Heat from Transformer and POE (Power over ethernet) Switches

I have an 18 x 18 x 6 stanless steel box to house two power supplies, and and a small 5 port switch. I am receiving my bandwidth from another wireless provider via a canopy SM. I am going to use a AP to transmit to a group of people with in a cluster about a mile away. My problem right now with the transformer for the S/M is the amount of heat it is giving off. After running for 3 hours in the closed box, it reaches 142 degrees on an 80 degree day. I am working on reconfiguring the box to see if it helps with heat dissipation, but has anyone else experienced this? Also could I elimate the power supplies and buy an inexpensive POE switch. Does the canopy equipment use the same standard?

No, the IEEE PoE standard is 48V, while Canopy uses 24V. I believe Canopy may be able to accept up to 40V, but I haven’t confirmed this with Tech Support; nor am I willing to risk a working unit to satisfy my curiosity. Even if you could use a PoE switch, you’ll likely have the same problem with excess heat.

The Canopy US power supply does generate a lot of waste heat. You could try the International power supply instead; I believe it’s a switching power supply, rather than a transformer, but they’re hard to find in the US.

The best thing you can do to keep the enclosure temp down is to shield it from the sun. Do not, however, insulate the enclosure unless you also provide some type of cooling – even fans on the bottom. There’s a free software tool available at www.hoffmanonline.com for calculating enclosure temps and coolong requirements that you might find helpful, or at least educational.

Is the enclosure necessary? Can you install longer Cat5 cables to put the power supplies and switch inside a building?

Have we discussed this project before?

Actually this box is in a building out of the sun, not insulated. And yes we have discussed this before.

I have it working reliable now. Here is what I did.

I completely revamped the layout of equipmet in the box.

First off I removed the 5/8" plywood backer in the box that I was using so I could easily mount things like the switch. I did this for 3 reasons. The first one was to remove the insulation value The second was that if something bad would happen with the excess heat I wasn’t supplying more fuel. Thirdly it moved the Transformer another 5/8" away from the front cover. (The transformer was only about 1.125 inches away from the front cover.)

Secondly I removed the higher quality (tripp-lite) surge protector, and replaced it with two cheap surge protectors that where only about 1.125" think. This gave me even more space in the front of the transformer.

Thirdly I placed the surge protectors and the transformers lower in the box so the heat could rise out of the way.

Next I painted the inside of the cover black hoping to absorb some of the heat and conduct it further away from the transformer.

Finally mostly as a last safety precaution I mounted the box using four 4" conduit spacers to keep the metal box further away from the wood studs, but it problably provides a little better circulation

The results say it all. I bought a small IR Thermometer for 49 bucks at Radio Shack. With this I was getting to about 142 degrees F before all these changes and the SM seemed to cut out after running only for 2.5 hours. Now I am running at 124 degrees F and for 4 days steady. (These temperatures are the temperatures of the actuall side of the transformers, not the box temperature.)

I hope this can help the next person, because I never thought all those things would drop the temperature 18 degrees.

I do have one more additional question. If you have just an ap at the main site and a sm at the remote. Do you need to do anything to get timing to the ap, or is that just if you are creating a big conglomeration of a network?



GPS timing is only needed to synchronize physically separated units so they will transmit at the same time. A single cluster of APs doesn’t necessarily need a GPS signal; it just needs a single, accurate, 1-pulse-per-second signal for all the co-located units.

I’m glad you improved your enclosure problem. Thanks for the update.

Did you consider mounting the transformer closer to the enclosure wall, even getting a tube of silicone heat-sink paste from Radio Shack and pressing the thing against the wall? This may increase the transformer temp a bit, but should lower the enclosure temp. All the heat has to exit via the walls; you may as well put the heat there directly.

Also, I’m not sure the temps are high enough for the black paint inside to help. Surface color and texture affects radiant heat, but not convective heat. If radiant heat is an issue, it’ll likely help just as much to paint the outside of the enclosure. There’s also a big difference between flat black and gloss black because of the surface texture.

The best things you did were adding the rear surface for cooling and moving the heat sources to the bottom. However, if you moved the Ethernet switch higher as a result, its temp could actually increase. Hopefully, the added surface area on the back more than compensated.