Measure movement of follower arm

Thread Starter

Quad2

Joined Jan 29, 2018
13
Hi - this has been puzzling me for some time. I am under a NDA (Non-Disclosure Agreement) on this project so I can discuss any specifics. However I am working on the electronic control systems of this device. It is a project I first started with about 5 years ago - in that time we have produced a device that works and performs 'OK'. We would like to make it better and that is basically where this post comes in.

The device is used to sort items of different diameters (ranging from 5mm to 25mm), the items are being produced at the rate of about 1 every 150mS. As each item is produced it passes under a sprung arm which forces the item to sit at the bottom of the channel where it is then moved onto a compartmentalized conveyor.

I would like to measure the deflection of this sprung arm somehow. I have been considering using a coil based oscillator with a metallic pin attached to the arm such that it moves in and out of the coil. Or some kind of optical system, where light levels are reduced based on movement. I have a small budget so I cant use anything fancy like a camera! The arm is moving a maximum of 90 degrees. Anyone got any suggestions!
 

John P

Joined Oct 14, 2008
1,782
You haven't said what receives the data, and it makes a lot of difference!

But assume that it's a microcontroller, potentially attached to a computer. You could use an analog sensor like the optical system you mentioned. Or a Hall-effect sensor (magnetic) or just a potentiometer, though with repeated rapid movement over the same range, a pot doesn't sound right. But an optical encoder could do it. Your magnetic device with an oscillator should work too, though it sounds a little more complex than the other things. A simple position sensor can be implemented lots of different ways!
 

panic mode

Joined Oct 10, 2011
1,791
nobody cares about your proprietary application but - you did not share any of key factors for the sensing task at hand - such as size requirements, resolution, accuracy, signal range, operating voltage, cost etc.

there are tons of ways to measure this but due to speed of the process, i would prefer contactless and no moving parts.
linear inductive distance proximity sensor comes to mind.
they provide analog output 0-10V for example, can directly sense metal (like this arm), are fast more than enough and usually have at least 10-bit (1024 counts) resolution.
for DIY there are chips that sense rotation of external magnetized target, etc.
 
Last edited:

btebo

Joined Jul 7, 2017
89
I agree with panic mode - contactless might be best since things are moving so quickly...

There are some amazing vision sensors out there that can do something like this. I've used some very simple ones on our machines - but there are very sophisticated ones out there. Look at Keyence www.keyence.com. They have excellent tech support and even though I have only bought several hundred dollars worth of parts, a salesman has come by and shown me some really neat stuff.
 

Thread Starter

Quad2

Joined Jan 29, 2018
13
nobody cares about your proprietary application but - you did not share any of key factors for the sensing task at hand - such as size requirements, resolution, accuracy, signal range, operating voltage, cost etc.

there are tons of ways to measure this but due to speed of the process, i would prefer contactless and no moving parts.
linear inductive distance proximity sensor comes to mind.
they provide analog output 0-10V for example, can directly sense metal (like this arm), are fast more than enough and usually have at least 10-bit (1024 counts) resolution.
for DIY there are chips that sense rotation of external magnetized target, etc.
I am quite aware nobody cares about the project, however having signed a legally binding agreement stating I will not make public any of the details of this device, and given a court is never a nice place to be, I am sure you can appreciate my caution.
 

Thread Starter

Quad2

Joined Jan 29, 2018
13
You haven't said what receives the data, and it makes a lot of difference!

But assume that it's a microcontroller, potentially attached to a computer. You could use an analog sensor like the optical system you mentioned. Or a Hall-effect sensor (magnetic) or just a potentiometer, though with repeated rapid movement over the same range, a pot doesn't sound right. But an optical encoder could do it. Your magnetic device with an oscillator should work too, though it sounds a little more complex than the other things. A simple position sensor can be implemented lots of different ways!
Sorry got a bit ahead of myself there - ultimately this needs to end up inside a PIC32 processor. If the sensor as an analogue output then I would use either the PIC32 to convert or a separate chip depends on resolution. There is absolutely no real time interaction with a PC, although all data is buffered to a SD card which can be read in a PC, or viewed over a USB connection, but this has little to with the measurement.

The arm will move a minimum of about 10 Degrees to a maximum of about 90degrees, I need to be accurate to 1 degree, which has been the problem we constantly hit. A pot has given great results as we can build a nice bit of signal conditioning and stick it through a 16Bit AD. But after not very long at all that pot wears out!
 

Thread Starter

Quad2

Joined Jan 29, 2018
13
I agree with panic mode - contactless might be best since things are moving so quickly...

There are some amazing vision sensors out there that can do something like this. I've used some very simple ones on our machines - but there are very sophisticated ones out there. Look at Keyence www.keyence.com. They have excellent tech support and even though I have only bought several hundred dollars worth of parts, a salesman has come by and shown me some really neat stuff.
I was talking with a guy from Keyence recently about another project! Didn't cross my mind to discuss this project!
 

Papabravo

Joined Feb 24, 2006
12,695
I am quite aware nobody cares about the project, however having signed a legally binding agreement stating I will not make public any of the details of this device, and given a court is never a nice place to be, I am sure you can appreciate my caution.
Right - so why would any of us in our right minds want to help you? It just isn't worth the aggravation. Lancaster's advice is to go public or go home. After 5 years you could hardly be making rapid progress to market.
 

Thread Starter

Quad2

Joined Jan 29, 2018
13
Honestly, this just goes to show what a short sighted narrow minded, unforgiving place the internet is! Literally all i wanted to do was bat a few ideas about with like minded people! I'm not asking or expecting you to design the item for me. Yes 5 years is a long time, and it is selling nicely, I'm wondering if you have ever heard of continuous improvement! I guess probably not but it's something we like to do keep our sales figures up! Anyway as it seems that the information i have given you is insufficient to enable anyone (thanks to those guys who have actually provide some useful input btw) to suggest any suitable methods of measuring this deflection (Angle of deflection, approx sizes, and frequency, along with system clarification) I shan't bother this site anymore.
I was talking with a guy from Keyence recently about another project! Didn't cross my mind to discuss this project!
Honestly, this just goes to show what a short sighted narrow minded, unforgiving place the internet is! Literally all i wanted to do was bat a few ideas about with like minded people! I'm not asking or expecting you to design the item for me. Yes 5 years is a long time, and it is selling nicely, I'm wondering if you have ever heard of continuous improvement! I guess probably not but it's something we like to do keep our sales figures up! Anyway as it seems that the information i have given you is insufficient to enable anyone (thanks to those guys who have actually provide some useful input btw) to suggest any suitable methods of measuring this deflection (Angle of deflection, approx sizes, and frequency, along with system clarification) I shan't bother this site anymore. I will let you amateurs go back to twiddling with the knobs on your radio grams or whatever it is you do! :)
 

atferrari

Joined Jan 6, 2004
3,578
In this context, "deflection" of the arm, what does it actually mean? Angle? (Non-native, that's why I ask).

If you could translate angle to linear motion, I would consider LVDT.
 

LesJones

Joined Jan 8, 2017
2,417
The required resolution of the sensor will depend on the smallest difference in size between any two of the objects. For example if the smallest difference in size was 1 mm then detecting about 40 positions would probably be enough. (As there could be 21 different size objects between 5 and 25 mm diameter. But if there were any two objects that only differed in size by 0.01mm then there could be 2001 different sizes so you would need to have a resolution of about 4000 positions.

Les.
 

jpanhalt

Joined Jan 18, 2008
8,331
I am not clear on whether the arm that is being deflected is rigid and rotates, or is flexible and being bent.

If the former, then a rotary encoder would be high on my list of approaches.

If the latter, then an ultrasound or light-based position detector would be my direction.

Which is it? If flexible, how long is it? Can it be longer than the minimal distance from where it is fixed to where it contacts the objects?
 

John P

Joined Oct 14, 2008
1,782
People do sometimes ask questions here about projects that they can't or won't describe enough to get an idea about what they're doing! But here it seems clear enough, especially where Quad2 says "The arm will move a minimum of about 10 Degrees to a maximum of about 90degrees, I need to be accurate to 1 degree, which has been the problem we constantly hit. A pot has given great results as we can build a nice bit of signal conditioning and stick it through a 16Bit AD. But after not very long at all that pot wears out!"

This isn't a very complex machine, and I think we've got enough information to offer a few realistic ideas. Anyway, "Remember the human" as we're told. It's easy enough to roll your eyes and mutter something for your own hearing only, and move on to another topic.
 

jpanhalt

Joined Jan 18, 2008
8,331
<snip>, it passes under a sprung arm which forces the item to sit at the bottom of the channel where it is then moved onto a compartmentalized conveyor.
I would like to measure the deflection of this sprung arm somehow.<snip>
Those are the terms that led me to doubt whether a modern, say 16-bit, rotary encoder would work. Sure, that encoder would give <<1° resolution, but is the arm rigid or springy? If springy, then resolution will probably be less and dependent on the length. No answer yet. I see no non-human responses anywhere in this thread.

I believe we have all seen confabulators with vague ideas hide under the pretext of an NDA. That is no reflection on the TS, but I cannot see that an answer to my question would violate any NDA. Moreover, while an answer on a public forum might be patented, it might be very hard to enforce.
 

djsfantasi

Joined Apr 11, 2010
5,838
We know the following:
  • object sits in a channel
  • the space above the channel is clear
We can surmise the following:
  • the object is either cylindrical or spherical.
  • the object is likely cylindrical
Could a Contact Image Sensor as a linear array of sensors be paired with a laser LED with a diffraction lense that produces a straight line?

No complicated camera and image processing. The height of the object above the channel can be directly read and the diameter thus calculated.
 
Top