Shiny LED Panels Run Hot

LED panels are great for grow lamps and all sorts of other purposes, but they are liable to overheat if the back surface is bare aluminium. Simply painting the top of the panel could almost double its thermal capacity, allowing you to run it at much higher current.

Source: Brompton Technology

My interest in hydroponics was piqued when a friend was given a (no doubt rather expensive) gift of an indoor vegetable garden.

Now in my experience farmers tend to be rather frugal, practical types. When I was growing up in Oxfordshire, our next door neighbour used to joke that he would wait until he saw two pigeons on a branch before getting out his shotgun because a single pigeon was a waste of a cartridge.

I didn't want to waste my time on money on the wrong set up – I wanted to get the right hydroponic system with my first shot! So I set about finding out what I could about the equipment on offer. Specifically, grow lamps. And there were all sorts of new gear to look at.

Quantum Boards. Do They check out?

After researching hydroponic equipment for a little while, my skepticism antennae started to twitch. Grow lamps commonly use a kind of LED panel known commercially as a "Quantum Board", which looks to be an aluminium PCB studded with parallel strings of medium power LEDs. (See the pictures above.) There are variations on the theme, but generally the driver electronics are implemented in a separate module, which is sometimes mounted on the top of the panel.

Perhaps because the alternative is an HID lamp, these LED panels are usually marketed as running "cool". You can generally also read lots of user reviews, which may or may not be real, attesting to this. The idea is that PCB itself is the heatsink: with appropriately wide spacing of the LEDs, no supplementary cooling is necessary.

Well, I remember when Apple's MacBook Air was the cool new thing, and then I burned my lap.

The issue is whether or not the spacing is "appropriately wide". I contend that every panel that you will ever see is thermally limited to some degree. You have to run these things below their rated current to stop the LEDs burning out. The other option is to the cool them down somehow. That would explain why the same sites that sell Quantum Boards also advertise very large, expensive heatsinks.

It's not hard to imagine why LED panels might be undersized. Convenient though it may be, flat aluminium PCB substrate is an expensive and inefficient kind of heatsink.

Yet another online heatsink calculator

Wanting a concrete example, I chose the QB120 V2 boards more or less at random from the Horticulture Lighting Group website. It is advertised that "Each board can be powered up to 70 watts. No Thermal Interface Material or heatsink is required." Let's evaluate those claims.

The panels measure 240 x 280 mm and feature 120 highly efficient Samsung LM301H LEDs capable of producing 40 lumens or so at a gentle 65 mA. At that current each LED has a forward voltage of about 2.7 V and throws off about 140 mW of heat. The LM301Hs can be run much hotter, up to 200 mA.

I set about making a widget to find out what kind of temperature the panel would reach when running at these currents. The spreadsheet takes a few short cuts for the sake of simplicity. It calculates the temperature of the panel rather than the LEDs on it, for example, and assumes that it is the same temperature all over, when in fact aluminium sheet is not a perfect conductor of heat – as you can clearly see on picture at the top of this post.

My first attempt at the calculator also completely ignored the effects of radiation,* until I discovered that it can account for a substantial proportion of the heat dissipated by surfaces that are cooled by natural convection. (If the heat sink has a fan on it, not so much.)

Plugging in the numbers, we find that at 65 mA per LED the panel would generate 17 W of heat and rise to about 35°C above the ambient temperature without additional ventilation, a tolerable burden of heat. At 200 mA... let's just say, the life of the LEDs would be significantly curtailed. While in a narrow sense it is true, as the website states, that "each board can be powered up to 70 Watts", doing so for more than a short time would risk ejecting the magic smoke (fabled eponym of this blog).

The one big unknown is what the back (or rather, top) surface of this particular panel looks like. As far as I know single-sided aluminium PCBs almost always have a shiny, brushed finished on the reverse. With an emissivity coefficient of ε = 0.09, the bare metal would dissipate only about 10% of the heat by radiation. If the top surface was anodized or painted, however, it could radiate as much as half of the heat the LEDs generate, effectively doubling the thermal capacity of the panel.

I've punched in the specs for a few other LED panels, and the story is the same each time. Granted, the calculator isn't perfect, and I would certainly like to back up the results with some experimental evidence. If the calculator is accurate, however, the implications are clear. (You see that shiny hood on your brand new Mars Hydro TS1000? Paint it.)

Please play around with the spreadsheet. If your favourite lamp cuts the mustard, be sure to let me know in the comments!

Paint your heatsink

Manufacturing an LED panel with an unfinished aluminium surface on the reverse is an obvious design flaw, a compromise made for the sake of lowering the cost. The good news is that there is an easy fix. You can paint it yourself. (I wouldn't try anodizing an expensive PCB unless you are seriously brave, or ridiculously rich).

From what I can gather the colour doesn't matter, only that the finish is dull. Personally I would go for matte black. Aerosol paint is best because you can layer thin coats, and because if necessary it can be wiped off with a rag and a little acetone. When painting aluminium it is recommended to degrease the surface before sanding with a fine grit and applying an etching primer. When the primer is dry, finish with the top coat.

Footnote

* So... I would have done it sooner but radiated heat is proportion to the fourth power of absolute temperature, so I had to figure out how to solve a quartic equation and then implement a numerically stable algorithm in Google Sheets.