Sensors for PET bottles?

Thread Starter

jjotjjotjjot

Joined Apr 17, 2011
7
Hi! I need to detect PET bottles for my project. I'm looking for a sensor that I can connect to a microcontroller. I've seen some commercially available ones such as Omron's E3-ZB, however they are quite expensive.

I've noticed that PET sensors are photoelectric in nature, can anyone explain how they work? I would like to understand how they work and maybe I could create a simple circuit for it? Or are there other ways of sensing PET bottles? Thanks in advance!
 

wayneh

Joined Sep 9, 2010
16,114
Figure out what wavelength you need and c'mon back. PET absorbs at specific wavelengths and that's what you'll need to find. Reverse-engineering is a great tool. There are a range of LEDs to choose from as light sources, and you'll need a detection circuit that helps increase your sensitivity.

Do you need this thing to really work, or just demonstrate the principal for a science fair?
 

BMorse

Joined Sep 26, 2009
2,675
Since most PET bottles ARE clear, it would be hard to detect them reliably with plain photoelectric or even prismatic photo-reflective sensors, the type of sensor that would be able to detect those are quite expensive, ..... trust me on this one, I design and manufacture Reverse Vending Machines that sort and count Aluminum cans, glass bottles and PET plastic @ www.Kansmacker.com
 

Markd77

Joined Sep 7, 2009
2,806
I think I read somewhere that PET bottles are quite clear to UV compared to other plastics. I think it was something to do with leaving river water in them out in the sun to sterilise the water.
 

Thread Starter

jjotjjotjjot

Joined Apr 17, 2011
7
Figure out what wavelength you need and c'mon back. PET absorbs at specific wavelengths and that's what you'll need to find. Reverse-engineering is a great tool. There are a range of LEDs to choose from as light sources, and you'll need a detection circuit that helps increase your sensitivity.

Do you need this thing to really work, or just demonstrate the principal for a science fair?
I really need this to work since I'll be using it to sort out PET bottles from other materials. Will the wavelength vary depending on the surface of the PET bottles? Or if it still has water inside? Thanks!
 

Thread Starter

jjotjjotjjot

Joined Apr 17, 2011
7
Since most PET bottles ARE clear, it would be hard to detect them reliably with plain photoelectric or even prismatic photo-reflective sensors, the type of sensor that would be able to detect those are quite expensive, ..... trust me on this one, I design and manufacture Reverse Vending Machines that sort and count Aluminum cans, glass bottles and PET plastic @ www.Kansmacker.com
I'm actually trying to build a recyclable sorter that will be able to sort out aluminum cans, steel cans, PET bottles and glass containers for our final year project at school. I will use a microcontroller for the system. Can you give me some tips? Thank you very much!:)
 

wayneh

Joined Sep 9, 2010
16,114
Well, a key issue for choosing wavelengths (you may need more than one) is the absorption of likely contaminants. So you need to know how PET absorbs compared to glass and water, and any other likely contaminant. With a little looking I found UV and IR spectra of PET. For example:
PET Spectrum.jpg
One thing I noticed was a bit of talk about PET fluorescing weakly. If nothing else in the stream fluoresces, that could give you a good jump in selectivity.

But I suggest you get some help from your colleagues in the chemistry department, the guys with spectrophotometers. They'll likely have a nice database of comparative spectra for the stuff in your junk stream. They may also have good ideas for a strategy for your project.

The sad bottom line is that this is going to be really difficult for the obvious reasons you've already noted. Labels, dirt, dyes, full or empty, physical distortions, etc.

If you're reinventing a wheel, you might be better off to go buy the wheel. Go find a used one cheap.
 

BMorse

Joined Sep 26, 2009
2,675
I employ a "Clear object sensor" to detect the clear PET bottles..... which is very expensive! but, if you are trying to sort the PET from an array of different materials, a high speed camera with some imaging software would be much better, same type of sensors that are used in trash sorting conveyor lines, and the system just blows a jet of air under the PET to "blow" it into a different chute than the tin and aluminum materials...
 

strantor

Joined Oct 3, 2010
5,159
This sounds like a job for a human. There will be no room left for MANpower (read wideswept unemployment, ultimated demise of society) if even jobs such as digging through rubbish are taken & dominated by machines.
 

wayneh

Joined Sep 9, 2010
16,114
This sounds like a job for a human.
Pattern recognition is indeed a job that humans are wired to do easily, and is very hard to teach a machine.

I like the idea of a "clear object" identifier for a machine as a first pass sort tool, maybe followed by a density sort (to remove glass) and then a human.
 

someonesdad

Joined Jul 7, 2009
1,583
Send me a sample of a PET bottle and I'll measure it's UV absorption with respect to a UV LED I have. It should only take a few minutes. Send me a PM if you're interested.
 

BMorse

Joined Sep 26, 2009
2,675
The machines I design, sort the PET from aluminum cans, and the glass (the glass is also separated by color.) all other materials are rejected by the machines.....In order to reliably sort all types of items, you will have to have a system that can differentiate each commodity type from each other.... and as for PET, there is not an actual Sensor available in the market that can specifically detect PET plastic, so you will have to employ several methods to work together to be able to pick out just the PET from other materials..... I use several different types of sensors to be able to differentiate each commodity, and my machines also keep track of the different sizes, color, material types .....
 

THE_RB

Joined Feb 11, 2008
5,438
He said he needed to *detect* PET bottles. Ultrasonics will do detect them, and some sorting cab be done by size.

An additional optical sensor can detect clear/opaque, metal detectors can detect cans etc.

I don't think any one sensor will reliably discriminate PET from all other objects on the conveyor.
 

wayneh

Joined Sep 9, 2010
16,114
I don't think any one sensor will reliably discriminate PET from all other objects on the conveyor.
The standard is the human, which uses a single (pair of) sensor(s) and a lot of signal processing power. I wonder if a video system with a "facial" recognition system tuned to PET water bottles instead of eyes and noses might do a decent job. I wonder how you could get a computer to recognize clearness? So simple for you and me.
 

THE_RB

Joined Feb 11, 2008
5,438
That's why I thought of ultrasonics. They will "detect" the clear bottle just fine. Then an optical sensor may likely fail to detect a clear bottle. Those two would get a lot accomplished especially if the conveyor system was optimised to run only one item, at a time through a bank of sensors (like a commercial sorting machine). The ultrasonics may detect size quite well too, in sorting items.

If they just had a big conveyor full of junk and needed to spot the PET bottles in it then I think you might be right an advanced visual recognition system will do an OK job but is not trivial and not the type of thing to ask for help with on a hobby electronics forum! ;)
 
Top