Hi all,

I couldn't find an introduction forum so I'll shortly do it here, my name is Niels and I'm a mechanical engineer working in the automotive industry, I'm mainly working in a theoretical environment developing optimization and control strategies for hybrid and electric vehicles. In my spare time, I like to tinker and work on my Lotus 7 replica, with electronics I can often find my way around making use of the internet resources but I'm not an electrical engineer by a long shot.

I'm currently however building a project where I need to drive 6 SYNIOS LEDs. Something which looked pretty straight forward at first but turns out to be not, maybe someone over here can help me out with this. I'll try to provide as much information as possible.

First of all, the project consists of multiple boards all driving 6 LEDs, they are all SYNIOS P2720 but are different in luminous intensity and electrical spec. The forward current ranges among them from 200mA until 700mA. The goal was to have a generic architecture in which I would be able to use the same STCS2 driver for the different boards and change the current sensing resistors according to the used LEDs. Just to be clear, each board would consist of 1 LED type only, for clarity I'll explain my problem with a specific LED which at this point causes issues.

For this particular board, I'm trying to run 6 "KS DMLS31.23-6J8K-68-M3W3" LEDs which are rated at 700mA and 2.43V, the board design comes basically directly from the STCS2 datasheet for which I've adapted the current sensor resistors accordingly. According to the datasheet, this is calculated by taking the ratio of the STCS2s feedback voltage over the LEDs forward current \[ R_f =\frac{V_{FB}}{I_{LED}} = \frac{100[mV]}{700[mA]} = 143 m\Omega \]. Note, that I'd like to drive 6 LEDs in total, therefore I've created two parallel branches each taking 3 LEDs in series, which in my mind would result in 1400mA (at least that is what I've learned way-back in school), this would mean \[72m\Omega \] instead of \[143m\Omega \]. However, I tried this and results in nice and smokey toasted LEDs (Trial and error ftw?).

So, I've replaced my resistors with the \[143m\Omega \] which actually works perfectly fine for a limited amount of time. After this time the LEDs shut down and will go on after a few seconds again, this is repeating endlessly. I'm pretty sure to have reached the limits of my STCS2 in terms of its dissipation capabilities. Does someone have a "simple" but robust advise on how to tackle this whenever I already have a PCB design, can I increase my thermal conductivity enough via an external path? Or, do I have to redesign my board, adapt the electronics and create an onboard heat spreader?

Thank you
Niels

 

These LEDs will die at 120C and will have a short life near 120C. You need to get 2 watts out of the part and keep the temperature down. I did automotive design for a while. Cars get real hot and cold and you need to remember that just because something works at room temperature does not mean it will work out doors. With out an aluminum heatsink I don't think you can run at full power at automotive temperatures.

 

These LEDs will die at 120C and will have a short life near 120C. You need to get 2 watts out of the part and keep the temperature down. I did automotive design for a while. Cars get real hot and cold and you need to remember that just because something works at room temperature does not mean it will work out doors. With out an aluminum heatsink I don't think you can run at full power at automotive temperatures.

That’s a good hint although at the moment the LEDs are running at approximately 1.65Watts (estimated), the thermal shutdown seems to occur at the STCS2 due to poor thermal conductivity of my board.

 

I have made LED boards where 95% of the PCB is copper. (both sides) I do anything to pull the heat away from the LED.
Some boards are mounted on a sheet of aluminum.
I have not used it but some PCBs have a sheet of aluminum inside.

 

Thanks for your answer, I've been looking a bit further into the thermal "management" of these kind of LEDs. I'm however wondering if in that case also a direct thermal path or sinkpad suffice would suffice. Or if there is still a need for an external heatsink?

 

In my opinion, you may calculate the temperature of the sink if you can define the limits of a box and the convection and radiation constants. You may consider some ventilation also.

 

See also:
How to calculate the temperature rise in a sealed enclosure | Heat Sink Calculator-Blog: Focused on Heat Sink Analysis, Design and Optimization

 

I'm at the moment also looking into switching towards a boost-buck converter instead of the linear approach, this should increase the efficiency and reduce heat to be dissipated. The reference you give will for sure help in figuring out how to get the heat away from my LEDs, thanks!