Dimmable LED kitchen light (Arduino, APDS-9960, RGBWW led)

[mfalkvidd]

@TheoL I didn't know they existed. What's the difference?

[BartE]

@mfalkvidd a video with cool 8 bit music :-) nice

[Boots33]

Great looking light setup...... now if you could just get the the sensor to control the kids as well I might be able to get them to do the washing up too. ;)

[TheoL]

@mfalkvidd Haven't checked all details. But for most less accurate but much cheaper. I believe I paid about 8 times less for the 9930 an aliexpress.

[gregl]

@mfalkvidd - Nice work.
Q: How close to the sensor does your hand need to be? and can this distance be controlled?

[Nca78]

Great project, I have ordered a 9960 and I can't wait to receive it and play with it, this could be a good start.

@TheoL said:

@mfalkvidd Haven't checked all details. But for most less accurate but much cheaper. I believe I paid about 8 times less for the 9930 an aliexpress.

I didn't know about this one, thank you about the information ! From what I see quickly the 9930 is only a proximity detection sensor, it cannot sense movement like the 9960. It can sense distance so it could be used for this application but it has shorter range, 10cm instead of 10 to 20 for the 9960.

[Cliff Karlsson]

Could this also have been done using a ultrasonic distance sensor?

[mfalkvidd]

@gregl said:

@mfalkvidd - Nice work.
Q: How close to the sensor does your hand need to be? and can this distance be controlled?

The sensor starts registering at about 25cm/10in distance and reports maximum proximity at about 5cm/2in (anything closer than that results in a reading of 255).

setProximityGain might affect this, but I haven't tried.

[mfalkvidd]

@Cliff-Karlsson said:

Could this also have been done using a ultrasonic distance sensor?

My experience with the HC-SR04 is that consecutive readings vary too much. A smoothening algorithm (take the median of 5 readings for example) might work, but one reading of the HC-SR04 takes 60 ms which already is 3 times longer than the sample rate I use for the ADPS-9960 so the dimming would probably "stutter" a lot. The speed of sound is simply too slow :)

[sundberg84]

@mfalkvidd - Great work! This is exactly what I had in mind - now i can follow this great example and spare me some hours! Thanks!

[mfalkvidd]

I would like to remove the need for holding still to "lock" the light level, but it is a bit hard since the sensor will get several readings while I remove my hand which will set the light level at all sorts of values.

Setting a timeout that "resets" the level to an earlier value when the sensor stops detecting should be possible, but will require some experimenting to get right.

Implementing a PID regulator might be useful as well, but would require revisiting some long forgotten math skills.

[Yveaux]

@mfalkvidd Nice project man!
Did you consider protecting the sensor somehow from direct air contact? Kitchen stuff stends to get real greasy over time...

[mfalkvidd]

@Yveaux said:

@mfalkvidd Nice project man!
Did you consider protecting the sensor somehow from direct air contact? Kitchen stuff stends to get real greasy over time...

Thanks!
I did not consider that. I think the sensor will sense transparent plastic and glass but I haven't tested, so maybe it is possible to put some protection in front of it.

[TheoL]

@mfalkvidd I have so far not been able to cover the sensor. It's very sensitive. So if somebody knows a material that can cover the sensor while keeping it operating, I'll be be much obliged.

[scalz]

@mfalkvidd
I have played a lot with 9960 few weeks ago, as I have a project with it, I will show you maybe.. ;)

• theoretically 9960 is better than 9930. If you look at datasheet of 9930 it's mentioned as proximity (no gesture engine in it). whereas 9960 is.
• sometimes I had to tweak few offsets in lib for sensitivity or thresholds (on my side, because there is a very nasty while loop in the gesture process..)
• for covering it, you need a material with 80% min of IR transmittance or it won't work well or not at all..
  So with this kind of sensor, without a special material, you need a hole (enough large, see datasheet).

[Nca78]

@scalz said:

• for covering it, you need a material with 80% min of IR transmittance or it won't work well or not at all..

And the datasheet says 1mm thickness maximum for the glass/plastic and also 1mm maximum air gap between sensor and glass. They also suggest if you put a glass to put a rubber layer with just two holes for the led (1.5mm) and the photodetector (2mm) to avoid crosstalk (reflection of the IR light from the led will generate fake proximity or movement detections).

In short it sounds much easier to not use a cover, or else to make a "big" hole for the full sensor size.

[scalz]

@Nca78 i think behind a glass plate it's less reliable, even if using rubber layer because this sensor can be sensitive to light fluctuations etc, imho not the best sensor behing a glass plate but i have not run so much attempts on covering it, so..
but it's a fun sensor :)

[mfalkvidd]

With the cost of the sensor down to $7.40 (from $15 when I first bought one) and even lower when buying multiple sensors I think it is better to just replace the sensor when it becomes too dirty.

[mfalkvidd]

Today I learned that the APDS-9960 will detect steam. The kitchen went dark when I poured up boiling pasta and the steam ventured under the sensor. Exciting times ;-)

[Yveaux]

@mfalkvidd said:

APDS-9960 will detect steam

More general, it will detect heat (IR radiation), so putting your toaster below it might also turn the lights off (dark-roasting :metal: )