I’ve noticed a bizarre behavior when my AC begins a cycle. Initially the fan turns on and stays on, and then every 20-30 seconds there’s a massive draw of power, which I assume is the compressor, but it only lasts for a few seconds and it immediately drops back down to the consumption of the fan. After about 8 minutes of these big current draws, the compressor appears to stay on and consumes a steady 4k watts for the remainder of the cooling cycle. Does anyone know if this a typical behavior for a single stage AC unit? I’ve attached a screenshot of the initial spikes and the first 8 minutes from the start of the cycle until the compressor stays on. Thanks for any insight.
I’m not an AC expert, but I wouldn’t expect to see that many huge startup spikes… I have new (1 month old) single stage compressors and they look like this through their cycle in the power meter (BTW - the yellow line is my solar generation)
Here’s how Sense detects those waveforms in the device-level power meter.
I wouldn’t be surprised if you have a bad runstart capacitor, where the compressor keeps trying to start but misses for a number of times, then finally kicks in.
It looks to me like Summer 2019 is shaping up to be the time Sense acquired significant AC-fail awareness … well, the data at least.
I believe @kevin1 is right here but there will come a day when Sense will tell you before you notice such things. Yeah!
Will alerting be active by the end of Summer? This is another one for @RyanAtSense & the team to look at.
Meaning, @gangelm, if you repair your AC you should definitely pass data and repair experience on to Sense. Perhaps by the end of Summer you’ll get a notification along the lines of “Your AC looks fine for next Season!” or “Your AC looks like it might need repair. Would you like us to do some deeper analysis?”
Thanks @kevin1 and @ixu. The AC technician came out and verified the behavior. He said it was on its way to burning out another component, and based on the repair history, it has happened before! We recently bought this home and we used the same service as the previous owner. Luckily, we’re still under the service agreement so the cost was reasonable. He replaced the failing connector and added a super boost which sounds like an additional start up capacitor.
Final question for you two and @RyanAtSense. Should I delete the 3 devices that sense detected for my AC since the startup signature will most likely be different?
I would leave them alone for now if for no other reason than I suspect Sense might want to parse your data and do a comparison. Failure gold!
As for the future look, I’ll pass that to @kevin1 & @RyanAtSense.
Install a hard start capacitor and see if that helps
My quick thoughts based on my discoveries and this blog:
- Let @RyanAtSense know you have captured a failing AC unit, and keep the old device around for a few more days so you can easily refer to it for any follow up
- But don’t be afraid to discard it since Sense keeps around the data.
- Plus Sense won’t automatically clean it up unless the old detection is interfering with new detections I’m kind of experimenting with what happens to old device detection AC here:
Thanks, the second thread is relevant to me because I also have an Ecobee although my system setup is not as complicated as yours. I only have one single stage system. Sense has also identified devices AC, AC 2, and AC 3. The fan is detected independently when running by itself but it isn’t picked up when the compressor turns on with it.
Another thing you can do is rename that A/C to something else and then will be able to save the history longer
Good luck with your fixed AC unit. Keep us up to date on what you do and what happens with detections. I think I’m going to keep my less identified “AC” device until the end of Aug to see how things play out.
If you are interested in ways to combine Ecobee data with Sense data, LMK. I’ve written an R program to do all the mundging necessary. It’s a bit trickier than it might seem since both Sense and Ecobee are “unreliable” in the sense that they both present time periods with missing data, so the combining has to be NA-saavy, knowing how to deal with missing data with the least amount of pessimism.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.