LED driver DC2339A, possibility to chassis ground LED heatsink

I am driving a 20A LED (~5V, Luminus) with common anode back plate facing the heatsink.
I have a 20A driver board DC2339A (Analog). It uses LT3744, the corresponding datasheet indicates the heatsink should be isolated.

Is there any problem, in terms of performance or safety, to provide DC2339A a floating 12V supply and use optocouplers for input signals, so that I can chassis ground the LED anode (and heatsink)?

Datasheet LT3744 indicates an inverted configuration should be used, but this is less efficient and the corresponding board DC1930A only delivers 12A.

Thanks a lot.


I think the '3744 is probably one of the more exotic LED driver ICs I’ve seen lately. Kinda takes a bit to wrap one’s head around what’s going on there…

It appears to me that the '3744 datasheet implicitly associates the term “chassis ground” with whatever node the SNGD terminal is connected to, and doesn’t seem to present a side-by-side schematic view of what’s meant when speaking of “inverting” and “non-inverting” topologies. The excerpt below shows what a “non-inverting” implementation might look like. SGND and Vee are connected. If the anode of the LED load is connected to “chassis ground” (e.g. SNGD terminal) with a connection shown by the added red line, the converter output current would bypass the load and current sense resistor, which for reasons that should be apparent wouldn’t work all that well.

The “inverting” topology is exemplified by example below. Here, attaching the load anode to SGND allows the Vee node to be drawn to a negative potential relative to SGND through the action of the inductor; because Vee and SGND are not bolted together in this case, such a connection does not bypass the load and sense resistor.

In both cases, it’s implied that both the input voltage and control signals to the device are applied relative to SGND, though I haven’t noticed this to be explicitly stated anywhere in the text. For purposes of your question then, I’d ask what precisely you mean when you speak of “chassis ground”. I’d assume you’ve got an isolated AC-DC converter elsewhere to power things, and are inclined to refer to use that term in the routine sense of what gets connected to the AC safety ground.

To a first approximation, one may pick any node one’s heart desires on the isolated side to reference to AC line protective earth/chassis ground. If all the low-voltage stuff interfacing to the '3744 is referenced to SGND, I do not believe optocouplers would even be necessary to make things functional in a non-inverting configuration.

HOWEVER, one should carefully consider what the implications of doing that would be from an ESD/EMC standpoint. If one bolts the '3744 output node to the AC protective earth, it means that the isolated/low-voltage section’s “ground” is going to be flapping in the breeze with respect to earth ground at whatever the switching waveform looks like. If what you’re cooking has to pass emissions testing at some point, that’s not likely to be something that would work in your favor…

Thank you for clarifying the down sides of my grounding suggestion. As “chassis ground”, I did mean the AC supply ground from an isolated supply.

I saw my suggestion was driving the LED adequately but I just was not clear why this was not a good idea. I understand a non-grounded heatsink might also generate noise, but I imagine the emissions from the driver would be worse. I will go with an isolated heatsink and AC supply ground the SGND of the “non-inverting” configuration that is in the DC2339A board.

Thanks again.