm probably misinterpreting some of these replies but I really appreciate the responses and don't want to waste peoples time.

I just wanted to confirm I'm not talking about a voltage drop across S and D (but I see where that is also an issue now) but a voltage drop from the battery that pulls the whole circuit voltage low and resets the MCU.

If I've misunderstood this than apologies.

[jacau]

I searched and found that the only PMOS in SOT223 package I have is ZXMP7A17G

Thanks gamalot for the testing and the link to the datasheet.

I can see from the graphs (as Simon pointed out) that the voltages and current I need stay within the specs for that part.

With my current PMOS, using a 5ohm load resistor on the drain and adding a 47uf cap on the gate (and GPIO PWM to ease it on) I get 400mV drop @3V/550mA to my NBIOT module and 300mV brief drop when the PMOS turns on which is suitable for both the MCU and NBIOT module.

So everything works now so I can test the rest of my circuit while I choose a better suited PMOS.

Thanks again everybody.

[Zero999]

I know I'm probably misinterpreting some of these replies but I really appreciate the responses and don't want to waste peoples time.

I just wanted to confirm I'm not talking about a voltage drop across S and D (but I see where that is also an issue now) but a voltage drop from the battery that pulls the whole circuit voltage low and resets the MCU.

If I've misunderstood this than apologies.

The original post implied it was the voltage drop across the MOSFET and you did say the test was 3.5V so wasn't dropping low enough to reset the MCU.

If the battery voltage is dropping too low, then it's because the battery's internal impedance is too high and is dropping a voltage. This could either be because the battery is inadequately specified or bad and needs replacing. The MOSFET is a distraction.

[jacau]

The original post implied it was the voltage drop across the MOSFET and you did say the test was 3.5V so wasn't dropping low enough to reset the MCU

Yes my error, I removed a paragraph from my OP that also included that my standard battery was a 3V primary lithium so that extra 500mV stopped the MCU brownout. Hard to get help when I'm missing information and not clear.

If the battery voltage is dropping too low, then it's because the battery's internal impedance is too high and is dropping a voltage. This could either be because the battery is inadequately specified or bad and needs replacing. The MOSFET is a distraction.

Thankyou for confirming that and helping me solve my issue! After reading your post, I discovered one of AWG14 alligator cables had a dry solder joint. Replacing that still sees the PMOS drop (to be expected I now know) but now no battery voltage drop!

So battery voltage drop is within what I can measure and voltage drop across the PMOS is 220mV @ 600mA. Well within the specs of my NBIOT module so I can continue testing.

Looking at a BSS215 series as they work within my requirements at 2.5V but will do more research.

Thank you all again.