TRADFRI transformer 10W

[Mariusthvdb]

https://www.ikea.com/nl/nl/p/tradfri-driver-voor-draadloze-besturing-grijs-50356187/

[bramstroker]

This one can not be supported for LUT mode, because we can not know what load is attached. Could be some ledstrip drawing 5 watt of power. Or a strip drawing 8 watt of power on full brightness.
You can use the linear mode and setup the min and max value yourself.

[Mariusthvdb]

yes, I see.
it would use the actual brightness to calculate whatever load is used depending on the min/max set in the config... magic

[bramstroker]

yes exactly. it will just interpolate between the min and max depending on the selected brightness.

[nepozs]

It has 0,18W of standby power.
I thought about an additional database for such devices with the remaining parameters to be completed by the user.

[Mariusthvdb]

right, so I have this now:

  - entity_id: light.boekenkast_marte
    platform: powercalc
    create_utility_meters: false
    linear:
      calibrate:
        - 1 -> 0.3
        - 3 -> 0.7
        - 13 -> 0.8
        - 26 -> 0.9
        - 38 -> 1.0
        - 51 -> 1.1
        - 64 -> 1.3
        - 76 -> 1.4
        - 89 -> 1.6
        - 102 -> 1.8
        - 115 -> 2.3
        - 128 -> 2.5
        - 140 -> 2.9
        - 153 -> 3.3
        - 166 -> 3.8
        - 178 -> 4.1
        - 191 -> 5.1
        - 204 -> 5.9
        - 217 -> 6.4
        - 230 -> 8.1
        - 242 -> 9.4
        - 255 -> 10.8

but how to add the light == 'off' state, being the standby value of 0.18 W? can we set a standby_power too when using the linear mode? Not according to the documentation. I feel this would be a good addition to the integration options, since the linear mode isnt truly applicable for these lights as you can see in the values.

FR: @bramstroker please add standby_power to linear mode and calibration settings.
I did try to follow https://github.com/bramstroker/homeassistant-powercalc#linear-mode-with-additional-standby-power but got an error on yaml mapping values...

changing to below (moving the standby_power up to above the linear: made the config check ok)

  - entity_id: light.boekenkast_marte
    platform: powercalc
    create_utility_meters: false
    standby_power: 0.18
    linear:
      calibrate:
        - 1 -> 0.3
        - 3 -> 0.7
        - 13 -> 0.8
        - 26 -> 0.9
        - 38 -> 1.0
        - 51 -> 1.1
        - 64 -> 1.3
        - 76 -> 1.4
        - 89 -> 1.6
        - 102 -> 1.8
        - 115 -> 2.3
        - 128 -> 2.5
        - 140 -> 2.9
        - 153 -> 3.3
        - 166 -> 3.8
        - 178 -> 4.1
        - 191 -> 5.1
        - 204 -> 5.9
        - 217 -> 6.4
        - 230 -> 8.1
        - 242 -> 9.4
        - 255 -> 10.8

[bramstroker]

It should just work for linear mode as well. You need to define standby_power on the highest level as this applies to all modes.
Did you try that already?

- entity_id: light.boekenkast_marte
  platform: powercalc
  standby_power: 0.18
  create_utility_meters: false
  linear:
    calibrate:
      - 1 -> 0.3

[Mariusthvdb]

duh, we crossposted..... and yes, I can confirm this to work correctly!

magic.

would you believe to be the best way (using linear, with calibrate and standby)? I mean, its not really linear at all...

[bramstroker]

Yes this would be close enough to reality as you defined a lot of measure points.
powercalc will calculate between the calibration points.

For example:

- 128 -> 2.5
- 140 -> 2.9

When you have brightness set to 134 the power will be 2.7

When you need more accuracy you can create your own LUT files with the measure tool. You can configure this LUT file with custom_model_directory option.

[nepozs]

In my opinion LUT file isn't really needed in this case.

[nepozs]

This is a great example of how we can approximate a (pseudo) nonlinear characteristic using only a few calibration points:

in reality probably very short red line could be discarded (but I don't know how estimation in powercalc works without point with brightness 1).

    linear:
      calibrate:
        - 1 -> 0.3
        - 3 -> 0.7
        - 102 -> 1.8
        - 178 -> 4.1
        - 217 -> 6.4
        - 255 -> 10.8

values should be slightly corrected to better match measured characteristic, but I would like to show idea.
(point 102 should have smaller value and points 178 and 217 higher)
Of course it's a must to measure power draw of bulb/light as accurately as possible before optimizing data.

@Mariusthvdb Could you move this thread to "ideas"?

[Mariusthvdb]

not sure, but Bram can I suppose.

[nepozs]

As author of thread you should have this possibility (under gear wheel icon next to "Category"):

[bramstroker]

moved to ideas