Edit: Photo below.

We are using an N-channel MOSFET for load switching.

This is the schematic:

enter image description here

The gate is driven through MCU GPIO pin (3.3V.) Vdd on the drain side is also 3.3V. Due to the current circuit topology, when the pin is high, Vgs should be 3.3V, which is higher than the threshold value for the MOSFET, and everything should be ok.

I have noticed two strange things:

  1. When the GPIO signal is high, I measured the voltage between the gate and the source with the multimeter, and surprisingly the values that it shows are not 3.3V, but 0.7V The transistor is functioning properly, though. If I separate the connection between the GPIO and the gate, then I can read the expected 3.3V value. Is it normal or does something decrease the GS voltage?
  2. In the datasheet, the Rds should be around 50mOhm. I measured the voltage drop across the drain and source legs of transistor, it is in mV scale. Then I measured the current flowing through the load. I divide voltage over this current, and get around 600mOhm, which is higher than in the datasheet. I want to understand the reason for this.

If GS voltage is to be considered as 0.7V, then perhaps such high resistance is normal because it is below the typical threshold value. How can it be 0.7V if I provide 3.3V on GPIO pin?

We are using these MOSFETs for different kinds of loads (LEDs, ICs and etc.) The stange behavior that I described (high resistance and insufficient GS voltage shown on multimeter) is present in all of them.

P.S. The MCU is an nRF52. enter image description here

  • 2
    \$\begingroup\$ Can you include a photo of your physical set up? Did you buy your parts from a reputable source or a random eBay/AliBaba seller? \$\endgroup\$
    – The Photon
    Aug 11, 2021 at 14:16
  • 3
    \$\begingroup\$ Something is wrong. Even though this MOSFET has a Vgsth of 0.7V~1.2V, it is not a diode like a BJT. It shouldn't lock Vgs to that voltage. It should be the 3.3V you are driving it with. Where did you buy this? \$\endgroup\$
    – DKNguyen
    Aug 11, 2021 at 14:17
  • 2
    \$\begingroup\$ BTW, you do not care about threshold voltage (Vgsth) when the MOSFET just barely conducts. You care about Vgs to achieve rated RDson. Even 10V MOSFETs have a Vgsth of only a couple of Volts. Unsure if you got lucky or worded your post wrong. But most would have ended up with a non-logic level MOSFET thinking the Vgsth = 1V was enough to be driven by 3.3V. \$\endgroup\$
    – DKNguyen
    Aug 11, 2021 at 14:21
  • 2
    \$\begingroup\$ No, a clear photo of the markings on the transistor. \$\endgroup\$
    – Andy aka
    Aug 11, 2021 at 15:12
  • 1
    \$\begingroup\$ I have updates. The initial information given to me regarding origin of these parts was incorrect. These parts are not from digikey, but another supplier that is possibly unreliable. I replaced current mosfet with another model n-channel mosfet that we had in our possession, and this time everything was ok. The Vgs was 3.3V with another model. Thank you for your assistance. \$\endgroup\$
    – UserRR
    Aug 12, 2021 at 11:26

2 Answers 2


When performance defies logic and nothing seems normal, validate your assumptions and compare with datasheet.

The test results indicate an inverted view of the part with the diode on the gate rather than on drain, yet your fuzzy photo implies correct orientation.

Start with an ohmmeter on part removed and verify the substrate diode is on the correct pins and polarity.

enter image description here

ref http://www.aosmd.com/pdfs/datasheet/ao3400.pdf


The MOSFETs that we have been using were bought from unreliable source, and my conclusion is that they are fake parts. After replacing the MOSFET with different model N-Channel Mosfet, everything was ok this time. The Vgs values were 3.3V as expected.

Thanks to everyone for your help.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.