Open window detection. (may or may not be a bug)

[ChrisMarriott38]

Not 100% sure if this is working right. As the lounge temps just dropped a load with the window open. However I don't know if this is related to the API update issue curently experiencing.

Does the Open window detection get the data from the API ie- TADO's interpretation? OR is Tado CE doing it's own calculation for a noticed more accurate quicker temperature drop.

If so, I do wonder what tado's interpretation is vs a potentially more accurate integration version might be.

For example Tado might report it after 15 minutes? But maybe an integration based one with an override might be good? Then you input a value for how long in minutes a window/door is detected as open. OR a Temperature value in C rapid drop to then use as a sensor to prevent heating.

[hiall-fyi]

@ChrisMarriott38 Good question! Let me clarify how Open Window Detection works.

Current Implementation

Tado CE uses Tado's API data directly - we don't do our own calculation.

When Tado's servers detect an open window (based on rapid temperature drop), they set openWindow in the zone state. Tado CE simply reads this and exposes it as a binary sensor. So the detection timing, sensitivity, and logic are all controlled by Tado's servers, not by Tado CE.

Why We Don't Calculate It Ourselves

1. Tado's detection is already good - They have years of data and ML models
2. Avoids duplicate logic - No point reinventing what Tado already does well
3. API efficiency - We'd need more frequent polling to detect rapid changes
4. Consistency - Matches what you see in the Tado app

Your Enhancement Suggestion

You suggested custom detection with configurable:

• Detection time (minutes)
• Temperature drop threshold (°C)

This is an interesting idea! It would require:

• Tracking temperature history per zone
• Calculating rate of change
• Comparing against user-defined threshold
• Creating a separate "Tado CE Open Window" sensor (alongside Tado's)

Pros:

• Faster detection than Tado's servers
• Customizable sensitivity per zone
• Works even if Tado's detection is disabled

Cons:

• More complexity
• Needs frequent temperature readings (more API calls or relies on existing polling)
• Could conflict with Tado's built-in detection

My Recommendation

For now, I'd suggest using Tado's built-in detection. If you find it's too slow or not sensitive enough:

1. Enable/Disable per zone - In Tado app: Settings > Rooms & Devices > [Zone] > Open Window Detection (on/off only, no sensitivity adjustment available)
2. Use a physical window sensor - More reliable than temperature-based detection
3. Create a HA automation - Use a template sensor to track temperature rate of change and trigger your own "open window" logic

Note: The issue you're seeing (window open but sensor not triggering) might be related to the polling interval issue in #99 - if polling is stuck at 120 min, the sensor won't update frequently enough to catch the open window event.

[ChrisMarriott38]

Yea so what's come to my mind and i've found as an issue is that with the 100 APIT limit, to actually get a good accurate open window detection, you need faster polling. My wife had the patio door open, and its about 50cm from the Lounge thermostat. It detected a fairly rapid temp drop, BUT i've not seen anything open window related show up.

Now I dont know if that was due to my APIcalls being 2 hourly, but I have noticed the sensor/entity going to unavailable lots.

So then it made me wonder IF people could select the Tado method. OR in-integration as another option.

However, it might be difficult still, Unless....HomeKit's temperature sensor entiy was selected or used for anything LOWER than the API polling rate. My thinking being if someone has faster polling, ie 20k rate at 1 minute updates, that is faster than HomeKit. But if your on 1h or 30 min polling, then the HK version updates i think from 5 minutes up to 15 depending on the amount of temperature change. I see variable sensor updates.

You can see here. The fairly rapid drop and below the stats intended low point, caused the heating to fire, which i manually intervened at.

You have some interesting points for sure. 17 minutes feels a bit too long to have the open window detection be of use. But i'll have to do some testing. So i do think it may be related to the API issue. But not sure on the Unavailable aspect. Will investigate more.

• Also open window from Tado only works with AutoAssist to be automatic heating off. If you don't have auto assist the sensor "should" show open window but it doesnt prevent heating.

• Yea i have wondered about adding further window/door sensors around the house.

[hiall-fyi]

@ChrisMarriott38 Thanks for the detailed analysis!

Current Limitation

You've identified the core issue - open window detection depends on how frequently we get temperature updates. With the API rate limit (100 calls/day), polling intervals can stretch to 30min-2hr, which is too slow to catch rapid temperature drops reliably.

Tado's server-side detection (openWindow in API) should still work independently, but we won't see the update until our next poll.

HomeKit Integration - Future Improvement

I'm exploring HomeKit integration as an alternative data source. This could significantly improve time-sensitive features like open window detection:

Data Source	Update Frequency	API Impact
Tado API	30min - 2hr (rate limited)	Uses quota
HomeKit	~5-15 min (local push)	No API calls

Benefits of HomeKit approach:

• Local communication (no cloud dependency)
• Push-based updates (no polling needed)
• Doesn't count against Tado API rate limit
• Faster temperature readings = better open window detection

What This Means for Open Window Detection

With HomeKit integration, we could:

1. Use HomeKit temperature for faster change detection
2. Implement in-integration open window detection (configurable threshold)
3. Keep Tado's openWindow as a secondary/confirmation source

This is still in the exploration phase, but your use case is exactly why we're looking into it.

Current Workaround

For now, if you need reliable open window detection:

1. Physical window/door sensors - Most reliable option
2. HA automation - Create a template sensor that monitors your HomeKit temperature entity for rapid drops

Example template sensor approach:

template:
  - trigger:
      - platform: state
        entity_id: sensor.lounge_homekit_temperature
    sensor:
      - name: "Lounge Rapid Temp Drop"
        state: >
          {% set current = states('sensor.lounge_homekit_temperature') | float %}
          {% set previous = trigger.from_state.state | float %}
          {% if (previous - current) > 2 %}
            on
          {% else %}
            off
          {% endif %}

---

I'll keep this issue open and link it to the HomeKit integration work. Your feedback on detection timing (17 min being too slow) is valuable for setting the right thresholds when we implement this.

[ChrisMarriott38]

Superb thoughts on the above! And yes even tado's open window detection i'm sure somewhere i read is 15minutes based? Not confirmed though.

I have wondered if where you added the outside weather sensor thats user selectable in settings. If that sort of option would work on a per zone basis. That you could user select your Home kit device? Not sure if that helps, but this is a good usecase for the open window feature and making it more reliable/workable with the API limitations.

[hiall-fyi]

Hi @ChrisMarriott38 Great suggestion about per-zone sensor selection! That's an interesting direction worth exploring.

I'll do some testing to see how HomeKit temperature sensors behave - update frequency, reliability, etc. Once I have a better understanding of the feasibility, I'll update you here.

Thanks for the suggestion - it's a good use case for improving open window detection reliability!

[ChrisMarriott38]

Very happy to help test these bits if its any help :)

It would be very interesting given the stats we have around the house to see if they trigger open window detection and us being able to use that sensor to toggle on and off the overall heating system.