Sensor for a New Project

Projects General Guidance
1

Visualize a large conveyor as in my crude sketch below. Material is vacuumed to the conveyor table with air fans, a carriage cuts patterns; when the pattern is complete the material is advanced and the operator trims out scrap and pulls the cut pattern from the far end of the table. The conveyor doesn't move very fast, maybe between 6-12" per second, which becomes important in a minute.

conveyor-sensor
conveyor-sensor1500×1000 121 KB

Note the "split." Part of the manufacturing process for the material as they are putting it on rolls, they will cut out a section in the middle of the roll and use it for QA testing. They use wide masking tape on both sides of the material to re-assemble the material and complete a full roll before shipment. The split is generally around 2" wide.

The problem: often the operator is at the far end of the table pulling the finished pieces and doesn't see the split as it comes off the roll and onto the table. The result is the pattern gets cut across the split (as shown) and is now scrap, the pattern needs to be cut in a contiguous piece.

My idea: It doesn't seem outside the realm of possibility to mount some sort of sensor, driven by an Arduino, over the near end of the table to detect the split. If a split is detected an alarm will sound and the operator can pause the conveyor and "skip over" the split.

The question: what sort of sensor would work for this?

The material is a nylon based fabric coated with PVC. It can be any color, from white to black but is generally a color - yellow, red, blue, orange. What we need is something that can be calibrated with the current color and detect a significant change as the "split" passes under it. For example, calibrate the sensor to a yellow material, when the split passes under it the alarm sounds.

As mentioned the conveyor doesn't move real fast, but the sensor would have to be sensitive enough to detect a change in material. I imagine lighter colors like yellow and white may be a bit challenging.

The manufacturer of this device doesn't have anything in the works, and the material manufacturers are a bit proud of their QA methods and have no intention of guaranteeing rolls without splits. :smiley:

Edit: battery powered at first, likely a 5-12 volt wall wart will be fine to power it. The "sensor alarm" will likely be clamped to one of the devices over the table and we'll be able to mount the sensor alarm within an inch of the material.

2

Map Wheel.
https://www.amazon.com/Calculated-Industries-6025-Digital-Measure/dp/B002O16VKQ

3

You might try something like the AS7262 multiband color sensor (a simple spectroscope), which could be calibrated on the fly to detect sudden color changes in materials.

Keep in mind that what a color sensor sees is very different from human eye response, and the colors and intensity of light reflected from any material depend very strongly on ambient illumination. I imagine that considerable experimentation would be required to make a reliable "split" sensor.

Another option would be a camera and image processing. It is easy to detect boundaries and straight lines, so a computer could be taught to recognize the "split" in images.

1 Like
4

No matter what you decide on for a sensor, you are going to need to test to see what actually works. Since the masking tape I am familiar with has a matte surface, I suggest a test for reflectivity of IR from the tape will be noticeably different from the product material.

1 Like
5

What's the budget?

6

Both of these are great ideas, thank you. The reflectivity is a good point, what I didn't say is sometimes it's not masking tape on the cut side, it's a "paper tape" similar to the brown packaging tape. The cut side is always the uncoated side, so this could be a more simple solution instead of looking for a "color change." I already have some IR sensors I could dink around with.

Not much LOL . . . actually if I can show that it just might work, they'd pay whatever is required but have to get a proto in place. This material is hella expensive so the goal is to stop cutting across splits and throwing it in the dumpster.

I've already looked at industrial sensors and didn't find anything suitable, even if I found one they're off the hook.

7

Sounds like hot air balloon material!

8

It's actually for the best river rafts in the world. :smiley:

9

I was thinking something like this may be worth trying:

1 Like
10

Okay I'm back . . . been struggling with component issues on my "fun" project LOL . . . after even more research . . .

Over at Pololu I did pick up a pair of QTR-1A reflectance sensors, analog version. The problem I see with these is that they need to be within 1/8" of the surface to get a strong reading on a spinning black and white disk, as little as 3/8" away and the signal drops off. Might work, might not, I can probably get that close or create a "wheeled" case over the material. I see potential failure since the yellow stocks are very close to the tone/reflectance of the "tapes." Pololu says these are their "old line" of sensors.

So I also picked up a QTRXL-HD-01RC "long range" sensor that claims to read as far away as 80mm. Trouble is the text on the product page is the same as the QTR-1A, showing the same graph . . . probably just web page laziness. This one is a digital version.

I liked what I saw with the AS7262 but a couple things made me gunshy. Primarily their implementation relies a lot on SparkFun products with very few resources to hack it together any other way. The other was price, this is an experiment for where I work (and am not paid enough in the first place LOL) and I don't even know if I can make it work.

I wound up getting an Adafruit APDS9960 which is not only a spectral sensor, it is demonstrated as a gesture/proximity sensor as well. What really sold me was the video demonstrations (showing simple breadboarding) and extensive setup support and documentation . . . and of course it was only $7.50 USD.

Will post back if any combination of these works to any extent on the toughest materials. My plan is to power it with batteries for portability, experiment with the detection then add a buzzer or speaker for an alarm . . . then eventually a wireless TX so if the operator is out of range their receiver will alarm.

Or maybe I'll throw one of these 100 watt amps at it and wake everyone up. LOL

11

Have you done any tests to compare the IR reflectivity of the various colors of stock material and the IR reflectivity of the tape? (The natural light colors may be similar, but the IR reflectivity may be considerably different. Or not.)

1 Like
12

Tests . . . with what? :smiley: That's kind of why I am getting both a reflective and color sensor. I don't even know if either of them will work. Worst case scenario is a yellow material we use commonly, the "tape" is the color of paper packing tape. There will be some contrast issues there, it may not work. But it's worth throwing under $20 at it to find out. :smiley:

13

Oh, sorry. I thought you already had the QTRs...

1 Like
14

Hey no worries and thank you for responding! This should be a fun one to play around with, will update the thread when they get here.

1 Like
15

OK, assuming the 'split' is 2" and your conveyor runs at 12"/second. This means the split passes underneath the sensor within 2/12 seconds = ca 150ms.

For full scale 16-bit color resolution, your sensor requires an integration time of 175ms (datasheet, page 4). This means that you're going to have to make do with (significantly) less than the 16-bit resolution the sensor has to offer.

You may or may not have sufficient color resolution at shorter integration times, you may need to sample the color a couple of times and even out the readings to get rid of fluctuations and spurious readings. The color of the split and how different it is (in either/or red, green and/or blue component) from the base material will affect how sensitive the system will need to be.

If your detection window is 150ms or thereabouts, you may have to settle for something like a 20-30ms integration time, so you can still get a few samples of the split material if that passes.

You could help things if you can influence some characteristics of the split tape; e.g. make it very different from the base material in terms of UV reflectance or something like that. This would allow you to make/buy a sensor that's specifically selected for that property. It would make things a little easier.

On the other end of the spectrum would be some kind of machine vision solution; may sound like overkill, but on the other hand, it may end up being easier to get up and running in an industrial setting. Besides, machine vision is getting pretty common these days, which means costs are coming down as well as the implementation complexity.

16

Well everyone, it looks like the APDS 9960 FTW.

I used a Nano AT, 1620 LCD display, Piezo buzzer, a few LED's and a Goodwill mouse to hack together a working prototype. Details in the video description. Runs on 3 AA's and seems to work well up to 3 - 4" above the surface.

It could stand a lot of improvement, like a remote buzzer for the operator to carry (it's usually loud in this environment) and electronics design but it works!

The point may be moot though, I may not be with this company much longer but it was fun building this thing.

1 Like
17

Thanks for the update, and sorry to hear about your job situation.

1 Like
18

Thanks but nah, I think it's good. Officially I'm retired, the job was just extra money. I could continue working there doing "busy work" but the company is not projecting enough sales and I think it's time to go. Shame though, this thingy would have made life a lot easier LOL

1 Like
19

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.