bottle inspection machine

[BlogRahul]

I found a video on Youtube showing how to inspect defectives  bottles on the conveyor

Link for video

 

inspection cabin tell which bottle is defective and rejection station send the defective bottle into rejection bin. I wonder how defective bottle get rejected on belt I don't understand logic because inspection cabin and rejection system are far away. How does rejection system know that which bottle need reject at rejection station

Does anyone have any idea

[wraper]

How does rejection system know that which bottle need reject at rejection station

Line moves at known speed, so the time when defective bottle needs to be removed is known. Also there may be an optical sensor which detects the bottle and can also be used for counting them.

[BlogRahul]

Thanks! This means that belt speed should be always constant. We will not get the correct result if the belt speed is more or less.

Do you have any idea if the belt speed is adjustable then How the rejection system will work ?
 

[wraper]

I don't know about this exact system but it's quite easy to regulate the belt speed in closed loop or measure its speed.

[mikeselectricstuff]

An inspection camera can also potential measure the speed

[Ian.M]

Its trivial for the inspection station to queue rejections and delay actuating the rejection ram till the belt has moved the required distance to the rejection station.  If there's a transition to another belt in-between, a bottle counter just before the rejection ram may be required.

[BlogRahul]

Its trivial for the inspection station to queue rejections and delay actuating the rejection ram till the belt has moved the required distance to the rejection station.  If there's a transition to another belt in-between, a bottle counter just before the rejection ram may be required.

I really don't understand how the rejection will happen when the belt speed is adjustable?  I found that the encoder used in this kind of machine

[tooki]

Speed x time will always give you distance. Or you could literally just measure distance: suppose you put an encoder on your belt that gives 1 pulse per cm. If you know the rejection station is, say, 250cm from the vision station, then you count 250 pulses and then trigger rejection. The speed of the belt doesn’t matter.

[Carel]

I once worked at a factory that produced machines for the yeast (and penicillin) industry. Yeast was extruded, cut to pieces, packaged and sealed. If it was sealed well, the package was hard.

So a short distance further down the line a pneumatic device attacked the package, if it went too far due to softness, a gate opened behind the package and off it went with the same device. 1980. The best solutions are simple. And it  was fun to see.

[bdunham7]

Speed x time will always give you distance. Or you could literally just measure distance: suppose you put an encoder on your belt that gives 1 pulse per cm. If you know the rejection station is, say, 250cm from the vision station, then you count 250 pulses and then trigger rejection. The speed of the belt doesn’t matter.

On a long-looped conveyer, you could also have a reference ID of some sort, visual or otherwise, for each bottle position.  So if the inspection unit flags a defect, it also flags the ID of the spot on the conveyor and then the kicker can get it when that ID gets to it.

[quemajico]

Bottle conveyors allways have an encoder, to seek preciselly the position on the conveyor, so the controller can track the exact position at the inspection point and follow the bottle to the rejection point.
Problems arrise when the line adjust speed and can occur some slip or displacement to the bottle. May times thre are also photoelectric switches to try to match the supposed position whith the real one.   
Usually also exist displacement registers to track all the bottles from the inspection point or points to the pusher, for each bottle, at the point where enters the machine, a register tracks the position and the status of the inspection (maybe more than one), and with this status, the machine selects also the rejection point, as sometimes, there are different pushers to save or destroy the bottle.
Best regards

[BlogRahul]

Speed x time will always give you distance. Or you could literally just measure distance: suppose you put an encoder on your belt that gives 1 pulse per cm. If you know the rejection station is, say, 250cm from the vision station, then you count 250 pulses and then trigger rejection. The speed of the belt doesn’t matter.

The hard part of this system is to understand how an empty bottles are rejected at the right place on the belt.

I assume we count 2000 pulse from inspection to rejection  I don't understand if all bottles are empty from  inspection to rejection point, then how the bottles will be removed 

[Steffalompen]

I don't understand if all bottles are empty from  inspection to rejection point, then how the bottles will be removed 

Either quickly with one pusher, with multiple pushers, or stop the line entirely because something is obviously very wrong.
A line of bottles resemble a shift register, and that is a useful clue to a possible path to minimize uC programming.

[BlogRahul]

A line of bottles resemble a shift register, and that is a useful clue to a possible path to minimize uC programming.

Can you please explain how the rejection will happen based on location?

There is no such microcontroller in my knowledge that has inbuilt shift register features. Will shift register Ic have to interface with micro ?   

[Steffalompen]

There are so many possibilities and I don't know your production line. I think unless I need a uC for something else to do with this I might forego with it entirely and use the bottles as clock pulse and ADC their status into a sequence of rejection with s.regs.

[BlogRahul]

I have searched more about this topic on the internet but I don't get much information. I don't know how the location of each bottle will be tracked. I don't understand the logic of shift register with microcontroller.

[Steffalompen]

By chance I went and googled "programming shift register" to see if I could find one done only in code.
The first result was programming a PLC for a bottle filling line 

Say there's 8 bottle positions between your scanner and your mechanism. When the scanner reads an anomaly it puts it into the register as a one, 00000001. With incrementing bottles it shifts it left 00001000 etc, perhaps another anomaly occurs 00010001, it reaches the end where the mechanism reads the one and outs the bottle 10001000.
This gets rid of timing an accurate delay, and of juggling potentially many anomalies each with it's own increment counter in code.

[BlogRahul]

By chance I went and googled "programming shift register" to see if I could find one done only in code.
The first result was programming a PLC for a bottle filling line 

Say there's 8 bottle positions between your scanner and your mechanism. When the scanner reads an anomaly it puts it into the register as a one, 00000001. With incrementing bottles it shifts it left 00001000 etc, perhaps another anomaly occurs 00010001, it reaches the end where the mechanism reads the one and outs the bottle 10001000.
This gets rid of timing an accurate delay, and of juggling potentially many anomalies each with it's own increment counter in code.

Thanks for the reply. My problem is that I don't understand the program for PLC. I only work with microcontroller

My question is Do I need to connect shift register Ic with a microcontroller ? I don't know any micro that has inbuilt shift register features


 

[Steffalompen]

I realise you are using uC and not PLC, but the principle is the same.
By now I hope you have looked beyond code and that you are onboard with the very simple analogy between bottles on a conveyor and digital shift registers. They're basically the same.

Yes. (and they can be connected in series)
Or it can be done in code.

[Psi]

The conveyer is likely driven by a stepper motor and the control system is advancing the belt position by a number of steps per second.

Hence the speed of the production line can be adjusted to whatever is needed in real-time but the control system can always maintain tracking on how far bottles are moving through and know exactly when they will pass any future point along the line.

The example above is pretty trivial.  It's more impressive seeing a stream of widgets falling off an edge while a computer scans, identifies and then aims and fires a blast of compressed air at exactly the right time to alter that parts landing position into a different bin.

(There is a video of this but I cant seem to find it)

[BlogRahul]

I realise you are using uC and not PLC, but the principle is the same.
By now I hope you have looked beyond code and that you are onboard with the very simple analogy between bottles on a conveyor and digital shift registers. They're basically the same.

Yes. (and they can be connected in series)
Or it can be done in code.

Which Serial-in, serial-out shift registers  IC will be suitable with 8 bit uC for the project

just google searched and got 74LVC595A is an 8-bit serial-in/serial

[tooki]

You don’t need a shift register. You need to think about what a shift register does, and why doing something similar in code would solve your problem.

[rpiloverbd]

What I can understand from the animation, either there are multiple sensors in the rejection block, or they are using image processing.
I could see many types of defected bottles here. Such as-
Empty
Half empty
Wrong cork
scratched cork

A filled bottle can be differentiated from the empty one by optical sensor, the same sensor that we use in line followers. A pressure sensor/weight can do the task too. Obviously, a filled bottle weights more than an empty or half empty one. Black cork can be detected by optical sensors too.

If they are using a raspberry pi with a camera, they can detect all types or faulted bottles by image processing. The best way to do this is to apply machine learning.

[BlogRahul]

You don’t need a shift register. You need to think about what a shift register does, and why doing something similar in code would solve your problem.

I assume 8 bottles fit between the scanner and rejection point.  there is a variable that stores 8-bit data ( 00000000 ). scanner scan the bottle and put 1 in a shift register if bottle defects otherwise position would be 0 and position will sift for next bottle

This 00010010 sequence show that bootles at 2nd and 5th position defect 

The rejection point only checks the last bit if it's 1 or 0. if the bit is 1 then bottle will removed from line

But Its not undersood how the bit positon will shift for ecach botle between scanner and rejection

 

[Renate]

As stated above, there could be multiple sensors at multiple locations. Maybe one looks at caps at one location and another checks for correct filling level at another location.

This doesn't alter the algorithm.

Code: [Select]

Sensor A
 |
 |     Sensor B
 |     |
 |     |     Sensor C
 |     |     |
 |     |     |     Reject
 |     |     |     |
 |     |     |     |

 0  0  0  0  0  0  0  everything ok

 0  0  1  0  0  0  0  B detects

 0  0  0  1  0  0  0  shift

 1  0  0  1  0  0  0  A detects

 0  1  0  0  1  0  0  shift

 0  0  1  0  0  1  0  shift

 0  0  0  1  0  0  1  shift & reject for B failure

 0  0  0  0  1  0  0  shift

 0  0  0  0  0  1  0  shift

 0  0  0  0  0  0  1  shift & reject for A failure

 0  0  0  0  0  0  0  shift & everything ok