feature/LED1624G9

[marthubner]

I made a measurement of IKEA LED1624G9 using a script running with two bulbs. I then manually added some values to the end of the hs.csv file approximately corresponding with xy_color: [0.458, 0.41] which is bulbs default warm white color. Power consumption is significanly higher for this color and near colors and the measuring script did not catch this spike at all.
Another problem is that this bulb does not work with color_temperature (https://github.com/Koenkk/zigbee2mqtt/issues/6388) but powercalc does not create any sensors until both hs.csv and color_temp.csv files are provided so I made an empty dummy color_temp.csv file.

[bramstroker]

Thanks.

I made a measurement of IKEA LED1624G9 using a script running with two bulbs. I then manually added some values to the end of the hs.csv file approximately corresponding with xy_color: [0.458, 0.41] which is bulbs default warm white color. Power consumption is significanly higher for this color and near colors and the measuring script did not catch this spike at all.

How do you know power consumption for the mentioned color is significantly higher than the measure script picked up?
When this is the case I'm also worried by the accuracy of the other measured values.
Could be a bug in the measure script somehow, but not sure where. It is working fine for me when using Hue bridge and Shelly plug.

Another problem is that this bulb does not work with color_temperature (https://github.com/Koenkk/zigbee2mqtt/issues/6388) but powercalc does not create any sensors until both hs.csv and color_temp.csv files are provided so I made an empty dummy color_temp.csv file.

Could you have a look into Developer Tools and retrieve the state attributes for the light? I'm interested in supported_color_modes field.

mpty dummy color_temp.csv file.

[marthubner]

How do you know power consumption for the mentioned color is significantly higher than the measure script picked up?
When this is the case I'm also worried by the accuracy of the other measured values. Could be a bug in the measure script somehow, but not sure where. It is working fine for me when using Hue bridge and Shelly plug.

I have that particular color set as a main warm white room light because the bulb performs best with that setting. It is also color with which the bulb turns on when reset. All other colors appear substantially dimmer. There are some discussions about this on github for example here.
It seems that this bulb is designed to provide much better results only at very narrow section of color spectrum and the script just is not precise enought and it missed that exact spot where some dedicated warm white diodes kick in at full brightness. I don't know if there are also some other spots like that but I think that if so, they will not be as significant as this one.

Another thing is that it is also not possible to hit that color with home assistants color wheel, it always slips somewhere next to it. You have to set specifically xy_color: [0.458, 0.41] to get the best appearing light and also higher energy consumption.

This is where the color lies on the wheel:

This is the graph of my shelly's power meter from running the measurement. The spike on the left is consumption with the subject color at full brightness and it just never got there during the measurement... (The gap in the middle is where the script stopped for some reason.)

Could you have a look into Developer Tools and retrieve the state attributes for the light? I'm interested in supported_color_modes field.

It states color_temp as supported but it is probably a mistake in zigbee2mqtt because the bulb does not do anything when trying to change color_temp and there are also multiple discussions about color_temp not working.
dresden-elektronik/deconz-rest-plugin#4140 (comment)
Koenkk/zigbee2mqtt#6388 (comment)
Koenkk/zigbee-herdsman-converters#3218 (comment)

[bramstroker]

You can ignore the failed action for "Release drafter", need to have a look into that.

color_temp.csv does not validate correctly. Right now it crashed the validation tool. I think you need to add a header row, and at least 2 other rows.

[bramstroker]

And thanks for clarifying.

[nepozs]

This is where the color lies on the wheel:

In my opinion this particular color loooks like warm white (maybe 2700K?) but bulb probably works in color-temp mode.

[bramstroker]

Yeah the color_temp mode seems to not work correctly for this light looking at the linked github issues.
When this is resolved we should update the library to include the correct color_temp.csv (with measurements) or remove it from the library.

[bramstroker]

Will merge for now. @marthubner Could you please have a close look into the related github issues when there are any fixes for color_temp mode?

[marthubner]

@bramstroker Yes, I will keep my eye on the issue with color_temp.

[bramstroker]

@marthubner Another powercalc user stumbled upon issue with this light profile not working correctly because color_temp was missing. See #2047
Back in the time there was an issue with color temp mode reading this PR comments.
Would you be able to provide measurements for this color mode to fix the profile?

[marthubner]

@marthubner Another powercalc user stumbled upon issue with this light profile not working correctly because color_temp was missing. See #2047 Back in the time there was an issue with color temp mode reading this PR comments. Would you be able to provide measurements for this color mode to fix the profile?

Ok, I'll try.

[bramstroker]

@marthubner Another powercalc user stumbled upon issue with this light profile not working correctly because color_temp was missing. See #2047 Back in the time there was an issue with color temp mode reading this PR comments. Would you be able to provide measurements for this color mode to fix the profile?

Ok, I'll try.

@marthubner are you still able to do this?

[marthubner]

@marthubner Another powercalc user stumbled upon issue with this light profile not working correctly because color_temp was missing. See #2047 Back in the time there was an issue with color temp mode reading this PR comments. Would you be able to provide measurements for this color mode to fix the profile?

Ok, I'll try.

@marthubner are you still able to do this?

Hi, sorry for the delay, just started the measurement process. If all goes well, I will post the result tomorrow.