Multi-servo product metering

JRoss

Lifetime Supporting Member
Join Date
Sep 2006
Location
Dillsburg, PA
Posts
184
I'm looking for more process oriented advice, but for those who care I'm using an AB ControlLogix L3 processor and Kinetix 6500 servos.

My application is a metering system for feeding stacks of product into a wrapper. I have to line the product up with the lugs on the flighted conveyor of the wrapper, so we're using servos to do this. This system is using six servo driven conveyors in series. The final one in the sequence is speed-matched to the wrapper, and the other five are used to gradually meter the product so that the spacing matches the flights on the wrapper.

My plan is to build on a similar system that I've done which only had a single servo following an encoder directly mounted to the flighted conveyor. In that case, I used the encoder as a master axis and "geared" the servo to it starting with a 1:1 gearing ratio. Then each time the product made a metering eye on the downstream end of the metering conveyor, I would calculate the lead/lag and adjust the gearing ratio up or down to compensate for error. The first couple of products would invariably be off, but product would quickly be adjusted to the correct position.

This system needs multiple servos because we are dealing with stacked product instead of single product, so I can't adjust things very fast, or product will go flying. But I can't wrap my head around the order that things should be chained together. Should I slave Stage 5 to Stage 6 (final conveyor), Stage 4 to Stage 5 and on back the line? Should I slave all the stages to Stage 6? Should I do something else entirely? Any advice would be much appreciated.
 
I went to a A-B servo class last year and I remember the instructor talked about setting up a virtual drive and letting the real world drives follow that one. You may want to get an Allen-Bradley technical engineer in there. They have came out with a lot of new servo drives and controllers in the last few months. Some of the newer ones are even cheaper and give you more bang the buck. Another option may be going to one of their 3 day classes. You sound like you could skip the basics and enroll in one of the more advanced classes that deals with setting up profiles.
 
Peter - A little elucidation? If it were easy to me I wouldn't be asking... If you mean it's easy to slave to virtual, it is extremely easy in the Rockwell software. If you mean the application is easy, tell me how you would approach it.

Thanks!
 
I'm guessing this is less a control probelm and more a mechanical problem.

Are you receiving products at the initial conveyor with the correct spacing already or do you have to adjust for the correct spacing for each product?
Is there a point where you will have one and only one product on a conveyor at a time? Or is the number of products and spaces that will fit on a conveyor variable?
Do you have absolute product control at the conveyor transitions? That is, do you know EXACTLY where one conveyor lets go of a product and the other conveyor takes it over?

Keep in mind that the Logix motion platform supports move superposition. You can be geared to a master and still perform a time based move on the same axis. The position command values are simply added and become a composite position command for that axis.

Keith
 
Product is being stacked by hand upstream then goes through a series of side-by-side conveyors to speed it up and take it down to a single lane before it gets to the metering servos. So when I receive the product, it doesn't have any particular spacing. Final spacing should be 6.5". First three staging conveyors are 16", next two are 12", final one (master) is long and mates up with the flighted conveyor of the wrapper. I have sensors at the upstream and downstream side of each of the six servo conveyors (total of 12 sensors).

On a previous project I using the superposition, but my implementation caused the metering conveyor to jump around under certain circumstances. I found it better to adjust the gearing ratio, which made things change more slowly.
 
We have something similar. Three belts feeding into the lug chain of a wrapper. Each belt has a photoeye wired to that belts drive registration input. The lug chain is the master for each belt. The amount of correction is limited to prevent over correction.
You can gear the belts to the master but I would use a linear cam. Hope I'm not way off base with what you are trying to do.
 
I have done this with a stacked product - totally random positioning - rslogix V17 or V18 Kinetics

Method that I used

All speed up /phasing belts locked to a Virtual Axis, the virtual axis phasing to the wrapper is adjustable by the operator.
Each phasing belt has a fixed gear ratio to the master (ratio set up during commissioning) - smooth CAM tables (curtsy of Peter N) are used to adjust for each clip phase error on every belt.

Most of the correction is performed on the first detection of the product (the product is moving at the slowest speed and can cope easiest) and I have an earlier PE on the first belt to give advance warning of where the next product is.
 
MichaelG - That sounds exactly like what I need. My program is close to complete using a similar method. I've got a virtual master that will be synced with the wrapper (more on that below) and am gearing each of the phasing belts to the virtual with a "fixed" gearing ratio to be determined.

Here's where my method diverges. For each phasing belt, I am calculating the phase difference between the wrapper and the product leading edge, and then adjusting the gearing ratio up or down by a small percentage based on the phase difference.

I think it will work on this different application, but I don't know, so I'd like more information about the way you are using camming. If you would be willing to actually share your code, that would be fantastic!

One other question: how did you have the wrapper tied in to the phasing belts? Were they controlled by the same PLC, or did you have an encoder on the flighted conveyor, or some other means?
 
JRoss, I was busy and had only a short time to write. I was hoping that the word phasing would be enough.

I would gear everything to the down stream line. The phasing command can be used to close or create gaps. The phasing command should be something simple like the slave ( upstream ) line will offset its position by distance S while the master moves distance M. The distance S is determined by the needs to open or close a gap. Distance S usually varies from move to move. The distance M should be a long as possible and still get the phasing done in time. This command is used all the time by our customers that are moving logs end wise and need to create gaps between the logs after cutting so they can be easily diverted down stream.

If there a multiple upstream lines the the same phasing command should be issued to all upstream lines.

I am not a fan of cam tables in applications like this. Gaps vary so the slave is always making a different offsetting move. The controller should be able to do phasing on-the-fly without the need of tables.

Here is another example where veneer must be aligned to make a wooden billet that is be be pressed. The second part of the video shows the sheets of veneer that move under two photocells that provides alignment data to the motion controller. The motion controller then issues the phasing command. In this application I think both sides could be the master and the belts that were behind were always sped up to be in phase with the leading belts.
http://deltamotion.com/peter/Videos/Auto Lay Up.wmv
All the sequencing was done by the motion controller. There was a PLC5 that moved the billet underneath.
 
Peter - Thanks for the information, very helpful. I'm a bit unfamiliar with the term phasing, so I didn't immediately latch on to it. Rockwell allows the use of an MAM (motion axis move) instruction on top of gearing for "phase advance/retard". I played around with it on my previous project, but had a little trouble controlling it at the speeds we were trying to achieve. Since the product was coming thick and fast, with only one metering belt, I found it simpler to adjust the gearing up and down to get the result I wanted. Perhaps I need to revisit the use of phasing on this project. I may try it both ways to see what happens.
 
Rockwell allows the use of an MAM (motion axis move) instruction on top of gearing for "phase advance/retard". I played around with it on my previous project, but had a little trouble controlling it at the speeds we were trying to achieve.
I don't like superimposing moves in this case either, at least not superimposed moves that aren't also geared. If the MAM isn't geared to the master then what happens when he master stops or slows down? The MAM keeps on going possibly bumping into something.

One must also do some calculations to calculate the right speed, acceleration and deceleration for the MAM. The user must specify the means on how to achieve the goal. I prefer to have commands where you specify the goal and the controller then calculates how to achieve it. We have 3 different ways of doing this. One is the phasing command.

The IEC phasing command requires two parameters a phase offset and the distance the master needs to move for the offset to occur. If the master slows down then the phasing slows down too. If the master speeds up so does the phasing. Everything is synchronous.

There should be a way of gearing the superimposed move too.
If there is a phasing command that offsets positions and is geared to the master then this really should be easy. I should make a video.










This is why I like the way we do phasing with a special command.
 
You hit the nail on the head as to why I previously chose to advance and retard using gearing adjustments. Not as accurate, but a lot less to go wrong. I'd love to have a phasing command as you describe it!
 
Originally posted by JRoss:

I'd love to have a phasing command as you describe it!

Then change controllers. Peter would be more than happy to line you up with what you need. What makes more sense? Spending an additional $2000 on a controller and ending up with a system that does what it is supposed to do or wasting $2000 in engineering effort hacking a system that doesn't do what you want to make it work only something like you want it to? I know as engineers we tend to undervalue our own time but this is a case where you might want to place a little more value on it.

Keith
 
odd this is exactly what I have achieved with the rslogix controller

each conveyor is geared to the virtual axis
the phasing was done with a CAM move so that the phase change was based on the master axis distance traveled
- The CAM move is ontop of the gear
oh and the calculation for the CAM move - Master distance fixed (almost one product cycle), slave distance the required phase distance

rockwell is not smart enough to provide simple commands where you say "as the master moves x distance phase shift y distance" on top of a gear, you have to "add" the available commands together MAG + MAPC

If you just use MAG commands to change the gear ratio the speed change to the new ratio is a linear ramp without jerk
Rockwell is definitely NOT a motion controller
 
Last edited:

Similar Topics

This is a question for anybody out there who has an in depth understanding of AC motors and how they work with VFDs, particularly in a...
Replies
16
Views
3,868
I have an application that allows an operator to jog a servo forward or reverse, slow or fast speed with push buttons. As I understand the MAJ...
Replies
3
Views
2,820
Hello, I had a question about benchmarking the Rockwell FactoryTalk View SE software. I've installed in the neighborhood of 30 different...
Replies
2
Views
247
Hello, I'm trying to test Acopos multi using Automation Studio. Ready LED is green blinking, how can i know what error of Acopos on Automation...
Replies
1
Views
97
Good Afternoon , I'm sure there are many Ishida Muti-Head Weigh Systems . We have a Ishida CCW-M-214W Multi-Head system . We are...
Replies
1
Views
412
Back
Top Bottom