Cooling LED drivers and LEDs on the same heatsink is bad?

[treez]

We are driving twelve 40V, 4 Amp UV LED arrays from twelve 160W LED driver PCBs. (one 160W LED array per driver) The LED driver PCBs are double sided SMT boards, (all components on the top layer) with thermal vias to bottom layer cooling copper lands.

Should both the LED Driver PCBs and the UV LED MCPCB’s all be thermally coupled to the *same* water cooled heatsink? After all, the UV LED MCPCB’s will run much hotter than the LED driver PCBs, and so thermally coupling the LED driver PCBs to the same heatsink as the UV LED MCPCB's may just end up with the LED drivers running hotter than they would if the LED driver PCBs were not thermally coupled to this heatsink?

So what do you think?… is it a waste of time thermally coupling the LED driver PCBs to the same water cooled heatsink as the one that the UV LED MCPCB's are mounted on?

The LED driver PCBs do actually get screwed to protruding “platforms” of the water cooled heatsink, but we wonder if its worth actually thermally coupling the LED driver PCBs to these platforms with electrically insulating, thermally conducting heat transfer pads?

 

[andre_luis]

It´s not clear for me if these LEDs operate on pulsed mode, what matters significantly on requirements for dissipation of a big amount of energy in a short period.

By the way, could you provide some drawing detailing layout of the heatsink and proper placement of LED arrays ?



+++

 

[treez]

LEDs are on all the time.

 

[chuckey]

Your LEDs dissipate 1.9 KW, you don't say what the drivers dissipate. Unless its too much for a reasonable heat sink, I would think keeping the water cooled heat sink as simple as possible would be beneficial. Perhaps you could mount the drivers on a bracket bolted on to the water cooled heat sink to reduce the size of the drivers heat sink.
Frank

 

[dick_freebird]

If you needed isolation, and the controller / driver is on the
primary side, sharing the heat sink might either ruin that or
add some more risk elements (insulator failure) to the long term
integrity.

There could also be some positive feedback type issues
with the load heating the driver (and drifting its setpoint,
etc.) I suppose.

 

[treez]

The 12 drivers dissipate about 2W each. They have a Vin of 48V, and the isolation smps is upstream of them and provides them with this 48V input voltage.

 

[mrinalmani]

After all, the UV LED MCPCB’s will run much hotter than the LED driver PCBs, and so thermally coupling the LED driver PCBs to the same heatsink as the UV LED MCPCB's may just end up with the LED drivers running hotter than they would if the LED driver PCBs were not thermally coupled to this heatsink?

I think the kernel of the question is related to over-heating of the driver board, if thermally coupled to the LED board, because LEDs as it is operate at elevated temperatures.

I do not completely agree to this. Here we are interested not in the case/junction temperature of the LED, but in the temperature of the coolant ie. water. Even if we take the worst case scenario of the coolant becoming boiling hot due to the LEDs, a carefully designed interface between the driver board and the coolant should not have a thermal resistance greater than say, 5W/C, per driver. This puts the driver case at 110°C. Which should be acceptable for almost any switch.

For a properly designed cooling system, ΔT of coolant will probably be limited within 25°C, and this should run the driver cooler than separate heat sinks.

 

[treez]

I don't think our cooling system cool the coolant that well.....I think the coolant gets up to 70c

 

[mrinalmani]

75°C is generally taken as a loose standard for heat-sink body temperature. This results from (50°C ambient + 25°C ΔT)
However, it is common to allow for a greater ΔT in non-convective heat sinks. (An 800VA Sukam inverter has a ΔT of around 45°C for the MOSFET heat-sink) This puts the body temperature of the sink at around 80°C on a "normal" day.

If the water temperature is limited below 75°C, in my opinion it would we as good an isolated heat-sink, if not better. Over-sizing the heatsink will result in temperatures below 70°C, but is it worth it?