I am working on a design for a 16-channel LED driver. At the moment the basic design is based on three main components:

The circuit needs to initially drive a 110 V LED light with 250 mA LED strings, but should to be adjustable to drive sets of other LEDs e.g Ledengin LZ7


simulate this circuit – Schematic created using CircuitLab

Without further components I think this design should work in principle with M1 getting an RDS(on) of ~30 mΩ (other similar MOSFETs had a larger RDS(on)). The RDS(on) can be reduced further if I can supply a higher voltage (14.1 mΩ at 6 V and 12.6 mΩ at 10 V).

(This combination of R2, M1, Q1, R2, and D1..n are repeated for each channel though R1 will be varied dependent on different D1..n on each channel.)

The PCA9685 can supply a limited current of 10 mA per channel and if I'm reading it right 25 mA total, and it can operate PWM at up to 1526 Hz.

Simulation at 1526 Hz PWM (assuming I did it right) shows if I have R2 = 10 kΩ I get an inrush current of 500 μA which should be no issue.

If I reduce R2 to 1 kΩ I get 5 mA which is likely too large to operate more than 4 channels at once, and 100 Ω gives 50 mA which is too large for a single channel and the PCA9685.

How do I determine the best value for R2?

Would the inefficiency resulting from a large R2 make it worth putting in a gate driver e.g a MAX620EWN+ (would need 4) or a cheaper alternative?

Or is there another cost-effective way to make this more efficient?

Simulation model and output (the New macro are a PLW7070GA Spice model)

simulation model

simulation outputs

  • 1
    \$\begingroup\$ If you only need 250mA of current, then you are really over spec'ing your RDSon. Also, your FET is only rated for 100V, not 110V+margin. \$\endgroup\$
    – W5VO
    Commented Jun 12, 2018 at 21:50
  • \$\begingroup\$ correct me if im wrong but isn't the voltage across the MOSFET (Source - LED Forward Voltage) \$\endgroup\$
    – fireblade
    Commented Jun 13, 2018 at 13:33
  • \$\begingroup\$ so for a 110V supply as long as the voltage drop across the LED array is >10V+margin this wouldn't be an issue ? and would ideally choose a voltage source to minimize the voltage across the mosfet to me more efficient \$\endgroup\$
    – fireblade
    Commented Jun 13, 2018 at 13:39
  • \$\begingroup\$ When the MOSFET is off, the current through the MOSFET and LED chain will be close to 0A. That makes the voltage drop across the LED almost 0V, and the rest is across the MOSFET. Picking a bigger MOSFET (lower RDSon) means you have larger gate capacitance, slower turn on/off, and more power dissipation from switching. Unless you're using a switch-mode driver for controlling LED current, "efficiency" is irrelevant - you will burn the difference between the forward voltage of the LED and your power supply somewhere. \$\endgroup\$
    – W5VO
    Commented Jun 13, 2018 at 17:29
  • \$\begingroup\$ that's interesting as the simulation of the circuit in TINA-TI doesn't show that simulating a set of PLW7070GA leds and CSD19534KCS, i only get a ~0-30V swing across the mosfet is there something wrong with the simulation? (ive added simulation info to post) \$\endgroup\$
    – fireblade
    Commented Jun 13, 2018 at 18:42

1 Answer 1


R1 is going to be more reliable in setting the LED current. Q1 will start stealing volts from the base of the FET when Vbe (=Vr1) is 0.65V. As such, the above will attempt to adjust the LED current to 650mA. R2 is serving to limit this, but the semiconductor parameters keep R2 from easily controlling the current...across devices and conditions, the current limited by R2 will change dramatically.

I'd say stick with R2=100 and R1=0.65V/Iled (where Iled is the current you want to push through the LEDs).

  • \$\begingroup\$ I know that R1 is used to limit the current through the LED, however that is not what i am asking. Im asking about the value of R2 and its effect on the current through the PCA9685 also already stated that setting r2 to 100 would, as i understand it, likely overload the PCA9685, if you believe this is wrong can you explain why \$\endgroup\$
    – fireblade
    Commented Jun 12, 2018 at 21:33
  • \$\begingroup\$ The 100 would stress it by about 150% from spec. This is probably fine for a hobby project but unacceptable for volume production. How much current do you need? You could set R1 as needed, and then just decrease R2 until R1 is in control. \$\endgroup\$ Commented Jun 13, 2018 at 12:11
  • \$\begingroup\$ The reason R2 has such an effect is that with a 1K resistor, you need only 15uA of base current on the 2222 to draw the gate voltage nearly down to the threshold. There's multiple places that could be coming from; without being there, I can't really speculate, but check the board for contaminants and try breaking some paths if you want to chase it down. \$\endgroup\$ Commented Jun 13, 2018 at 12:27
  • \$\begingroup\$ currently there is no actual board an im working from simulations. I am trying to workout how to make the Mosfet work most efficiently while still protecting the the PCA9685. Idealy this would be capable of operating all 16 channels simultaneously which due to single pwm clock would mean a total of upto 16x the inrush current of any individual channel. \$\endgroup\$
    – fireblade
    Commented Jun 13, 2018 at 14:08

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.