Looking for a PWM dimmable LED driver

[benyig]

Hello,

I want to control the dimming of several power LEDs and LED strips at once. Normally, I would do it using a transistor like a MOSFET, but the estimated high current makes it impractical due to too much heat.

Instead, I'm looking for a PWM dimmable LED driver that I can control from the PWM capable digital pins on my Arduino board. Constant current and constant voltage drivers are both OK. So far I couldn't find any that seems to be compatible. But I'm probably missing something

Are there such drivers available at all?

Thanks for any advice!

[PaulRB]

Power LEDs require constant current drivers. Led strips require constant voltage drivers or power supply. I don't think you can use the same drivers for both.

[benyig]

I don't plan to. We have constant current LED strips too (like this one ). We can leave the power LEDs out and use only constant voltage LED strips so we can adapt to both situations, whichever is the better solution.

[Grumpy_Mike]

I would do it using a transistor like a MOSFET, but the estimated high current makes it impractical due to too much heat.

Then all you need is a better FET. Look for ones with a low drain to source resistance when turned on. The lower this is the less power you burn when you switch it on.

If you think this is wrong then post your calculations as to how much heat your FET has to dissipate.


I don't think you can dimm those constant current strips.

[benyig]

I admit I'm not an expert. All I know is we tried to use the IRLZ34N at around 1.7A, 12V CD and it went to 60-70 C very fast although it is supposed to be a low drain-to-source resistance FET. It seems enough to be concerned about when it will have to handle 8-10A continuously in the full setup. But maybe it's nothing to worry about, I don't have enough experience with this kind of light control, coming from the world of DMX.

[Grumpy_Mike]

No 1.7 Amps will only give you a power dissipation 0.12W, so you must have done something wrong.

[benyig]

Thank you, you may be right. But my original question still stands as it would be much easier to just buy LED drivers that can be controlled by an Arduino and don't need the extra tinkering.

So, are there LED drivers like that?

[Wawa]

We have constant current LED strips too (like this one ). We can leave the power LEDs out and use only constant voltage LED strips so we can adapt to both situations, whichever is the better solution.

That strip seems to be a mix of common CL resistors and a tiny linear regulator (I don't see any inductors).
I would still call this a constant voltage strip (with a little bit of regulation).

I don't see any voltage window limits in the specs where the regulation would still work or fry the linear chip.
This strip should be PWM-ed low-side with a locic level mosfet.


There are many low voltage CC LED drivers available for the Arduino,
but they need to be matched to the LED you're driving.
Give us a link to the 'other' LEDs you want to control.
Leo..

[benyig]

That strip seems to be a mix of common CL resistors and a tiny linear regulator (I don't see any inductors).
I would still call this a constant voltage strip (with a little bit of regulation).

I don't see any voltage window limits in the specs where the regulation would still work or fry the linear chip.
This strip should be PWM-ed low-side with a locic level mosfet.


There are many low voltage CC LED drivers available for the Arduino,
but they need to be matched to the LED you're driving.
Give us a link to the 'other' LEDs you want to control.
Leo..

The problem is that in this case the voltage (or if we go CV then the current) would be quite high because we need to dim a lot of lumens at once.

For example, if we used 12V CV strips like this then we would have to go up to 8A to get the amount of light we need. What kind of logic level MOSFET can deal with that?

[Grumpy_Mike]

What kind of logic level MOSFET can deal with that?

Most types.
https://www.adafruit.com/product/355

[benyig]

Most types.
https://www.adafruit.com/product/355

Thank you! It's nice to see a sentence as clear as this one: "you can switch at least 15 Amperes without a heat sink - more if you use PWM" for a beginner like me.

As I don't know enough to do the math even with the help of the calculator, could you tell me that with this MOSFET what temp should I expect if it's constantly fully on and if it's constantly at about 50% approximately? At 12V 8A. Thanks!

[Grumpy_Mike]

Well the web page says:-

that means that you can switch at least 15 Amperes without a heat sink

The thermal resistance junction to ambient is 62 ^oC per watt. That means for each watt you dissipate you get a 62 degree rise in the case temperature.

So 8A at 50% is just the same as 4A continuously, the voltage is irrelevant. Assume you have applied 10V to the gate the maximum on resistance will be a maximum of 8.2mR so that means:-
Power = I²R
Power = 4*4 *  8.2 e-03
Power = 0.1312 W

So for each watt you get a 62 temperature rise therefore for 0.1312W you get :- 62 * 0.1312 = 8.13 ^oC rise above ambient.

For only applying 5V, the 8.2 becomes 16 e-03 so this is about 16^oC above ambient.

[benyig]

Thank you! So, following your calculation, what really matters are the R_DS(ON) and R_JA values. That means that with this guy we would get to cca. 278 ^oC above ambient at 8A and V_GS 5V. I hope I didn't make a mistake...

That seems too much, but now at least I know what to look for when choosing a MOSFET. Thank you again, Mike, for clearing things up!

[Grumpy_Mike]

Well I only get 182^oC above ambient.

8 * 8 * 0.046 = 2.9W so 2.9 * 62 = 182^oC

But yes in general you should not run the junction much above 120^oC for a decent life. This ofcourse assumes no heat sink.

at least I know what to look for when choosing a MOSFET

Excellent.

[benyig]

Well I only get 182^oC above ambient.

8 * 8 * 0.046 = 2.9W so 2.9 * 62 = 182^oC

But yes in general you should not run the junction much above 120^oC for a decent life. This ofcourse assumes no heat sink.
Excellent.

You're right, I just doubled the R for 5V instead of actually looking at the table, well... One more thing that I've learned (hopefully.)