So many motion controllers and intelligent servo drives are available today with ProfiBus, DeviceNet, Ethernet, Modbus, etc. People are even having luck with sending motion commands in ASCII strings to get flexibilty from move to move.

Are there still good reasons to buy motion control cards for PLC racks and control a drive with +/- 10V?

 

There are many applications in which the fieldbus based systems simply cannot maintain sufficient coordination. If you are running a conveyor system, or many pick and place systems then the fieldbus systems may be a good choice. If you are requiring tolerances below 0.001" in multiple dimensions, or very high speed coordination even with low tolerances then the fieldbus systems are typically not capable of meeting the spec.

Davis Gentry
Senior Applications Engineer
Delta Tau Data Systems
[email protected]

 

I always use to say PLCs are good for doing I/O and Motion Controllers for doing Motion. Motion controllers are becoming more advanced and incorporating the ability to do the PLC functions... avoid (2) controllers in the system. I hated ladder logic PLC motion - especially if the process has a lot of motion. Simple motion in ladder logic was ok. I think some of the PLC companies relized that not all liked doing motion in ladder so the gave the ability to have a motion program in the PLC that you can execute and pass variables.

Other thing is most Motion control programmers could not program ladder logic and vice versa.

Some motion controllers allow you to get expansion I/O - maybe enough for the machine. Some have ethernet I/O expansion ability and a PLC task in them. Some more advanced motion controllers - PC/Software based allow you to do many axes of coordinated motion, PLC, Vision, Robotics..etc.

> Are there still good reasons to buy motion control cards for PLC racks and control a drive with +/- 10V? <

I hate to say, but I think so yet. Depends on the application and what your trying to do. If you have a machine that is process intensive with a lot of I/O points and little motion along with the need for HMI or OITs, or some network communications...PLC is going to give you that where your motion controller will not. Now you could if you want like many do, just use an intelligent servo drive and communicate to it via your field bus or if the motion is canned - just simple I/O for Go/Done.

 

"Are there still good reasons to buy motion control cards for PLC racks and control a drive with +/- 10V?"

The only reason I can think of is if you are dealing with existing equipment. For example, if you have a servo-controlled machine and all you want to do is replace the PLC, you're stuck using the old +/-10V that the existing servos use.

IMHO, a new system should NOT use +/-10V. As the Delta Tau person mentioned, any fieldbus will do if you aren't coordinating motion (or only coordinating loosely). If you need coordinated motion you have to use a motion fieldbus. Most major PLC manufacturers support some sort of motion bus. Allen-Bradley and Schnedier / Modicon support SERCOS, Siemens has Profibus-MC. Major motion control vendors also have bus systems. Delta Tau has MACRO, Motion Engineering (now owned by Danaher) has SynqNet, several people are doing FireWire, etc.

I don't think the technical differences between various motion buses matter a lot. Pick one you like and go with it. The +/-10V analog command with encoder feedback belongs in a museum with the horse and carriage.

-James Ingraham
Sage Automation, Inc.

 

Thanks for your reply, but my question deals more with interfacing a PLC to a motion control system.

PLC scan rates are so slow that a rack mounted motion controller has to have its own processor anyway. This being the case, does it ever really make sense to put this intellegence in the PLC and restrict functionality to the function blocks that the PLC manuf. develops?

Seems better to me to have the motion control stand alone funtioning as a peer or device on a PLC network. Heck, discrete wire digital is often enough.

I wonder if Controls Guys from the "PLC World" to whom motion might only be a fringe application are getting their money's worth when they buy a PLC rack mount motion control card because they feel comfortable with the vendor.

 

The old +-10V analog servo command is unfortunately the only industry-wide standard that exists. It may be a pain to wire, but it performs well and is wide open. Almost every drive supports it.

For the above reasons, I use it all of the time. Nothing else meets my requirements.

Bill Sturm

 

I agree, motion in the PLC stinks, but you have to look at it from the owners standpoint.

If it is all in one box, you only need one software package to run it and if you plant already has a gazillion of those boxes - you only need to train your people on one system . . . that has a lot of appeal.

I am a big advocate of keep the motion engine separate from the state logic engine. I like to think that different vendors have strong products and weak products and doubt that any single vendor has a lock on the best of everything - pretty much guaranteeing that you are going to compromise on performance if you get it all in the same box.

My favorite solution today is a mix of vendors.

State Logic managed by Control Logix
ControlNet connection to a Delta Tau UMAC that is controlling a host of ACVector and ACServo drives via (+/-10VDC) analog torque command.

If the motors are small enough, we go direct PWM to the amplifiers - but when you deal with smoke stack industries with 10 - 100 HP vector drives, direct PWM is not a practical option for getting off the shelf drives.

 

If you're using UMAC, why not use MACRO? And are you really getting your money's worth buying the UMAC with a network option? Since you've already got the ControlLogix, why not use A-B PowerFlex 700S for vector and servo control? Coordination can be done over SynchLink.

I certainly agree with "best of breed" approach, instead of the "shoehorn in a PLC 'cause that what maintenance knows" approach. I would never try to run CNC off a PLC, or a nuclear power plant off of Linux. (Heh-heh; just wanted to see if Curt and Jiri are paying attention.) Still, I do NOT buy the argument that +/- 10V is "best of" anything. Yeah, yeah, widely supported, blah blah blah. If a new medical technique is invented that works ten times better but requires retraining doctors, would ANYBODY argue that we should keep using the old techinque?

-James Ingraham
Sage Automation, Inc.

 

Well, I'll comment on the UMAC with or without MACRO option. MACRO is a good option if your machine is large or modular in nature for the same reason any other fieldbus is good. It greatly simplifies wiring and the cost savings there more than make up for the additional cost of MACRO. If the machine is modular it can also save significantly on commissioning time as the wiring between modules drops to MACRO, power, and estop. If the machine is relatively compact, and all of your controls are in one cabinet then MACRO is probably just unnecessary added complexity.

As for using ControlLogix - that's fine if you're running a conveyor system. Or a slow pick and place maybe. If you want fast and/or accurate coordinated motion (try drawing a very fast circle with an X-Y table controlled by ControlLogix - as of a year ago the A-B demo I saw turned that into an ellipse) then you need a better multiaxis controller. Naturally I prefer ours, but there are a large number of controllers out there that can be used in most applications and do better than ControlLogix. My
assumption is that with time this will change. A-B will probably put the money and time into ControlLogix to make it a truly capable multiaxis controller. Anyone want to comment with their own experiences to rebut my year old stuff?

Davis Gentry
Delta Tau Data Systems

 

James -

I'll grant you the PowerFlex drives (when controlled via. a decent motion controller they turn in respectable performance), but for motion apps that require accuracy along a path as well as accurate definition of the path (not the same thing) AND the ability to introduce registration corrections on the fly, easy recovery from machine faults, bla bla bla... hard to beat a dedicated motion controller. The UMAC is tops in this regard.

This month we will wrap up a couple of machines each with 100+HP of connected ACVector drives (Control Techniques UniDrive SPs) with ControlLogix on top of UMAC via. ControlNet and a PanelView+1000 all connected via Ethernet to the Internet so that we can upload/download to everything in the box. This really is the best of all worlds - maximum performance in state logic, motion control and drives.

So far we haven't been able to get a MotionLogix system to run this app - but we still give it the college effort every year. We are getting closer. We are locked out of a few big accounts until we either convince them to use something other than the big blue, or until such time as we can coax a few more features out of the box and get it to work.

Until that time comes, we continue in our "Thinking outside the Octagon" approach -

Putting motion in ladder is a sin.

 

Ken said: "Putting motion in ladder is a sin."

In my view, LADDER is a sin. But I'm stuck with it. Of course, Rockwell's people will tell you to use Structured Text instead if it makes you feel more comfortable. I suspect you won't like it any better than you like their ladder logic; I know I don't.

Doesn't change my other point; +/- 10V is a sin. Yes, Davis, even in a small confined cabinet. Analog signals are subject to noise, even over short runs. Accidentally flipping the polarity of the wires causes a runaway. Complexity of wiring is a general problem; there's usually some breakout module for connecting an expensive high-density connector to invidual wires, since 4 axes requires roughly 60 wires. (8 for feedback, for 2 +/-10V, 1 enable, 1 fault, 2 limits, and a home switch.) It is impossilbe to retreive fault data through the analog out / feedback in system. With a network, those faults can be spit back to an HMI. (True, more useful in a larger system than a smaller, but even in a small space you might have a policy against opening an electrical enclosure without lockout / tagout, which would of course obliterate the fault information.) You can set more parameters on the fly. For example, you might normally have torque limited to %50 until some part of the cycle that is known in advance to need more.

Analog needs to go away, and it can't go fast enough for me.

-James Ingraham
Sage Automation, Inc.

 

I agree that analog servo interfaces should be superceded with some sort of digital network, but (and theres always a but) there are no universally accepted digital servo interfaces in the industry.

Someone needs to invent Modbus for servos.

Some say Sercos, but in reality, only a few vendors support it. It also requires intelligent drives with a position loop.

Firewire has no standard software layer.

ME has developed Syncnet, just another limited choice, so far.

Delta Tau tried to standardize Macro, but few bought in.

What can we do?

Bill Sturm

 

James,
For what it is worth - the +/-10V interface is far from dead and while I agree that parameter setting and fault diagnostics are nil with this interface - performance / noise issues are not a problem with well designed systems and well designed systems shouldn't require parameter changes as the motion controller typically runs the drive in torque mode and can do torque limiting far better programatically than could be managed over a network.

When we do want more information from the drive, it is only an RS-485/DeviceNet/Profibus connection away - and in most cases this is more for configuration of a drive out of the box to make life easier for the maintenance folks. As far as drive fault retention, I am not aware of any brand name drives on the market today that do not have a non-volatile drive fault history buffer.

We have integrated Sercos systems, (Sanyo Denki, Indramat, and Serclose with Rockwell) . . . none of these systems can perform at the level of a +/-10VDC system in a high performance coordinated motion control system. ref. www.janicki.com - other integrators have tried and failed to achieve what we have done with these machines. They have resorted to using all manner of networked, fiber-optic based drives and controllers only to eventually call us in to replace them with the analog interface - although not for the reasons you might expect.

Sercos / Synchlink are fine for simple motion apps with pick and place or conveyor control or web handling apps with nominal registration requirements. But they are all in one sense or another spliner apps or at best geared following or cam following architectures. Trajectory management is dramatically dumbed down when run over a network and used for describing coarse waypoints - sure you have great PID servo performance, but your are closing the loop on an ill defined trajectory that is constructed in the drive rather than in the motion controller.

It is in my customer's best interest (hence our best interest) to use the best drive / motion system design we can come up with . . . on smaller systems, nothing beats direct PWM control whether for air bearing stages with linear motors or ACVector or brushless servo . . . as the current levels go up and exceed 50A per axis with 480V systems, off the shelf direct PWM doesn't exist - so we go with standard drives (eg UniDrive SP or AB PF700VC, etc.) = +/-10VDC. I estimate that 150-200 axes / year leave our company in multi-axis motion sytems and fully 70% of them are +/-10VDC.

So in the end, the main thing that endears me to the +/-10V interface is its ability to directly command torque from a PID loop that has coordinated trajectory data that is clocked in at the servo interupt - something that is simply not possible with any networked interface without some level of splining of the trajectory. The only thing better than the analog interface in this regard is the direct PWM interface.

If path accuracy is important across multiple axes of control - you have very few choices for a high performance interface between the controller and the drive. Our niche is high performance motion control - if it can be done with a networked drive . . . great, by all means use it and enjoy it's advantages.

 

Ken,

I agree that "the +/-10V interface is far from dead." I just WISH it would die.

I'm not sure I agree with your comment that "well designed systems shouldn't require parameter changes." You may be right, but for several years now I've been going in the direction of smart drives, and you've been going in the direction of a smart controller with dumb drives. I suppose I have to admit that I have not spent much time designing sytems ("well" or otherwise) using analog in the last 5 years.

As for adding "an RS-485/DeviceNet/Profibus" connection in addition, it adds cost, complexity, and an additional fail point if used during operation. If it's just for "static" configuratopm changes (out of the box, someone swapped out a different motor, etc.) then it's semi-irrelevant; you've still got the 15 or so wires per axis I mentioned in a previous post.

And even if the drives "have a non-volatile drive fault history buffer", that doesn't help you GET the history buffer. If you have to lock out the panel to open it, then that drive is sitting there with its drive faults locked in memory with no way to retreive them.

"We have integrated Sercos systems... none of these systems can perform at the level of a +/-10VDC system in a high performance coordinated motion control system."

This surprises me greatly. Not that I ever thought SERCOS was a panacea, mind you. Still, my experience with +/-10V is that the conversion from digital to analog in the card, and than from analog to digial in the drive, along with inherently limited accuracy of an analog signal, leads to "ceiling" beyond which analog just can't go. (Again, as of 5 or so years ago.) Certainly Indramat and Industrial Indexing Systems would publicly argue the point. (They'd have to, wouldn't they?)

"Sercos / Synchlink are fine for simple motion apps with pick and place or conveyor control or web handling apps with nominal registration requirements."

Admittedly, this is the area I am most familiar with. I guess I have to defer to your experience on this.

Still, Indramat makes CNC controls on SERCOS, including a nice picture on page 5 of their IndraMotion MTX brochure that looks remarkably similar to the shot on the Janicki home page. Granted, it's not real, and it's a lot smaller.
The very odd URL that ends in a question mark is:

http://www.boschrexroth.com/country...oducts/brc/a_downloads/71237AE0104_final.pdf?

GE Fanuc, who I beleive makes more CNC controls than anyone else in the world, swichted to a proprietary digital communication scheme more than a decade ago, and I have yet to hear anyone complain about their perfomance.

"on smaller systems, nothing beats direct PWM control"

I buy that. It saves you the D-to-A to A-to-D conversions, is noise resistant, and means you have a "drive" that doesn't really HAVE any parameters. And, of course, no network latencies.

"I estimate that 150-200 axes / year leave our company in multi-axis motion sytems and fully 70% of them are +/-10VDC."

God, I LOVE getting that kind of information. This is why I bother to read control.com; to run into somebody who has real-world experience doing something similar-but-different who can give a different perspective on things. Thank you.

For the record, of the 50 axes / year we ship, 50% are Profibus, 30% are A-B "Sercos", and the rest are retrofits of existing systems that have to be analog. We don't have co-ordination requirements most of the time.

-James Ingraham
Sage Automation, Inc.
www.sagerobot.com

 

Well,

Delta Tau does have a direct PWM output that is better than +/- 10V, it gives the torque command but using a PWM signal instead of analog. It works at the full speed of the PMAC, and all the coordination is done in the PMAC controller in hard realtime. It doesn't give you fault networking and such, but at least you don't have to use an analog interface if you are concerned with noise.

I wonder about profinet and ethernet powerlink. I am skeptical, but some people (like B&R) are claiming to already be doing nice coordinated moves. I guess there should be a spec for what entitles a tough coordinated move! )

~Ken

 

Good discussion -

I don't think there is yet a perfect solution in the area of motion control interfaces.

Lots of good arguments to have distributed control and for many applications, it provides the best solution - but when coordination of various axes is needed to a very tight uncompromising standard - only centralized control will do.

The interface is a stepchild of the philosophy of control you select which is of course a function of how tight you really want to control the related motion of each of the axes.

The degree of path accuracy degrades slightly with the implementation of coarse waypoints delivered via some manner of network solution and movement of the final trajectory generation and PID loop closure to within the drive.

FWIW, I have grown to like using the nuclear bomb of controllers (Delta Tau) and the more apps our company does with stand-alone and distributed controls, the more stark the contrast I see in directing your motion axes like a classical symphony orchestra vs something that looks more like an after school Jr. High concert band.

 

With direct PWM, you do not need to get fault info
from the drive. All of the info should be already in
the control. The only feedback from the drive is a
current signal, if I understand correctly.

There seems to be two camps, the direct PWM method and
the networked position data. I just wish that either
or preferably both of them would standardize on an
interface that the majority of the industry supported.
I think that both methods could coexist in the
market. I don't think either method by itself would
satisfy all motion vendors.

Bill

 

B&R is doing some incredible things regarding coordinated motion over ethernet powerlink. I personally have developed a 5 axis screw machine and a 6 axis bag making machine. Both of these machines had been attempted with other hardware/software and the B&R was faster, more repeatable and cheaper. You should check them out. Developing the software was somewhat complicated but the results possible were worth every bit of the development time.

BTW:
I know what you mean - what really constitutes a tough coordinated motion application???

 

This is one of those areas where as Einstien said "Things should be made as simple as possible, but no simpler!" As long as people keep trying to make motion "point and click" simple with general class, compromise, solutions there will be problems. Simple motion is simple. Coordinated, high speed, precise motion is not. You can
hide all the complexity but all that means is that a product will work great in some situations and fail miserably in others and all will be mysterious. The hard applications will still have to be solved the hard way and will still require knowing what you are doing. It's like point and click data acquisition, getting meaningless data
is far easier than it's ever been :^)

Regards

cww

 

Delta Tau is the way to go for most if not all systems, though you have to work to make it "maintenance types" friendly, but from an engineering perspective you can't beat it whether it is 0-10VDC, PWM, Macro, etc.