So, to preface, I have a grow lamp setup with dimmable leds. There is a teensy 3.5 controlling twelve FQP30N06L mosfets which drive twelve different vero 29 COB leds at around ~70 watts each. There are four different mean well lrs-350-48 (350 watt 48 volts) power supplies, split with 3 led's each. The default pwm frequency of the teensy is 488.28 hz. The leds are being dimmed all of the time so that I can control the color temperature within the room to simulate day/night cycles as well as winter or summer lighting. For example, in a set of 3 there might be two 5000k leds and one 2700k led, and the 5000's are running at 60% duty cycle while the 2700 is running at 30% duty cycle, etc, so that the load of all three never exceeds 300 watts. The issue this creates, at 488.28 hz is that it is impossible to film anything in the room due to horizontal banding and shutter interference with my gopro. Since gopro doesn't allow you to adjust shutter speed for timelapses, I was hoping to increase my pwm switching frequency. However, I noticed that my FETs heat up quite a bit at frequencies higher than 1000 hz, and at 7000 hz one of the power supplies failed. So basically, I need to figure out what the ideal switching frequency is for LED lighting and wanted to ask if anyone else has had experience dealing with this issue, and what you did to solve it.
TL;DR Need to film timelapse in room with LED lighting. Need to pwm led's for dimming control. Teensy 3.5 default switching frequency is 488.28 hz, which causes shutter interference/banding lines with go pro. Frequencies higher than 1000 hz cause mosfets to heat up. 7000hz caused power supply failure. What frequency do you guys use to pwm leds?
If the FQP30N06L stay cool at low frequencies and get hot at higher frequencies, I think that must mean the teensy pins do not have enough drive current to switch the mosfets on & off fast enough, so they are spending a high proportion of their time in a partly-on state, which has a higher resistance, and hence the heat problem. Perhaps you need some mosfet gate drivers between the teensy and the mosfets. These are basically current amplifers, able to source or sink more current to get the mosfet gates to switch fully on and fully off faster, so less time in that partly-on state, and less heat.
Here is a simple driver circuit. R2 5,6 - 10K, R3 should be between 27 and 50 Ohms. T1 and R6 should be about 100 Ohms although it will work just fine without them. This should drive the FQP30N06 as fast as the Arduino will let you. What happens you lose drive through the assumed resistor connected between the arduino and the gate of the MOSFET. A schematic would have saved the guessing. The current needed to charge the Miller capacitor etc is in the amps when you start switching fast, your design limits it by the available charge current. When it is in the linear range it works just like a resistor and dissipates the energy as heat. Be sure the power supply can handle a PWM load, not all can. If you put a scope on the gate you will see it never gets to the point of fully saturating the MOSFET at high frequency.
PaulRB:
If the FQP30N06L stay cool at low frequencies and get hot at higher frequencies, I think that must mean the teensy pins do not have enough drive current to switch the mosfets on & off fast enough, so they are spending a high proportion of their time in a partly-on state, which has a higher resistance, and hence the heat problem. Perhaps you need some mosfet gate drivers between the teensy and the mosfets. These are basically current amplifers, able to source or sink more current to get the mosfet gates to switch fully on and fully off faster, so less time in that partly-on state, and less heat.
Thank you for your response! the teensy is certainly limited in voltage (3.3v) and current (30ma I think), so that makes sense it could be an issue at higher frequencies.
gilshultz:
Here is a simple driver circuit. R2 5,6 - 10K, R3 should be between 27 and 50 Ohms. T1 and R6 should be about 100 Ohms although it will work just fine without them. This should drive the FQP30N06 as fast as the Arduino will let you. What happens you lose drive through the assumed resistor connected between the arduino and the gate of the MOSFET. A schematic would have saved the guessing. The current needed to charge the Miller capacitor etc is in the amps when you start switching fast, your design limits it by the available charge current. When it is in the linear range it works just like a resistor and dissipates the energy as heat. Be sure the power supply can handle a PWM load, not all can. If you put a scope on the gate you will see it never gets to the point of fully saturating the MOSFET at high frequency.
Thank you, this makes sense. I may see if I can find a cheap scope to try and obtain a better understanding of what exactly is going on. I do believe the power supply might be an issue, even before the mosfets overheat. I did have one of them fail at 7000 hz, although I can't guarantee it was the fault of the pwm. I will have to look into it more and see if I can find a schematic of the mean well lsr-350-48, but I haven't had much luck. Unfortunately, a power supply that can handle the higher pwm will probably be way more expensive. I feel like my best choice at this point is going to be to stick to the lower pwm and find a camera that can adjust its shutter speed to match. Thank you for helping me work through this, much appreciated.