cum grano salis — nobody likes a clever bastard

I handle my own network infrastructure and most of my PC and networking equipment is in a repurposed Ciprico RAID cabinet. It has a 240V fan tray on the top and I don’t really want it running all of the time because what’s in there right now doesn’t put out as much heat as a half ton of circa 1997 9GB disks would. I had a lot of options for purchasing something to do this, but being armed with the knowledge that most electronics equipment contains only a few bucks of parts and that the design of this would be a walk in the park, I just built my own.

Note: If attempting to build this, thinking of building this, or succeeding in building this kills or hurts you or others, destroys anything, or otherwise causes you to think I am to blame for something, I am not. It is your fault and you accept that I am not liable for anything.

I originally wanted to run the unit off of the switched line power, but to keep it simple, I ended up using an old cell phone charger as the power source. That goes into an affordably priced and very versatile Maxim MAX604CPA voltage regulator for 3.3V to run the show.

Another great part, the Maxim MAX6509HAUK-T, has all of the brains. It’s a temperature switch with programmable hysteresis and threshold. I set it to the low 2° hysteresis by connecting the pin to ground. There is no hope that a 10° hysteresis would be useful since the fan tray would never be able to pull in that much ambient cool air. I don’t have supplemental cooling and I just need it to maintain a reasonable temperature in the rack. If it gets too hot, my UPS will start complaining and I can handle the rest manually. The temperature threshold is set with R2. There are two simple formulas for calculating R2’s value in kΩ. T is in °K for both equations.

For temperatures between -40°C and 0°C:

For temperatures between 0°C and +125°C:

When the temperature switch is tripped, it pulls the transistor to ground. That transistor turns on the LED and triggers the MOC3043 optocoupler, turning the triac on. The dangerous side has basic transient and inductive kick/overvoltage protection with a metal oxide varistor in case it is used for inductive loads. Sizing the heat sink for the triac wasn’t too difficult, but finding the heat sink took longer. There are just too many available. I settled on the MK3306/S insulating kit and the 507302B00000G heat sink.

Most of my time was spent on the PCB. I didn’t have much to work with over here, so I decided on a single sided 2″ square piece of FR-4. It all fit pretty nicely with all of the SMT parts on one side. I have this stupid trace that goes for a little walk around the opposite side of the VRM to the transistor, but it works and that’s what matters.

I managed to stick this into a PBL3 enclosure with two IEC outlets. I cut some slits into it to let the air and low voltage power in. I don’t have pictures of it because it’s in the rack, but it looks like a black box… with wires coming out of it.

Bill of Materials

I didn’t include the little passives or the transistor in the BOM. They’re only maybe a few cents each and mostly interchangeable. If you’re interested in the exact parts used, feel free to contact me.

So, the total came out to somewhere around $19.14, assuming the passives and other parts were around $5. Who knows what this would cost retail, but it’s probably a lot more and would be a lot less fun.

I did this project in May and it has been running for six months without a hitch. There was a noticeable drop in power consumption as soon as this went into place, so I met my goal. It hasn’t burned the house down, either!

I’ve packaged the gEDA files for anyone who is interested. They are located here.

240V Thermostat by cum grano salis — nobody likes a clever bastard, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.