Dimming a LED strip - does the MOSFET really switch off?

[Dr. Katzenfritz]

Hello community,

I'm trying to get a 555 based LED dimmer to run which I built with some odd stuff I had lying around, thus the rather strange part/value selection. The schematic is in the first picture and shoud be pretty much the standard circuit - a 555 timer IC generates a PWM signal whose duty cycle is controlled by a potentiometer and switches a p-channel MOSFET IRF9540N which powers the LED strip (connected to J2 in the picture). On the bench it's powered by a lab supply set to 12V (12.2V in the scope screenshots below).

I decided to drive the MOSFET directly with the 555 as I didn't have much space on my small board for a transistor driver, I hope this isn't the root cause of the following problem:

The dimming works in principle, but at minimtum pot setting, it's still very bright. A quick look at the voltage across the LED strip (second picture) shows that during the active part of the cycle is correctly at 12V, but it only falls off to ~7.2V in the inactive time. Meanwhile, the gate-to-ground-voltage (third pic) and gate/source-voltage (fourth pic) look alright to my untrained eye.
If I turn down the supply voltage, the active part voltage goes down accordingly, the 7.2V remain until the supply voltage falls below that level, then the voltage across the LED strip stays constant at the (decreasing) supply voltage level.

Can anyone explain that behaviour to me?

[sleemanj]

Double check you have not swapped source and drain.

Is the FET legit, was it sourced from a legit place, could it be fake or damaged.

Remove the 555 from the equation, tie the gate directly to 12v (that is, just remove R2 from circuit) and see if it turns off then, this will tell you if it's a problem with the fet not turning off hard enough, or if it's a problem with driving it.

How much actual current is passing during the off time.

[Seekonk]

LED act just like a zener.  Place a resistor across them to see what is actually happening.  A LED can operate with just micro amps.  I would lower the 10K resistor to just 1K and put a diode or two in place of the 56 ohm resistor. Likely in off there is enough geate voltage to cause some leakage.

[bson]

You have no current limiting other than the MOSFET channel resistance, so even with only very brief pulsing the LED will be so bright you can't effectively determine the dimming by just eyeballing it.  Add a series resistor that sets maximum brightness.  If that's not enough you can slow down the switching and rely more on the channel resistance by replacing the 56ohm gate resistor with something much bigger.  (1k?  10k?)

[Zero999]

You have no current limiting other than the MOSFET channel resistance, so even with only very brief pulsing the LED will be so bright you can't effectively determine the dimming by just eyeballing it.  Add a series resistor that sets maximum brightness.

It's an LED strip, which typically includes built-in current limiting resistors.

I suspect MOSFET leakage and the fact the circuit can't achieve a low enough duty cycle are more likely to blame.

[mikerj]

I suspect MOSFET leakage and the fact the circuit can't achieve a low enough duty cycle are more likely to blame.

Very likely.  Probably just a few uA of leakage, but enough to keep the voltage pulled up due to the addition of the LED Vfs in series.  Temporarily connecting a lowish value resistor across the LED strip (e.g. 2k or so) would prove this.

[Seekonk]

Sure, if you want to call the gate being driven leakage.

[Dr. Katzenfritz]

Sorry for the late response guys, and thanks a lot for all your suggestions! I will look through them, give it a try at the bench and report back