During one of our recent disasters, the county gave out free electric heaters as it was winter and our gas service was out for a couple of weeks while they repaired the lines. These heaters were understandably cheap stamped-metal affairs, but they served their purpose.
Fast forward a couple years and I’m trying to come up with a backup plan to keep my bird room warm if (when) the elderly furnace goes out. (I had already replaced the computer in that furnace but that’s a story for another time.)
The aforementioned Free Heater suggested itself, but when I auditioned it on a cold night the mechanical bimetallic thermostat did a pretty terrible job of regulating the heat. I like to keep the bird room at 70F, but even after adjusting the touchy control knob to turn “off” at exactly 70, there was so much hysteresis the room might end up at 65F or 75F.
Then I remembered that I had once gotten a couple of generic industrial PID temperature controllers in a trade. They even had built-in relays! Surely it would be easy to hack one of these into the cheap heater for precision control.
Narrator: it was not easy
I won’t bore you with every detail of this build, but here are the highlights:
- The PID controller ran on AC (good) and had an internal relay (good), but the internal relay was only rated to 1A (bad, as the heater coil was 1000W so needed around 9A).
- Since the internal relay wasn’t beefy enough, I decided to slave it to an external 10A relay. I found some on Amazon with the right specs, but when I installed one, whenever it was energized it made a loud buzzing sound. I checked everything and couldn’t find anything wrong. Luckily I had bought a pair, and when I tried the second relay it worked properly. Success! Except… a critical part having a 50% failure rate doesn’t instill confidence in something hot that will be left unattended at night.
- As noted, the original bimetallic thermostat sucked, but it did have an integrated gravity switch that would turn off the heater if it tipped over. Unfortunately it was tightly integrated into the original assembly so there was no way to include it in the new build. Another safety feature gone.
- The wiring turned into quite a rats nest, and since it was all high-voltage high-current AC, the connections HAD to be reliable. I ended up using crimp rings and a large bus terminal to make all the connections, which took up yet more room but was worth it.
- After all the trouble of installing it, it turned out that my PID controllers had a maximum cycle time of 100 seconds (I was hoping for more like five or ten minutes). Which implies that the relays will also click on and off every 100 seconds.
Relays are mechanical devices that wear out, and cycling them 864 times a day was troubling. I could (and may) replace the 10A relay with a solid-state relay, but I’d have to either find one with a 120V “coil”, or find a way to add a safe low-voltage supply to the already-full box. SSRs are also prone to failing “on”, which is yet another thing to worry about or design around (a surprisingly hard problem).
So what happened?
After all that, it’s worth noting that the hacked heater did work quite nicely, impressively holding the heat at a constant 70F or wherever you set it. (And this was without me fooling with the PID defaults; good job China!)
It also seemed reliable enough, but knowing how the sausage was made I don’t know if I’d be 100% confident leaving it running unattended overnight, which was the whole point of this endeavor. I could keep working on it to make it safer and more reliable, but at this point it might just be worth investing in a better heater.
Some projects proceed smoothly towards an elegant design; others, like this one, bounce like bumper cars from one problem to another, with the “solution” to each problem adding complexity and more things that can fail. I’ve run into this with spacecraft systems; if a backup system is likely to cause more problems than it solves, it’s usually a sign that you should rethink your original premise.