EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: girishji on January 15, 2023, 11:42:47 am
-
Strange behavior I am seeing. I am using a N Mosfet as a low side switch (see datasheet below). Source at ground, Drain connected to load at 3.3v. I expect Vds=3.3v when switch is OFF and ~0v when it is ON. This particular mosfet has 1.1V across DS when Gate is connected to ground (OFF)! It turns on fully (Vds=0V)when G is supplied, but it cannot work as a switch. I have bunch of other N Mosfet from other manufacturers and they all report Vds=3.3V when G is at 0, which is expected. Before I throw these away, anything I am missing?
https://www.tme.eu/Document/d39f23a42c12123e6ff90b0cbdb3e54b/BSH105.pdf (https://www.tme.eu/Document/d39f23a42c12123e6ff90b0cbdb3e54b/BSH105.pdf)
-
With a very low threshold the leakage current of this fets is relatively high. So 0 V is not yet in the fully off range. There should not be very much leakage current with 0 V at the gate, but it could be more than one is used from other types. Below the threshold the current does not go instantly to zero but is more approximated with a exp(U*e/kt) curve.
The DS calls for a typical 50 nA leakage current with 0 V at the gate. This is quite a lot for a small fet.
-
So the gate/source voltage is zero? Did you measure it? Does this behavior go away if you use a lower value gate pulldown resistor?
-
What was the drain resistance?
What's the meter's input impedance?
-
What resistance is the load? When the transistor is off it acts like a high value resistor, if your load is also high resistance then you will make a voltage divider.
-
I can replace the above mosfet with this one (https://www.mouser.pl/datasheet/2/115/DIOD_S_A0009865515_1-2543348.pdf (https://www.mouser.pl/datasheet/2/115/DIOD_S_A0009865515_1-2543348.pdf)) and it works correctly as expected. Both have very similar characteristics on paper. I also have a BSS138 and it works correctly.
Also, I tried another BSH105 and it has the same exact flaw. Unlikely I have 2 bad copies.
(For those asking whether Gate is really at zero etc., yes it is. I can even connect gate to ground directly, with or without a resistor (160R, 450k, does not matter), and Mosfet lets some current flow through DS in quiescent state. Load is just a LED and resistor (2k). I have a scope attached to all those terminals)
-
Only LED + 2K resistor? Hmm that's weird.
Do you have one of those cheap transistor testers + LCR meters you could put the part on? Maybe it's internally miswired or not even a mosfet.
Alternatively it could also be something else in your setup (tiny metal bridges between pads/traces, lots of RF power in circuit or nearby, etc). Might be worth cutting some slots in a bit of plain copper clad board (or use copper tape) and solder the SOT23 to a 2k resistor + LED there, then use a different power supply to power it.
-
Schematic of your circuit pls.
-
Does the LED light up when the gate is grounded? If not, maybe it's just because your MOS has very low threshold, nothing weird.
https://www.planetanalog.com/very-low-gate-threshold-fet-maybe-too-low/ (https://www.planetanalog.com/very-low-gate-threshold-fet-maybe-too-low/)
-
Don't panic, I can easily repeat it with a low threshold MOSFET (FDN337N), and the current reading on the sourcemeter is about 580nA (will drop to 300nA if I disconnect the Fluke 117 which is measuring 3.3V).
https://www.onsemi.com/pdf/datasheet/fdn337n-d.pdf (https://www.onsemi.com/pdf/datasheet/fdn337n-d.pdf)
---
Because neither the LED nor the MOSFET is turned on, the drain of the MOSFET is in a state close to floating. The input impedance of the multimeter will affect the measurement results.
I did further measurements with a Keysight 34465A and measured the drain voltage at about 1.1V at the normal 10M input impedance, and about 1.9V when switched to high input impedance (>10G).
-
I don't think you mentioned if the source lead was connected directly to ground ? I am assuming that it is connected to ground.
The Vg-s is what matters. Just go make sure, was looking for the words "gate to source" voltage here.
Thanks,
boB
-
Almost any BJT will be able to drive a LED, so no special MOSFET is required.
-
I think I nailed the problem based on what gamalot and kleinstein have said.
This low threshold mosfet does not behave well when Drain has <1mA flowing through it when switch is open. I connected Drain to a variable resistor with 3v3 supply. At 300R mosfet behaves well (first pic, ch2, Vmax=3.28v), and at 4k not so well (second pic, ch2 1.68v).
(Eventually this switch will be handling about 50mA at the drain, so this is not a problem.)
In last pic, GPIO pin (of a xiao ble) is connected to Gate through 160R and it is set high/low every 2ms. On a different note, anyone know how to quickly turn off a Mosfet? This one turns on in a few nanosec but takes >1us to turn off. Haven't figured out how best to discharge the gate capacitor quickly. Maybe I will post a separate question.
-
Probably not related to your issue but I thought i'd mention it.
I notice you have two wires sitting on your laptop in that photo.
It's pretty common for laptops running on mains to get some floating mains charge on their case, even when it's plastic.
You can sometimes feel this in your fingers as the surface 'feels' different when plugged in charging. Kind of like the friction of the surface changes. It can feel quite odd sometimes.
Mosfet gate wires near a laptop like that, even when insulated, could easily cause confusing things to happen.
-
In this case laptop is not connected to power. One of the USB-C is connected to Owon oscilloscope, and another one to xiao board. I did notice different feel when running fingers across surface of laptop (it is aluminum) when connected to main, and I wondered why. Thanks.
-
SMPS power supplies usually don't have perfect isolation from mains. For EMI protection they usually have 2 small capacitors (e.g. 5 nF range) to neutral and the live wire. This leads to a capacitive couples voltage of half the mains voltage. If there is 3 pin plug / connection this voltage could be reduced by a ground link.
-
I think I nailed the problem based on what gamalot and kleinstein have said.
This low threshold mosfet does not behave well when Drain has <1mA flowing through it when switch is open. I connected Drain to a variable resistor with 3v3 supply. At 300R mosfet behaves well (first pic, ch2, Vmax=3.28v), and at 4k not so well (second pic, ch2 1.68v).
(Eventually this switch will be handling about 50mA at the drain, so this is not a problem.)
In last pic, GPIO pin (of a xiao ble) is connected to Gate through 160R and it is switching on/off every 2ms. On a different note, anyone know how to quickly turn off a Mosfet? This one turns on in a few nanosec but takes >1us to turn off. Haven't figured out how best to discharge the gate capacitor quickly. Maybe I will post a separate question.
Because it is a single ended drive, the turn off time must be slower than the turn on time, and choosing a smaller pull-up resistor will improve it.
But if you get turn off times >1us, maybe that's not the gate capacitance at play, it's more like the capacitance of your oscilloscope probe, switch it to the X10 position if it's on the X1 position.