Society of Robots - Robot Forum
Electronics => Electronics => Topic started by: rbtying on January 31, 2011, 07:24:37 PM
-
Hey All,
I've got a sensor here called the "Rockwell Automation 42EF-D1MNAK-A2", no datasheet included. It says on the package that it is
DIFFUSE
SENSOR
0.5m(1.6ft.)
and on the right of that
INPUT: 11-30VDC
LOAD: 100mA MAX
OUTPUT: NPN
(2) COMPLEMENTARY LO/DO
So most of that makes some sense, but I'm really wondering how to interface to an "NPN" output. My goal for this sensor is to be used as a line-following sensor. Any help would be significantly appreciated.
-
Hi,
LOAD: 100mA MAX
OUTPUT: NPN
So most of that makes some sense, but I'm really wondering how to interface to an "NPN" output.
It's a regular "Open Collector" output.
Think of it as a BC547 (or similar) with the emitter grounded and the base driven from the rest of the circuitry, with the collector as the output, able to sink up to 100mA (non inductive).
An open collector output is used to keep noise down among others and if you have some length of wire to the receiving circuit, place a pull up resistor at the farthest end of the wire (as seen from the photo interrupter).
My goal for this sensor is to be used as a line-following sensor.
Sorry to say, but I don't think you'd have any luck there.
-
Thanks, Soeren!
I was thinking that something like that would be true. Oh well, explains a bunch. (Not my choice of sensor, FIRST just got a bunch of them for free, apparently).
A IR led and matched IR-sensitive transistor will work, right?
(If not line following, I wonder what these sensors are for? Perhaps collision detection?)
-
Hi,
I was thinking that something like that would be true. Oh well, explains a bunch. (Not my choice of sensor, FIRST just got a bunch of them for free, apparently).
Oh, if you get like 8 of them and if they come with the fiber optic lines and if you place the loose end of the fibers fairly close to the ground (i.e. < 10mm), perhaps you might be able to adjust them to a point where they can be used.
But frankly, I think they're way to large for anything but a really large robot.
They're nice for other purposes and you can use them to read simple bar codes (if they're printed large), again provided you have the fibers (which in their own right can be used for lots of nice things ;D).
The fibers are made of a bunch of fibers going into each "hole" (emitter and detector). They're shaped like a "Y" and are intertwined into a single bunch in the business end, so it lights up the spot that is looks at and nothing else - try it with a visible color LED at one branch, look ino the other branch and point the joined end at different stuff - if you use a white LED, you can see the color of what's under the joined end through the open (not lighted) branch.
A IR led and matched IR-sensitive transistor will work, right?
That'll be a cautious yes.
Remember to check the floor where it's supposed to work - and the tape used for the line.
Some materials may appear dark to the human eye but "white" when viewed in the IR part of the specter, so if you don't test it, you run the risk that your sensors won't see the line at all or with so low contrast that it cannot differentiate.
A high quality generic line sensor setup should be able to switch through colors for the emitters and have a "wide band" sensor - I did a design once with red, green, blue and IR LED's, so at the start it runs through the colors to decide which has the highest contrast on the floor/line combo and if the contrast lowers significantly, it makes a quick burst through the colors again (which helps on tracks where the floor material changes over the course.
If using just one color and not knowing the track, I'd go with around 550nm (yellowish green) LED's and photo transistors with at least around 70% of their peak sensitivity at that wave length, as that is close to the peak sensitivity of the human eye - and after all, it was a human who laid out the track and found the contrast to be good, so in most cases it should work (except if it's a white line on a sick green floor ;D)
Using red light, both red and white will appear "white", while green will appear "black".
Using green light, both green and white will appear "white", while red will appear "black".
Using blue light, both blue and white will appear "white", while yellow/orange will appear "black".
Etc.
Decades ago, I worked with the then emerging technology of bar codes and a large company had a huge run of printed labels made (and applied on the goods) with white lines on a red background and I was consulted since they could absolutely not read a single one of them - Their reading pens used a red LED! (the scanners used everywhere today was grossly expensive back then, so most applications used "bar code pens").
I explained the problem and the guy responsible turned pale by the thought of a new run and application of hundreds of thousands of bar code labels (and the likely flack he would get), but I got him off by landing a sale of a large number of pens using IR - he was happy, his boss was happy, my boss was happy, so who was I to cry ;D
The morale of this is to always check the contrast, whether it's about bar codes, line followers or any other kind of optic contrast detection.
(If not line following, I wonder what these sensors are for? Perhaps collision detection?)
They're mostly used in production lines, to check if there's a box or whatever passing by, for controlling start/stop or routing of objects.
If used without the fibers, some of them should work well as obstacle detectors (no distance info though).
-
try it with a visible color LED at one branch, look in to the other branch and point the joined end at different stuff - if you use a white LED, you can see the color of what's under the joined end through the open (not lighted) branch.
Of course you are not supposed to use a wide spectrum light source with optic fibers...
The different frequencies have different reflection rate... It works like the prism principle...
Anyways... Fairchild QRD1114 are nice reflective pair... Used them many times, from encoders to line following... The transistors are pretty fast at switching plus both transistor and transmitter can be removed from the case for added flexibility. The later can prove very useful when dealing with building a quadruped encoder.