Help with a basic control design

[geesamand]

I have an application where we create a back-forth motion in a cleaning machine. This is done using an explosion-proof rotary actuator (containing a 3600:1 worm reducer) that operates a crankarm to drive our machine back and forth in a slow, continuous motion. There is a control box typically supplied with this unit to provide safe operation of the actuator motor.

I need to eliminate the control box and change the motion to be incremental: i.e. to run the actuator once per day so that the machine moves the actuator crank 1/8 of a turn. Drift is not a problem (absolute position). I want it to be easy for customers to drive this with their plant DCS. I can see a couple of options:
1) Try to fit a rotary encoder on either the input of the actuator reducer or output of the reducer, and let the DCS use the encoder to operate in a closed-loop mode.

2) Have the DCS operate the motor for a set number of AC cycles (say, 900 cycles) that corresponds to the fraction of turn of the output shaft and leave it at that.

I'm not a controls guy. I'm asking for assistance because my proposal should be reasonable for a typical industrial customer running a plant in an area where the equipment must be explosion proof. Your input or pointers to other arrangements that might help with my problem are greatly appreciated.

David

 

[KllrWolf]

What type of motor is beig used - AC, stepper, or servo motor?

 

[waross]

For cheap and dirty I would try counting cycles, adding a few to compensate for slip. Add a position detector of some type to reset the count every revolution. This may be an automotive crank position sensor, an explosion proof limit switch or anything else that gives a once per revolution position signal.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[geesamand]

After some consideration, I've determined that drift is somewhat important as several of these machines usually work together and they need to remain pointed in the same direction. (They need not be perfectly aligned, but they can't go through periods of time where they are out of phase with each other)

I like the idea of using a single position sensor to pick up the "zero" location of the output shaft, because it will allow the control system to reset.

Here's the question: since this will be someone else's DCS, would making my customer program all of this in their DCS be asking too much? (Commercially, not technically speaking - I'm sure for many DCS pros this is still cake)

Dave

 

[itsmoked]

Be asking too much? Heck yes.

I think ideally you'd want to have the customer send a single pulse out to your gizmo and have it advance to the next position.

Include a button to allow manual advancement so in maintenance, setup, or commissioning they can re-sync it with the neighbors.

Use a EXP proximity switch which looks for holes, perhaps in a plate on the output shaft. Have your simple controller run until the next hole shows up.

For different situations you just provide a plate with a different hole count.





Keith Cress
kcress -

http://www.flaminsystems.com

 

[geesamand]

Clearly this part of the project will get reviewed. I was told it needed to be fully transparent and integrated into the customer's DCS. They didn't want to pay for autonomous controls. So the key will be giving enough transparency but also keeping things convenient.

 

[waross]

I was told it needed to be fully transparent and integrated into the customer's DCS.
Sales engineer who doesn't really "get" the details, right?
Getting past this source to find out what the customer really wants and needs may be a bigger problem than the design.
To itsmoked's suggestion add one hole at a different radius and another sensor to provide a check position.
Include a cheap PLC for control and offer an option without controls. See which sells.
All I can suggest for transparency is a plastic index plate. Or does transparency mean making it simple enough that the sales engineer can understand it. (Don't answer that.)

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[geesamand]

Sorry, transparency in this case meant visibility and communication to the DCS.

We've sold this system before with it's own control box. All it needs is power. It does not offer places for the DCS to monitor whether it's in the right position or if it's encountered an error.

Our customer has asked the sales guy to scrap the control box and let them control this by themselves. So one way to do this is to provide enough sensors to make it controllable but not too much.

We really don't have any controls experts in our company and so understanding the risks and challenges of redrawing the boundary of our scope of supply to cut through the controls.

 

[controlsdude]

I think you already have these devices from last design.

Then take the control box off from your last design, leave all the input sensors and output devices on the machine. Wire all these to a junction box, let the customer worry about the rest.

 

[geesamand]

The last design was also different in that it moved continuously and the controls basically made it move, caught errors, and allowed local manual control. It was as open as open-loop gets, really.

This time round they also want to change the motion to move to one of several discrete positions. So new bits are definitely required to ensure it's close enough to one of the positions and that the position can be read from the machine.

 

[waross]

Does anyone else remember the grey code? It was a disk encoding that gave unambiguous position information. With a binary encoded position disk, there may be an ambiguity as the sensors cross boundaries where the alignment is not perfect and more than one sensor changes state at the same position. It was a long time ago. Try Googling "Grey code". Three or four sensors on one encoder disk should suffice.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[waross]

Sorry it's Gray code.

http://en.wikipedia.org/wiki/Gray_code#Position_encoders

Four bits should do.


Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[controlsdude]

There are Indexing head or indexing tables is maybe what your looking for, if your trying to get a set motion or turn of a rotation?

http://en.wikipedia.org/wiki/Indexing_head

 

[djs]

A motor controlled with a good VFD can step amazingly well using limit switches or prox switches. I have used that system on packaging machines achieving better than 1/16" variance on a 30" index.