Gas Furnace analysis


Being a retired HVAC automation tech I find the Sense/Kasa combination an interesting way to analyze and control our electrical devices. Case in point is our two stage 90+ gas furnace. Sense found the blower motor fairly easy. But the draft inducer or igniter not so much. So I installed a HS110 on the furnace to see what’s going on. In the following images there is a leading and following 10W reading that is the transformer/circuit board/thermostat load.

Following is single stage operation.

This is two stage operation.

I have a steam humidifier that cycles the fan. This view shows humidification with a single stage heat cycle. Furnace only. Sense has identified the humidifier as a separate device.

My initial realization was that it’s no wonder that Sense never identified the furnace beyond the fan motor considering the limitless variations of this system.

Where do I go with this;

  1. Next winter I will analyze the 1 stage vs 2 stage vs humidification to find the best comfort level at an economical operation.
  2. The furnace/thermostat have a dehumidification function. Going to analyze this to see what operation is the most economical operation vs comfort. Sense has also identified the A/C condensing unit as a separate device.


Great to have an HVAC pro’s analysis of their forced air furnace power pattern. Totally informative. Thanks !



Thanks for that analysis. It was very informative. I have a similar issue to yours, except even harder to detect. My furnace(s), 2 identical Bryant 90+ ones, are variable firing ones. They currently are set to vary between 35% and 100%. On colder days out here, they never shut off. Great for comfort but I suspect will never be detected by sense.



Nice write-up, and your annotated power meter display is especially informative!

I will say that Sense has a good chance at detecting the inducer motor and igniter given enough cycles. Keep in mind that from the Sense ML point of view, the entire house one complex machine with many subcomponents. A furnace with three main components, two with inductive signatures, should be in the realm of detectability. The controls, on the other hand, would never be detected without smartplug assitance.

1 Like


We have a similar furnace and AC in our house. The main blower motor was replaced and I realized that there are 4 windings in it that only two are tapped. I had a look at the schematic and it had a lot of the information for loads and timing on the trace.

There is a startup pattern and shutdown which are quite consistent. The run time is determined by the load on the house.

Perhaps as the unsupervised AI gets the timing and period of the trace pattern, we’ll see it on our device list. It did detect the hot surface igniter for awhile and now only shows the blower. The change in winding doesn’t show up because there is no inrush for the energized motor.

BTW - we are using an ecobee thermostat so the is a possibility that the Sense system could correlate the call for heat, cool, or fan at some point in the future on the device list :sunglasses:



This is how Sense sees the furnace without HS110

Even when I zoomed in I don’t get the detail shown below. Sense has had over two years to move beyond the fan identification. I assume that since I’ve linked it to a HS110 sense will no longer try to identify it independent of the hs110.
And here’s what it looks like with the HS110.

1 Like


That is very interesting. I don’t understand why the smart plug would have any effect at all on the draw of electricity. What do you think is causing this?
Honestly it doesn’t look like the same device. The upper waveform or the one without the plug very closely resembles a heating element. My floor heater as seen in the screenshot specifically.
Am I understanding that this was first detected by Sense natively and then put on the plug? Did you then tell sense that the device was now in the smart plug?



Not sure what your exact question is, but the pictures you shared are exactly what I would expect. I have my forced air furnace on an HS110 and have it set to turn the air handler on for at least 5 min per hour (fan only). Your bottom HS110 screenshot looks pretty like a stream of furnace cycles (right waveform) while your top Sense-detected screenshot looks like a bunch of air handler cycles (left waveform).

The only difference is that the periods between your cycles are different, but your screenshots came from two different times, when presumably the outside temperature and thermostat settings were different.

1 Like


I selected a random screenshot of heating cycles under sense with and without hs110 as an example of the difference in detail. That’s all.

1 Like


Yup, much harder for Sense to find the other 3-4 components of the furnace cycle…



@samwooly1 - It looks that different because Sense plays a line between accuracy and giving a user a detection. I think many are being surprised how much different reality can be.

Frustrating at times but at least they’re getting heaps of data to get some corrections going :).

1 Like


I think @kevin1 may have identified the difference above.
The two original waveforms do not appear like the same device e is running in the same way. @kevin1 system and posted waveforms show the difference in fan only and fan while in heat cycle. I believe that is what is seen and the reason they look so different, because they are.
The original waveforms may be the same fan but the operation of the fan while in heating mode, starts up, fluctuates and shuts down differently than when on full power to circulate air.