Circuit safety measures for DIY 12V LED modules

[criegels]

I'm in the middle of a project building a few high power LED modules utilizing cheap 3W leds from ebay (the type with the metal star base).

Each module will be stuck onto a drilled aluminum sheet with a thermal epoxy and will have a 120mm fan cooling from the backside.



Since i'm a cheapass I'm employing an ATX power supply, so at 12v with each string of leds running at ~700ma; each module (of six strings) will dissipate roughly 50watts max.

The control circuitry is a simple constant current source being activated by PWM. I'm aware that the Vbe sense transistor will somewhat help reduce thermal runaway, but I thought it would be sensible to add a PTC thermistor above the collector of the base-current feeding transistor (see circuit diagram) in case a fan eventually fails. This guy would also be epoxied above one of the LEDS and buried/insulated under some silicone.



Any hints about further safety measures would be appreciated, somewhat of an amateur here. (I'm already committed to the 12V PSU, so please dont criticize).

Cheers!

[Siwastaja]

You should rethink the thermal design. Surface area to the fan-driven air is minimal really; 50W sounds a bit high for this design. A standard finned heatsink would be better, it would work with a lower fan rpm or possibly even fanless, and you would save the drilling work.

You can make your current design better if you use standoffs on the fan mounting so that some of the air blows sideways. This will also increase the air flow from a simple fan by reducing the pressure requirement, so you may actually get almost the same amount of air through the holes as before, with extra air blowing sideways.

[criegels]

Thanks for the reply Siwastaja,

The design is for a greenhouse and requires some airflow, so the fans serve a dual function as air circulators and coolers.
As you suggested, I intend to have the fans on standoffs as the control circuitry will be mounted under the fans on the opposite side of the sheet from the LEDs.

A la:


Some silicone caulk along the edges of the fan in the gaps between the standoffs will restrict airflow upwards from under the LED hood, aiding natural convection.

Specifically though I'm concerned about the circuits failure modes over time. The thermistor approach limits the total power to about 25w, which should be fine for convection sinking in the case of a fan failure. The 120mm area there under the fan is fairly large, and plenty of 20W (100w equiv) LED bulbs have minimal heatsinking.

Any other failure modes I'm not addressing?

[mikerj]

You've actually decreased the surface area and lowered the cooling performance by drilling all the large holes.  A plain sheet with air blown against the surface would be better, though still probably not sufficient for 50W.

[criegels]

You've actually decreased the surface area and lowered the cooling performance by drilling all the large holes.  A plain sheet with air blown against the surface would be better, though still probably not sufficient for 50W.

If you'll allow me to disagree; here is my argument for why it should be fine at 50w.

I've used .063in (1.6mm) aluminum sheet, the leds have a metal star base that is equally thick. Those perforations will maximize the airflow onto ALL sides of each led. So this should he an improvement in terms of surface area in contact with air at room temp. Then add in the fact that there is forced as well as convective air flow on this surface and... well i dont want to get into the math, but i do believe this is far superior to plain sheet.

Some simple 2d surface areas:
Hole radius: 5mm
Number of holes: 54

Area removed by holes (single side): 54×pi×5mm^2 = 4241mm^2
Area added by hole walls: 54×2pi×5mm×1.6mm= 2714mm^2

[Siwastaja]

Holes indeed allow air to the opposite side, but it's unsure at best whether the air moves laterally there; most probably, some turbulence-like effect provides cooling on the LED side near the holes, but it doesn't go very far, unless you install a protective glass which would divert the air sideways.

So, having holes could, in theory, double the surface area because the air can flow on the both sides, but the area loss from the holes needs to be subtracted from this number. The LED star heatsinks seem to bridge the holes and, in effect, make the holes smaller than they first appear, which is good. Maybe the holes could be slightly smaller to begin with, this would at least increase the contact area between the LED stars and the sheet.

In the end, the thermal design looks rather poor for 50W, and it's hard to say whether the holes have positive or negative effect, but probably they aren't going to either make or ruin the day.

[criegels]

Say, for example, one LED burns out (open circuit) in the control string. If that occurs, the other strings start pumping close to 2A of current... How can i fix this failure mode?

[criegels]

Although it removes the feedback loop that was present in the original circuit, if I intend to run the PWM signal at <200Hz - would the following circuit be stable and reasonable?

[CJay]

You've actually decreased the surface area and lowered the cooling performance by drilling all the large holes.  A plain sheet with air blown against the surface would be better, though still probably not sufficient for 50W.

Have to agree, I think that dissipating 50W through that thin sheet with all those extra holes is going to be difficult at best.

The OP needs to take a look at the LED streetlight and floodlight designs that are out there on the market, the cheaper ones have marginal cooling at best and they have an awful lot more metal to dissipate the heat than his design.

[criegels]

You've actually decreased the surface area and lowered the cooling performance by drilling all the large holes.  A plain sheet with air blown against the surface would be better, though still probably not sufficient for 50W.

Have to agree, I think that dissipating 50W through that thin sheet with all those extra holes is going to be difficult at best.

The OP needs to take a look at the LED streetlight and floodlight designs that are out there on the market, the cheaper ones have marginal cooling at best and they have an awful lot more metal to dissipate the heat than his design.

CREE also has a 10-year warranty on their products, which are passively cooled and for all I can tell are not PWMed.

[criegels]

I'm convinced that this circuit would be superior.

-Base stopper resistors prevent complete current utilization by a open circuit string.
-The set voltage is tunable with a trimpot, and there is at least some temperature reactivity in the set voltage.
-The current in the control voltage circuit is more isolated from the rest of the circuit with a darlington transistor.

Oh yeah, I'm planning to buy as bunch of these little things to stick on the backside within a 25mm standoff area below the 120mm fan.
http://www.ebay.com/itm/261948596432

[criegels]

Works good & doesn't get very hot