RGB Values Mystery .. .. .. .

[Musicmanager]

Morning Guys

A couple of nights ago I got incredibly bored and did a little mooching in a box of odd bits I bought at a Sunday Market some time ago. I came across a small module marked TCS 34725 and a quick google suggested it was a 'colour sensor module' for which I found quite a bit of info including a plan to hook it up to an Arduino, which I promptly did. The plan used the Serial Monitor @ 115200 baud so I added a small TFT to make it easier to follow .. .. .. .

It seems to work OK, but the values given for Red, Green & Blue are always high 3 digit and mostly 4 digit numbers not the 0 - 255 I was expecting. I had a good google and I can't find any reference to a colour scale with that sort of range and those numbers to me are a bit meaningless.

Does anybody recognise the scale please ?

MM

 

[Pommie]

What values do you get if you completely cover the module?

Mike.

 

[Musicmanager]

Hi Mike

R:26; G:15; B:16; C:28;

Lux: 0;

I think that's to be expected;

There is an LED on the board ( I turned it off for the pic ) and the sensor reads the light reflected from the object colour .. .. .. at least, I think that's what it does .. .. so completely covering will prevent any reflection.

MM

 

[Musicmanager]

Incidentally, ignore the 'Same' , 'Diff', & 'ORANGE' bits of the TFT; that's just me playing about with my ambitions !

MM

 

[Pommie]

If you're using an LED to illuminate it then you're limited to the spectrum of the illuminating LED. Maybe try a small torch (flashlight ) bulb.

Mike.

 

[Nigel Goodwin]

If you're using an LED to illuminate it then you're limited to the spectrum of the illuminating LED. Maybe try a small torch (flashlight ) bulb.

Mike.

As the LED is part of the board it's colour is presumably correct for the complete design - and a torch bulb gives a really horrible colour light, which wouldn't be correct for the sensors - not to mention torch bulbs have VERY short design lives (single digit hours!).

 

[Pommie]

Didn't realise it was a built in LED. If so then a rather neive design?

Mike.
Edit, one day I'll learn how to spell that word.

 

[Musicmanager]

I did try varying the white light source .. .. but it only affected the colour temp. & the 'Clear' reading .. .. ..

It seems to work perfectly .. .. I can get recognisable readings for each primary colour without a problem, and I suppose one solution would be to use the Arduino to divide the results mathematically, but that may affect accuracy some.

I was convinced there was an RGB scale I didn't know about .. .. .. .. .. .. .

Thanks for your thoughts .. .. .. .. .

MM

 

[gophert]

The device has 16-bit resolution so a 3-digit (decimal) number is not much.

https://ams.com/documents/20143/36005/TCS3472_DS000390_3-00.pdf

 

[Nigel Goodwin]

Didn't realise it was a built in LED. If so then a rather neive design?

Mike.
Edit, one day I'll learn how to spell that word.

There's an 'a' in it somewhere - but it's 'one of those words'

But as for 'naive', probably not, it's designed specifically for it's intended use, and (presumably) has a correctly specified LED fitted.

 

[ronsimpson]

This part looks like the same part I used but from a different company. " Texas Advanced Optoelectronic Solutions, Inc. / TAOS " Yes the same part. Used many 1000 of these.
It is hard to get "dark". Light leaks in and through the walls of a box. " 3,800,000:1 dynamic range "
I think the dark number should be in the 0 to 5 range.
Any way in production we had a dark test and then substracted that number from the readings.