MOSFET Design Review

Does this design appear to be correct?

I am looking to drive an array of 10 LEDs at 1A per LED, using a 2kHz PWM signal at a 5% duty cycle. The circuit will only be run for a few seconds at a time, and may be expanded for more LEDs in the future.

Questions I have are:

  1. Is the calculation for R1 correct? I was not sure if I am supposed to use Rds(on) and Vds(on) in the calculation as I did, or if this is incorrect.

  2. Are the values for R2 and R3 acceptable to allow a 2kHz PWM signal and keep the MCU happy? Microcontroller will either be a PIC or Arduino.

  3. Is it possible to use 10 parallel sets of 3 LEDs in series and omit the power resistor all together?

Thanks in advance.

steve108:

  1. Is the calculation for R1 correct? I was not sure if I am supposed to use Rds(on) and Vds(on) in the calculation as I did, or if this is incorrect.

TL;DR. See answer 3.

steve108:
2. Are the values for R2 and R3 acceptable to allow a 2kHz PWM signal and keep the MCU happy? Microcontroller will either be a PIC or Arduino.

They are standard minimum values, neither "happy" or not. R3 is placed wrongly, should be before R2.

steve108:
3. Is it possible to use 10 parallel sets of 3 LEDs in series and omit the power resistor all together?

Totally insane! The circuit is already badly wrong!

Yes, it is silly to do other than to use sets of three LEDs in series if you are powering from 12V. Each branch of three LEDs must have its own current limiting device corresponding to each LED in the group you illustrate. You can calculate the value of the resistor approximately (ignore the FET parameters, either it turns on fully or you have a major heatsink problem) but you may need to prototype and measure it - which you do not by using a multimeter in 10A mode, but simply by measuring the voltage across the resistor whose value you know.

Just wait till Grumpy Mike gets to this thread - he will tell you to use a current control circuit (transistors) for every parallel branch. :smiley:


And question for you: Why would you drive this at a fixed 5% duty cycle rather than simply using a lower current and more conservative design?

It is for a high speed photography project and I need to get as much light as possible in a very short amount of time

Just use a flash and long exposure in a dark room.
You can easily catch "slow" stuff like falling objects, water droplets splashing around etc.
Put the flash in the lowest power setting (1/128) and you'll get some very short exposures.
Look here for some actual measurements.
http://www.gock.net/2012/01/flash-durations-small-strobes/

Yes, that works but I am using IR LEDs. For high speed photography in complete darkness.

Use a resistor on each series of LEDs.

steve108:
It is for a high speed photography project and I need to get as much light as possible in a very short amount of time

Then you are going the wrong way about it.

You do not feed the gate of that FET with PWM you feed it with a single pulse.
As others have said you need resistors in each parallel channel and with 12V and an IR LED that typically has a Vf of 1.2V then you can have up to 9 in series, with 1.2V left to go across the resistor.

and omit the power resistor all together?

No.

I am looking to drive an array of 10 LEDs at 1A per LED,

So do you have the part number of this IR LED that can stand 1A?

Diito.
Always use a series resistor with LEDS. If you ran the circuit you posted you would blow all the LEDS on
the first shot.
You can parallel up to three LEDS per resistor but not more. The reason paralleling LEDS is strongly
frowned on is that if ONE of the LEDS goes OPEN, the remaining two assume the current of the blown
LED. This scenario can lead to a catastrophic chain reaction that can start with one failure and then
lead to successive failures. Once the current reaches the "tipping point" , the whole thing goes up in
smoke in the blink of an eye.
If you can avoid it, don't parallel LEDS.

SFH4550 will tolerate being driven the way I describe.

The reason for a continuously pulsed input is to get multiple images in a single exposure.

Thanks for the help everyone.

So do you have the part number of this IR LED that can stand 1A?

Forward voltage
Durchlassspannung
(IF = 1 A, tp = 100 μs)

SFH4550

Period for 2kHz is 500 uS,
The device is only rated for 100uS
which would be 10 kHz , not 2 kHz.

Where did you get the 500uS spec ? (because I’m looking at the datasheet and I don’t see it)

@OP,
If it’s Design Review, it is customary to include the part number of the device on a schematic.
Also, if you are asking for a design review, I am compelled to point out that while your math may
be somewhat correct, your understanding of electronics is not. It is customary to use 10 series
resistors (one in series with EACH LED) with the value 4.7 ohm /1 W . The combined resistance
of these resistors will be:

Rp= 1


1/R1 +1/R2+1/R3+1/R4+1/R5+1/R6+1/R7+1/R8+1/R9+1/R10

= 1


[ (0.21276595744680851063829787234043) X 10]

= 0.47 ohm/10W

This prevents a failure in one led from triggering a catastrophic cascade failure of the other leds because
each LED has it’s own current limiting resistors. (sometimes it is better NOT to share)

Just wait till Grumpy Mike gets to this thread - he will tell you to use a current control circuit (transistors) for every parallel branch.

So is Grumpy Mike like the “Boogey Man” we use to scare little children into doing what we tell them ?
:smiley:

See the table 'Permissible Pulse Handling Capacity' on sheet 6. It shows a 1 amp pulse 25us on time at 5% duty cycle.

25/500 = .05 = 5%

1/500us = 1/.0005s = 2000Hz = 2kHz

5% duty cycle of a frequency with a period of 500uS is 5% of the period of the 2khz pulse.
This is not a square wave. This is 25uS ON, 475uS OFF.
Your information is correct, but the waveform you drew to represent the pwm was a squarewave.
(ton=toff)

That's what confused me.

The pulse you are describing would be a very narrow pulse (like a spike) followed by a long OFF time.
Your math is correct.
How would you generate this waveform with an arduino ? (is that what you planned to use ?)

void loop()
{
read_shuttersw()
{
int swrd = digitalRead(shuttersw);
if (swrd ==HIGH)
{
 pulse;

}

void pulse()
{
digitalWrite(FLASH,HIGH);
delayMicroseconds(25);
digitalWrite(FLASH,LOW);
delayMicroseconds(475);
}

Yep that is pretty much what I had in mind. Seems simple enough I may just use a PIC

You should be able to do it with an ATtiny85 . It costs $2 and it supports delayMicroseconds.
You need an arduino UNO or a programmer to program it.

ATtiny85 supported functions

What are your other program criteria ? (what else does the uC need to do ?)