UPDATE:

If your MOSFET (or other chip) does not behave well AND you bought from not-really-known seller (even if his/her positive feedback ratio is over 99%) for a very competitive price, MAYBE you got a "factory FAKE" item (means: intentionally produced as FAKE in factory).

This realization took about 3-4 nights to me, because I just started using MOSFETs.... very VERY VERY big eye opening experience.

Original question:

I checked "MOSFET does not turn off", or even "does not turn of totally" related topics. There the solution was usually or other (parasitic way of power), or MOSFET driving problem.

I thought I understand MOSFET (voltage driven and which voltage counts), however I faced the same problem with 2 different simple circuits, so likely I missed something regarding MOSFET basics.

What I wanted to achieve: driving a 12V LED strip (draws about 2A).

1st I used P channel MOSFET - as switch. It was working, BUT very often the LED strip lightly dimmed and vibrated (MOSFET ON was OK, OFF was not perfect). I read somewhere P channel MOSFET gate needs little higher voltage than source to surely turn off..

In both circuits I created them in simulator (drawings came from there), so V1 (dataIN pin) is coming from a 3.3V microcontroller (which I realized only in 2nd circuit).

P channel MOSFET

So I discarded that circuit and made a simpler one with N channel MOSFET: (after taking apart, I checked the P channel MOSFET and .... just now I saw a Youtube video regarding real and false MOSFETs: Youtube regarding REAL and FALSE MOSFETs )

N channel MOSFET

BUT This 2nd circuit does exactly the same. I moved the pin to a PWM capable pin, so now I can dim the LED strip nicely, I can switch on the LED strip as well, but when I switch it off it "starts" to dim very lightly and "blinks" randomly. Again, I can make steady 5%-100% LED power and works nicely, but with 0-1% (I am interested in 0%) it does not switch off firmly.

In 2nd N channel MOSFET circuit I also checked the voltages:

  • logic high = 3 V (MOSFET gate), MOSFET DS = 0,2V - works as expected (however MOSFET datasheet states 2-4V is needed to turn on fully).

  • Logic low = 0,2V (MOSFET gate), MOSFET DS = 5V (7V drops in LED). I shorted R3 (150 Ohms) to make things even simpler. When LED strip started "very dim blinking", I even shorted gate to source (with multimeter) - blink continued and I measured about 1mA <- THIS when I decided to ask help, because this ruined everything I imagined.

I have to admit, that my power source is far from perfect, 2A drain in 12V made it 11,5V, but obviously does not make few volts spikes etc.

Very rarely I achieved the "total off" state (without modifying everything), but within few mins - few hours I got back the low light & blinking LED strip. I was thinking to make some capacitor to eliminate any ripple (if any).

The dataIN is about 50 cm wire, LED strip is 2-2,5 meters, wire to LED strip is about 1 meter. Again, I shorted (in 2nd circuit during logical low dataIN) gate and source - without reaching the desired TOTAL OFF.

  • P channel MOSFET is: IRF9540N
  • N channel MOSFET is: IRFZ44N

I know both has a diode between source and drain (connected as reverse bias), but as I heard it is a byproduct of the MOSFET production and should not make any trouble here.

If anybody could help me:

  • Why MOSFETS does not turn totally off in the above 2 circuits.
  • How can I modify them (currently I have the lower circuit only, but I can recreate the 1st circuit if the solution is there).
  • I added this question finally (after realizing that very likely I inserted a fake MOSFET from China to my circuit) - could be this my problem is? Therefore my 2nd circuit should not able to turn ON a real MOSFET or at least it is very borderline (as stated in datasheet).
  • What are the ratings for the LEDs? It could be that the 1k series res is too high and the LEDs can;t draw as much cuurent as they need. For the PMOS, it should work, albeit at (maybe) slightly higher power dissipation due to the 10k res, but not by much. The NMOS version needs higher gate drive, if datasheet says 2-4V you make it min. 5V. – a concerned citizen Jul 14 at 18:30
  • 3v on the gate is not enough to meaningfully turn on an IRFZ44N - the 2-4 volts is for a 250 micro amp load current. The given graphs for useful performance don't even start until 4.5v on the gate. So you either need a much lower threshold FET or a higher gate drive voltage. – Chris Stratton Jul 14 at 18:35
  • @a concerned citizen, @Chris Stratton Both of you are RIGHT. ... and THIS is why now I know for (almost) sure that I bought a fake IRFZ44N from eBay, because its turns ON very nicely! I will buy a IRLZ44NPBF from reputable local buyer and check... – V-Mark Jul 14 at 19:24
  • If you're concerned about fake parts, try to get them from digi-key or mouser. – Felthry Jul 14 at 19:55
  • What's the source of your LED strips? Do they have some kind of active power management (perhaps to short out defective diodes to keep the rest of the strip running)? I have a failed COB LED light that has an interesting failure mode: it is completely dark unless the power supply is at substantially lower than expected voltage, at which point a handful of the diodes light up and start flickering. I hypothesise that it's an artifact of a system that's trying to compensate for failed diodes but going wrong because there are too many of them. – Jules Jul 14 at 20:22
up vote 2 down vote accepted

I bought some N channel MOSFETs (IRLZ44N) - which has logic level gate drive from a reputable local store (for about 0,36EUR / 0,43 USD per piece). According to datasheet, for 3V gate voltage and 3A drain-source current the resistance goes below 0,1 Ohm...

I changed the N channel MOSFET (and only that) - and VOILA everything works like charm! As I already connected to PWM signal, at even 1% PWM LEDs are nicely (but barely) dim.

Now I am sure that the originally bought IRF MOSFETs from eBay are fake (IRFZ44N were 1 USD/5 pcs; IRF9540N 1,22 USD/5 pcs) - so a price tag difference was (the real in Hungary, including 27% TAX) about 0,23 USD per piece - I start to pull my hair out.... - and always check local sellers next time.

I also bought P channel MOSFETs (IRF9Z34N) and I am quite sure my other circuit would also work with a real MOSFET.

  • To keep in mind, thanks. – Dorian Jul 17 at 17:23

Most likely your LED-s are catching some noise, might be from the DC-DC converter or the WIFI router or even from the power source itself.

The drain source diode makes a half bridge rectifier for the noise leading to a small current going through the led strip.

Place a 1Kohm resistor in parallel with the LED strip and the flickering will go away.

  • I also thought that, but when I disconnect 1 leg of the LED strip (simplified proof of concept MOSFET off) LEDs are totally off. When I connect LED strip to the power source directly, LEDs are all light brightly (this part was working always). – V-Mark Jul 16 at 21:09
  • Whatever the reason is the resistor will turn the leds off. You might be right, tough, can you try with a resistive load to see if there is any drain current? – Dorian Jul 16 at 21:29

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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