Can't get Scribbler Barcode to work
I'm copying this from the orignal thread in the Scribbler forum. This is my last post from a thread I started on that forum. Please look at my thread by the same name in the "GUI programmers forum" for the posts leading up to this. I'm still looking for some help on this. Here is the transferred post:
Phil,
I didn't get as far as trying the concatenation of barcode identification. I loaded your proposed code for detecting the white lead-in to the barcode. I got really strange behavior. I added music after the barcode detection to indicate a false barcode. That music always played the instant the Scribbler entered the white area. So I added several more serial 8X all white detection to be certain the detection time was long enough. I still got the false barcode music as soon as the scribbler entered the white part of the paper. By the way, if I allowed the scribbler to continue, I got the false indication again after completely crossing the bar code and leaving the white area after the barcode. Just for the heck of it, I took a sheet of flat black poster paper and put that in front to the barcode printout. Now I got the false indication about the middle of the black paper. That doesn't make any sense. How did the program get around the need to find a white area before entering the barcode detection routine? Could it be that the line detectors are actually getting a reading off the black paper? I haven't had a chance to try the sensor test routine that is built into the GUI interpreter. That could tell me if the line sensor really is reacting to the black paper.
Can you give me any more suggestions for troubleshooting this program? I also tried putting LED ON routines on every all white detection false indication and never got any lit LEDs. I don't understand that either. It should certainly have happened on the black paper. These Scribblers (I have 2 of them) will correctly indentify a barcode if I use a simple detection program and start the Scribbler on the white lead-in of the barcode printout. So I've got to believe that this is cockpit problems and not a Scribbler malfunction. What's your take?
Dave W.
Phil,
I didn't get as far as trying the concatenation of barcode identification. I loaded your proposed code for detecting the white lead-in to the barcode. I got really strange behavior. I added music after the barcode detection to indicate a false barcode. That music always played the instant the Scribbler entered the white area. So I added several more serial 8X all white detection to be certain the detection time was long enough. I still got the false barcode music as soon as the scribbler entered the white part of the paper. By the way, if I allowed the scribbler to continue, I got the false indication again after completely crossing the bar code and leaving the white area after the barcode. Just for the heck of it, I took a sheet of flat black poster paper and put that in front to the barcode printout. Now I got the false indication about the middle of the black paper. That doesn't make any sense. How did the program get around the need to find a white area before entering the barcode detection routine? Could it be that the line detectors are actually getting a reading off the black paper? I haven't had a chance to try the sensor test routine that is built into the GUI interpreter. That could tell me if the line sensor really is reacting to the black paper.
Can you give me any more suggestions for troubleshooting this program? I also tried putting LED ON routines on every all white detection false indication and never got any lit LEDs. I don't understand that either. It should certainly have happened on the black paper. These Scribblers (I have 2 of them) will correctly indentify a barcode if I use a simple detection program and start the Scribbler on the white lead-in of the barcode printout. So I've got to believe that this is cockpit problems and not a Scribbler malfunction. What's your take?
Dave W.
Comments
You might try adding some delays between the checks for white. It could be that 8 (or more) checks happen too fast to be useful.
-Phil
I'll try that, but I'm skeptical because the white sensor is satisfied when rolling over all black poster paper. If I test for white 8 times and do that multiple times with delays in between, might it not be the case that the sensors will see white 8 times, delay, see it again, delay, etc. The problem, it seems to me is with the sensors themselves. Why should they ever see white when crossing black paper? This paper has a very slight sheen to it, but it is black. It could be that the sheen indicates the ability of the paper to reflect light and it might reflect enough IR to satisfy the sensors. But, if they are that sensitive, then the white sensing method will never get them over the vinyl floor. Because it is a waxed floor, it is really reflective.
If my speculation has merit, then my question would be how to make a challenge using the barcode printouts. My next experiment will be to push the Scribblers across the black paper when loaded with the sensor testing program in the GUI. I'll do the same observation over the floor.
Dave W
Even black surfaces will exhibit "specular reflection". This is the reflection that occurs from the slick surface itself, as opposed to the "diffuse reflection" from the underlying pigment. Black poster board, being very smooth, likely has a very reflective surface. Flat black matte board would be better, but it's much too expensive to use in this kind of app. Posterboard spray-painted with a flat black lacquer or enamel would probably work well, though.
And you're right about the IR. What looks dark to our eyes may very well appear bright in the infrared.
One thing you might try, if you think the line sensors are too sensitive, is to stick a piece of Scotch Magic Tape over the emitters. This stuff makes a pretty nice diffuser and may be just enough to kill the effects of sheen. (I haven't tried it, though.)
-Phil