high speed labeler application
veluxllc
Posts: 30
in Propeller 2
Has anyone developed a motion algorithm/routine similar to CAMMING devices?
I'm looking to develop a high speed labeler. I'll need touchscreen and nigh speed input response.
How these work is they get a trigger and quickly accelerate a feed motor/stepper so that it tracks a predefined speed or tracks an input speed via a square wave. This is usually a single channel from a quadrature encoder, though I'm likely to do quadrature inputs as well. Needless to say, the output speed compared to the quadrature input speed are not always nice numbers to play with.
Let me know if you've seen anything like this. Would be nice to use the P2 on it, but P1 may work as well.
Other options/ideas are Modbus and/or etherCAT communications as well as standard ethernet. There will be multiple analog and multiple motors on this beast as well. will be fun.
I'm looking to develop a high speed labeler. I'll need touchscreen and nigh speed input response.
How these work is they get a trigger and quickly accelerate a feed motor/stepper so that it tracks a predefined speed or tracks an input speed via a square wave. This is usually a single channel from a quadrature encoder, though I'm likely to do quadrature inputs as well. Needless to say, the output speed compared to the quadrature input speed are not always nice numbers to play with.
Let me know if you've seen anything like this. Would be nice to use the P2 on it, but P1 may work as well.
Other options/ideas are Modbus and/or etherCAT communications as well as standard ethernet. There will be multiple analog and multiple motors on this beast as well. will be fun.
Comments
Myself, Mickster and ManAtWork all have some experience I believe.
99% of this thing is already thought out. Just wish to make sure the whole gearing thing is done based on proven principles. trying to target a moving position when you have ramp up/down speeds that can vary... My assumption is you target the worst case acceleration to target position at highest speed then dilute that over the slower speeds......
May have to control a 10A AC heater element too.. what a pain this is going to be...
It could be fun, a prop with a small interface, responding to signals to index a stepper, controlling temperature and some pneumatic valves, lol.
Was your labelling done with the PROP? label size? Feed/label index rate/speed?
I haven't looked too hard in the Obex but I doubt you'll find such code.
The problem is you never know WHEN a box is going to show up on the conveyor and you also must label several items in the box. So, if 5 jars are in there, you have to label all 5 while they move, and center the labels. And just for fun, you don't know the dimensions ahead of time, so I'll be measuring the length of the box, and calculating the fire points and firing to place the labels evening in that run.
From time to time boxes show up on the belt with several items in them. Different items or always the same (jars of same size...)?
The labels are predefined or you need to print them "on the fly"?
If the items are not of same size and the boxes cannot be centered and aligned somehow on the conveyor you'll need a laser scanner or camera to idetify their positions and sizes. I think the machine vision will be the hardest part. Once you know the positions where the labels have to go the rest is relatively easy.
If the items can be "serialized" somehow, I mean they appear only one by one and not more than one somultanously side by side, you only need a Z axis servo if you dispense the label synchonized to the conveyor speed.
The worst case is if you have to process multiple items in parallel of different heights. Then you need a 3-axis machine which has to be a lot faster than the conveyor as it has to jump sidewards and folow the conveyor direction at the same time.
Are you specifically looking for a quick solution or an excuse to utilise the Prop?
Maybe I missed it but I still have seen no confirmation that the P2's Smart-Pin-Quad-Decode has been proven. Not that we absolutely need it because it's easily handled in code.
This sounds like a flying-shear type of project. I would probably use variable ratio "electronic gearing" because "tracks a pre-defined speed" might not be reliable. Will the conveyor motor be closed-loop?
I have tested it and it works. I don't actually use it because I haven't managed to generate timestamps for counter-changed events which would be necessary for high resolution speed measurement. So I had to fall back to pin polling and WAITPAT. But for simple position counting the quad decoder works.
Thanks for the info.
boxes coming in will be controlled laterally, so they will always go directly below labeling head.
I've actually done this exact application before and printed on labels AFTER they were applied, then scanned and verified them. The issue is, that was a 30,000 dollar labeler head. It was used because of a short development timeline and now, things are different.
as boxes come down, they'll pass a photo eye and be measured via encoder counts while the eye is blocked, then that measurement will be divided into the number of labels to be applied. Shortly after that, the labels will be applied. Second box can be read in while this these are processed, so a fifo count targets. What I did in the past was for each trigger point, I had a stack of registers that were set to a value for each apply. that value said how far away the next label application was from starting. All registers in this stack would be decremented at the reading of an encoder pulse. When they reached zero, it would trigger and be removed from the stack.
The conveyor is not closed loop. The speeds can vary as it's a simple single phase AC conveyor and some drive components are simple belts, so dust and belt tension will affect speed. Flying sheer is VERY similar to this, if you know of a reference.
Product all same height as it comes through, that's all taken care of. Funny, I JUST did a 3 axis machine for a customer using Mitsubishi simple motion module and servos. customer gave us his test run on hot and cold parts Friday... 100% good parts.
Do we know what the final P2 chips will sell for? Partner is curious, but I'm like... we get this working, a few dollars difference in the one chip is nothing since it replaces several others.
It would be good to start seeing some basic motion code adapted to this chip. It's completely capable of complex work. If analog control is stable enough you could probably drive galvos nicely.
In any case, I'd rather not buy motion chipsets for this since potential customers may need low level control of things. Easier to keep everything together and write one MMI than to do an MMI, interface for motion ships, driver sets, etc.
So any flying sheer references out there?
Marc
I would consider downloading every white paper and manual available.
The first ever dedicated motion chip was the HP HCTL1000 which was developed by Dr. Jacob Tal who then founded the Galil corporation (currently located 1.3 miles from Parallax according to Google).
It appears that your application would require nothing more than the electronic gearing. The conveyor encoder count would represent the command for the movement of the applicator. I use this technique often. It's quite similar to an encoder handwheel where the handwheel is the master that commands a slave axis to move at a specific ratio.
I agree. I'm quite familiar and have used such in my travels as well, but at industrial level. Just looking for proper reference code, any platform, that can ben adapted to do it properly.
My plan is a core to read and define WHEN labelling happens and monitor operator buttons and other signals, a core for actually reading encoder and driving the motor, another core for handling touch display and interface. There may be a core for ethernet and handling specific data and communications. Not sure if I'll try to implement modbus or ethercat on this one. Ethercat on this platform (an open standard) is very fast and would set this platform up for automation. Heck, even Arduino is being used as a control platform for $49 at automationdirect.
Thanks for the white paper, it's pretty high level compared to what I'm looking for.
Again, I know gearing is needed. Just looking for proper reference to do just that portion.
Marc