I will be using driving N-channel Mosfets with 5v PWM to switch 24v 2A(per channel) LED strips. I will use four per strip (RGBW) and have 6 strips, totalling 24 Mosfets on the PCB.

I am looking at using the FAIRCHILD SEMICONDUCTOR FDD8878

  • Transistor Polarity N Channel
  • Continuous Drain Current Id 40A
  • Drain Source Voltage Vds 30V
  • On Resistance Rds(on) 0.011ohm
  • Rds(on) Test Voltage Vgs 10V
  • Threshold Voltage Vgs 1.2V
  • Power Dissipation Pd 40W
  • Transistor Case Style TO-252AA
  • No. of Pins 3
  • Operating Temperature Max 175°C

I have calculated the temperature rise to be 6°C

I x I x R = W

2 x 2 x 0.015 = 0.06W

Junction-to-Ambient thermal resistance = 100°C/W

0.06 x 100 = 6°C

Would this be a good choice Mosfet? Reliability and cool running without heat sinks is my main concern.

For comparison a SOT-23 Mosfet I was looking at:


  • Transistor Polarity N Channel
  • Continuous Drain Current Id 5.8A
  • Drain Source Voltage Vds 30V
  • On Resistance Rds(on) 0.03ohm
  • Rds(on) Test Voltage Vgs 10V
  • Threshold Voltage Vgs 1.2V
  • Power Dissipation Pd 2.1W
  • Transistor Case Style SOT-23
  • No. of Pins 3
  • Operating Temperature Max 150°C

The temperature rise for this would be: 14.4°C

2 x 2 x 0.036 = 0.144W

0.144 * 100 = 14.4°C

This would save a lot of space on the PCB, but I would be concerned that it wouldn't dissipate the heat adequately.

  • \$\begingroup\$ I think the second one should be adequate. The 30V is also adequate because the FET would not see any real voltage even close to 24V because of the voltage drop from the LEDs (assuming the FET and LEDs are in series). Bigger MOSFET (higher current or voltage) has larger capacitance which leads to slower switching time and larger switching loss. If you are switching at 10 KHz or below then switching should not be a significant concern anyway. \$\endgroup\$ – rioraxe Mar 16 '16 at 18:34
  • \$\begingroup\$ @rioraxe The MOSFET will see all of the 24V when the LEDs are off. \$\endgroup\$ – Phil Frost Mar 16 '16 at 18:54
  • \$\begingroup\$ @rioraxe I'm using a PCA9685 which has adjustable switching from 24 Hz to 1526 Hz, so I guess there's no problem there. \$\endgroup\$ – davivid Mar 16 '16 at 18:58
  • \$\begingroup\$ @PhilFrost I agree that the MOSFET should be spec for the whole 24V. The off voltage is going to depend on the balance of the leakage current of the LEDs and MOSFET when both are off. The main thing is, 30V MOSFET is fine for switching 24V non-reactive load. There is extra cushion because there will be some voltage drop across the LEDs but that is secondary. \$\endgroup\$ – rioraxe Mar 17 '16 at 2:55

144mW isn't much power and 14.4°C isn't much of a temperature rise.

Subjectively, anything below 50°C is "quite cool". If you expect this device to nominally operate in an ambient temperature of 36°C or less, you will still be below 50°C.

And even if you are not, you still have a 100°C margin until you hit the maximum operating temperature. I'd say the SOT-23 device is quite adequate.

  • \$\begingroup\$ Great I'll start with that then. On another note does the 30V rating give enough headroom for the 24V supply? \$\endgroup\$ – davivid Mar 16 '16 at 13:40
  • \$\begingroup\$ @davivid I'd be inclined to look for a higher rating, since it may be difficult to know if the maximum drain to source voltage will be exceeded under transient conditions. In particular consider that the LED string has some inductance: depending on how fast you turn it off you may get some high transient voltages. You can carefully analyze all this stuff, or you can just spec the components with plenty of headroom and call it a day. For hobby projects where saving 5 cents is not significant, I'm most inclined to do the latter. \$\endgroup\$ – Phil Frost Mar 16 '16 at 16:46
  • \$\begingroup\$ Thanks for the advice. Would a voltage rating of say 50V be a good figure? or should I go higher? The led strings are going to be continually pwm'd from 0-100% \$\endgroup\$ – davivid Mar 16 '16 at 18:43
  • \$\begingroup\$ @davivid Yeah that sounds like a decent margin, absent a more detailed analysis. \$\endgroup\$ – Phil Frost Mar 16 '16 at 18:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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