suggestions micrologix1100 position control

Have a applicatioin that i hope someone on here may be able to give some advice on. we have built a prototype piece of equipment that involves fabric that is fed through a dryer. There are some rollers that are driven by motors across the top and some other rollers across the bottom that can rotate up or down depending upon the tension. The prototype as you can see from the pictures have two prox switches, one on top and one on bottom. As the tension changes, depending upon which switch makes, the motor either slowly speeds up or slow downs to try and keep the tension constant. It was done this way because that is the way the manager (boss) said it was going to be done.
Well now I have inherited the project and the manager (boss) now wants the prox switches replaced with sometype of a analog sensor. I have done some searching for a rotary type transducer that would give a 4-20ma and use this to control a pid loop to maintain that the rollers stay in the center at all times while the fabric is feeding.
To be honest, not sure if this is the correct way. Any suggestion from anyone who has done anything similiar would be greatly appreciated.
The concept I have thougt about is taking a reading from a encoder that is the master to determine the speed that each of these motors should turn, use that as the setpoint to maintain in a pid loop for each motor, but not sure what is the best way to use the value from the sensor to adjust the output from the pid loop.
any advice or suggestions to help me start thinking in the right direction would be appreciated.

Thanks

irondesk40,

Is this in that same steam "infested" environment that needed better than IP69k gear?

Is the master a fixed diameter? In other words, will the master speed feedback be directly proportional to your fabric linear travel rate?

If so, that makes it pretty simple. You could actually do the application with discrete switches rather than analog.

Basically, you need to adjust for stretch or shrinkage in the material, but otherwise it would be "electronically geared" to the master.

You'll want to match the ramp rates of the master as closely as possible especially if the line stops and starts frequently.

When running, calculate the "ideal" speed based on the master speed feedback.

Then, use a PID to adjust a "stretch" factor from, say .8 to 1.2 and multiply that by your ideal speed. Those numbers are WAGS, you'll have to adjust to your application.

But, the idea is to only rely on the PID for the unknown portion, not the whole speed reference.

Use as much precision as possible in scaling the base speed for best results.

Be sure to put the PID in manual mode when the master is disconnected from the slave. In other words if the line is stopped or there is no fabric in the loop, you don't want your integrator winding up.

If you want to try this with digital feedback, you can simply multiply your stretch factor by 0.99 when at one prox, and by 1.01 when at the other prox. (Again, those are example values which will need to be tuned...) Do that on a controllable periodc basis (self resetting timer) and be sure to clamp the results within known limits. If you do this, set up the switches/flags so that there is a dead spot in the center where neither is made and outside of that center sweet spot, one of them is always made.

A good rotary transducer:
http://www.meas-spec.com/product/t_product.aspx?id=2676

Paul

Thanks for suggestions, will have to read it a couple of times and hope it sinks in.

There is a master motor that will be set to run at a certain ypm. The other motors will be set to run at the same speed as the master, but the rotary transducer idea is to replace the prox switches that are there at the moment.
This has been a on going type project and this is now the 3rd version of the prototype and things keep developing with each prototype.
The current setup is just taking a reading from a master encoder that tells the speed the fabric is running, then the other motors are told to run that same speed, but if one of the prox switches get made, then that speed is slowly increased or decreased to get the desired result.
Now, what confuses me, is if you take the master speed and use that as the setpoint to try and maintain with a pid, then how to take the 4-20ma reading from a rotary type transducer to change the speed for that motor in order to get the dancer back to the desired position.
Will read your suggestions and hope it sinks in through my thick skull.

Thanks a lot

Now, what confuses me, is if you take the master speed and use that as the setpoint to try and maintain with a pid, then how to take the 4-20ma reading from a rotary type transducer to change the speed for that motor in order to get the dancer back to the desired position.

Click to expand...

No, the PID will be controlling the dancer position, therefore its process variable (input) will be the actual dancer position and its setpoint will be a value that represents the desired position (center) of the dancer.

Its control variable, or output, will be the shrink/stretch factor, or adjustment made to the base speed.

Thanks

OkiePC said:

No, the PID will be controlling the dancer position, therefore its process variable (input) will be the actual dancer position and its setpoint will be a value that represents the desired position (center) of the dancer.

Its control variable, or output, will be the shrink/stretch factor, or adjustment made to the base speed.

Click to expand...

Thanks for suggestion, have tried it on a motor test setup and works very well. Used the following sensor. Still need to do some more testing, but looks very promising.
Now have another question. Is it possible to take a Powerflex drive such as the powerflex 40, send a analog signal sensor to a drive to tell the drive the speed to run and then network that to about three more drives so that they run the same speed and then connect a sensor like I have attached directly to each drive and use it like a trim to either speed up or slow down the drive that it is connected to and not have to send a analog output from the plc to each drive and have a analog input for each sensor..
The reason I am asking is that the machine that I am going to modify has a micrologix 1100 and it is maxed out with expansion modules and if this is possible, then I would not have to change out the plc.
I hope this makes sense.
Thanks
Here is a link to the sensor.
http://www.am.pepperl-fuchs.com/products/product.jsp?product_id=24053

Yes, you can do that. The PF40 can be set up to retrieve data from a master using the DSI protocol (RS485). You then can set the P038 (reference source) paramter to the DSI port.

But, it might be easier and have better response if you simply send the same analog speed reference to all of the drives.

Thanks for advice
Now this project has to evolved to the point where now there will be a master motor that will be told to run a certain speed, then there will be two chambers (heat) there will be about 10 motors in each chamber and each motor will be controlled by a drive and also there will be a inclination sensor(dancer) at each motor.
There will be a total of about 20 drives. Now looking at the best way to send the speed signal to each drive, still looking at using the 4-20ma from the dancer to trim the speed that the drive has been told to run.
In the past when there was only a few motors being controlled by a 4-20ma signal from the plc, I just ran a seperate analog output from the plc (micro 1200's,1500's)to each drive.
But with at least 20 drives, looking for a possible better way to send the speed signal to each drive without having to have at least 20 analog outputs from the plc.
I have a 13 slot slc rack and a couple of 505 processors. Was also considering the new micro1400.
Any advice on how someone may have approached something like this by using device net, profibus,ethernet etc. would be appreciates.
We getting to the point where we have been doing quite a few projects in the last few months that required controlling multiple motors that it may be a good time to consider upgrading to controllogics, but our boss who is the director of engineering does not want to approve that kind of expense so we staying with the slc,micro line and rslogix 500
Thanks

irondesk40 said:

Thanks for advice
Now this project has to evolved to the point where now there will be a master motor that will be told to run a certain speed, then there will be two chambers (heat) there will be about 10 motors in each chamber and each motor will be controlled by a drive and also there will be a inclination sensor(dancer) at each motor.
There will be a total of about 20 drives. Now looking at the best way to send the speed signal to each drive, still looking at using the 4-20ma from the dancer to trim the speed that the drive has been told to run.
In the past when there was only a few motors being controlled by a 4-20ma signal from the plc, I just ran a seperate analog output from the plc (micro 1200's,1500's)to each drive.
But with at least 20 drives, looking for a possible better way to send the speed signal to each drive without having to have at least 20 analog outputs from the plc.
I have a 13 slot slc rack and a couple of 505 processors. Was also considering the new micro1400.
Any advice on how someone may have approached something like this by using device net, profibus,ethernet etc. would be appreciates.
We getting to the point where we have been doing quite a few projects in the last few months that required controlling multiple motors that it may be a good time to consider upgrading to controllogics, but our boss who is the director of engineering does not want to approve that kind of expense so we staying with the slc,micro line and rslogix 500
Thanks

Click to expand...

If each of the 10 motors is to follow the Master, I'd use one analog output to the first of the 10 drives and have the other 9 daisy-chained to the first one.
Analog in of first drive gets the speed reference from PLC, analog out of first drive to the analog in of the 2nd, analog out of the 2nd to the analog in of the 3rd, etc.
Sure saves a bundle on hardware as long as your drives have analog ins/outs available.Never mind.
I just realized each motor must have on its own speed reference

irondesk40 said:

...there will be about 10 motors in each chamber and each motor will be controlled by a drive and also there will be a inclination sensor(dancer) at each motor.
There will be a total of about 20 drives. Now looking at the best way to send the speed signal to each drive, still looking at using the 4-20ma from the dancer to trim the speed that the drive has been told to run...
...I have a 13 slot slc rack and a couple of 505 processors.

Click to expand...

Devicenet with a 5/05 can handle 20 drives no problem. We have one system with 46 drives on a single scanner. It would also be easy to migrate to C-Logix later and keep the devicenet network.

Do you have RSNetworx for Devicenet software?

Do you have to send the same dancer signal to all of the drives, or do you trim each one with its own inclinometer?

Do you have RSNetworx for Devicenet software?

No, I have rslogix500 and linx lite.

Do you have to send the same dancer signal to all of the drives, or do you trim each one with its own inclinometer?

The theory at the moment is to send the same master speed sigal to all of the drives at once and then have a inclinometer mounted on a dancer assembly at each motor to use to maitain the speed of that motor.

With 20 drives, you will want some sort of communication network to send the master speed reference. I think parallel times twenty with a voltage signal would be a problem, and series with a 4-20ma signal too. Someone may be able to correct me on this...

Ethernet could be used, but only for messaging with a 5/05. The 5/05 does not support Ethernet I/P for control of I/O. You would need to have delays between messages and it could take several seconds to update 20 drives when the speed command needed to change. We have a 5/05 with an ethernet network to only 7 PF40 drives and we must limit the frequency of messaging to prevent comms issues.

With a devicenet scanner handling the communications, I think you'd have the least problems and the simplest set up except for the purcha$e of RSNetworx.

You could also go for Controlnet, but you'd need RSNetworx for that too.

Paul

With a devicenet scanner handling the communications, I think you'd have the least problems and the simplest set up except for the purcha$e of RSNetworx

Have never used Devicenet, would you have to purchase this software in order to communicate to powerflex 40 drives from a slc505 plc?

You would need the software to configure the network and download that configuration to the devicenet scanner card. You can also use it to monitor the network and the devices on the network.