Subject:
|
Re: colour sensor
|
Newsgroups:
|
lugnet.robotics.handyboard
|
Date:
|
Wed, 26 Jan 2000 18:34:34 GMT
|
Original-From:
|
Chuck McManis <CMCMANIS@MCMANIS.COMnospam>
|
Viewed:
|
1403 times
|
| |
|
|
At 02:43 PM 1/26/00 +0000, E158 (M.FAZAL) wrote:
> hi chuck
> i tried using the photocells i got with the gleason research sensor motor
> kit. i used red green and blue filter, but the values are almost same (with
> a variation of 1 or 2). the variation is much more sensitive to the distance
> between the sensor and the surface. what am i doing wrong?
>
> you also mentioned optical sensor. is that the same as the photocell?
>
> cheers
> fazal
I have typically used phototransistors rather than CDS photocells but both
will work depending on the application.
Sensors are like computer programs, when they do one thing they are fairly
easy, if they do multiple things they are more complicated.
The problem you are having with the photocells relates to their optical
sensitivity.
CDS photo cells are sensitive to light intensity, so you can think of the
light intensity at a given "distance" from the cell as the DC bias. Now you
subtract that bias from the base line and you get the variation due to
color on top of it. In circuit terms you require an "ac" connection.
Now in my circuit that I used to locate colored "area" on a playfield, the
bias level was set by reading a known "white" area. I used a DC bias on an
op-amp to generate a zero reading on white. Then a magnitude comparator to
locate the "colored" area (red/green/grey). The output was essentially 1
"bit" (yes/no) on whether or not I was over a particular color.
In the early 80's there was a company that used four photodiodes attached
to a the top of a dot matrix printer print head that could "scan" color
pictures at approximately 100 dots per inch. Their sensor was, in my mind,
quite clever in that it used one of the diodes with no filter, and then one
each covered by red, green, and blue. The diode with no filter would set
the bias level and then they would extract the color information from the
other three. If you wanted to try something similar you could use a CDS
photocell without a filter in the negative feedback path of an opamp and
the filtered cell on the input. Such a set up should be able to get you
several levels of color information.
The bottom line is, specify what you need to detect, figure out ways to
eliminate as much background noise as possible, then amplify only enough to
get a detection. (to much amplification re-introduces noise)
--Chuck
|
|
Message has 1 Reply:
 | | Re: colour sensor
|
| (...) I posted to this yesterday, but my reply seems to have not appeared. What you need to have, is four sensors, not three. The fourth has no filter, and measures 'ambient' light. You then need to compare the return from this, to the other three (...) (25 years ago, 27-Jan-00, to lugnet.robotics.handyboard)
|
2 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
