Hydraulic Motion Control - When is a PLC not good enough?

[agarb]

I did my little speed test on my machine with good results. I have a 5" cylinder with the RMC70 SSI feedback moveing at 0.001 inch/sec. Trends looked smooth I did not see any ruffness at all. I would also add that this is a quality cylinder with low friction seals.

I'm curious, what is the resolution of your SSI feedback?

-----
My application with an HACD and 18 inch bore cylinder. This was a custom cylinder build by low-cost bidder.

Could do 0.1 inch/minute (0.0017 inch/sec) without any problem.
If command was 0.05 inch/minute (0.00083 inch/sec) the cylinder speed about 8% too fast.

I believe I mentioned this before. When I moved control from the HACD to the PLC, and gave a 0.1 inch/minute command, the actual speed was about 8% too slow. I did this 3 times with repeatable results. I didn't try to go any slower with the PLC as I didn't have time.

SSI encoder resolution of 0.00039 inches (0.01 mm).

Actual speeds were back calculated with a stopwatch by looking at postion change on the SSI encoder over a 1 minute period.

Edit: I didn't specifically look for or observe slip-stick. I do remember observing the control outputs to my proportional valve were repeatably dithering around 7% or so with a 0.1 inch/minute command.

 

[CharlesM]

I was getting the speeds back from the RMC plots. Most of the time I use Temposonics position sensors however this is my "test" machine that was made up of a lot of spare parts. It has Balluff SSI position sensors. I would have to check but I bet they are 0.01mm just like you have.

 

[agarb]

The thing I don't like is when this is taken in a vacuum. It should be intuitively obvious that a dedicated motion controller will perform motion operations better than a general purpose plc.

Keith, I believe your "take it into a vacuum" comment was referring the mention I made to Peter about private discussion of my future application. I was referring only to the potential application; as I don't want to reveal too much application specific data in a public forum on a project we have not officially sold, or may have non-disclosure agreements in effect.

Otherwise, I want to keep all other discussion here.

I still have some other posts to read through and digest. (Allscott)

 

[agarb]

Yes, Delta Computer Systems, Inc makes these controllers for Rockwell Automation. I wrote the code for the QS and we modified M02AE code to make the HYD02 and M02AS. That is why I had better know how they work. We make the modules and we make some profit. We don't do tech support for these products. However, I will answer questions on this forum. I have done so within the last two weeks.

If I can't sell and RMC I will try to push the Rockwell products because we still make money. When it comes to hydraulic motion control my opinions are biased.

Thanks, I am now seaching the forum and reading all the old posts on these modules. After consulting A/B list prices, my initial impression is that they were too pricey, then I realized that with our A/B mulitiplier and on a per axis basis they are likely cheaper than the HACDs. What woud be really nice is if there was a 1769 style I/O module that would work with the MicroLogix 1500 and/or CompactLogix.

 

[agarb]

Responding to allscott’s post… yes, I read it all and can’t really disagree with anything.

I think it is really about picking the right tool for the job. In some cases it looks like a PLC can be good enough. AMCI makes several SSI modules for most of the A/B PLCs; not sure what the prices are. Every project I work on has analog inputs and outputs. It would be rare when I didn’t have some spare ones. Otherwise, with up-front planning, I’d just order a 8 channel module instead of 4 channel, at a relatively small cost adder.

I guess the real benefit, in my mind, is more simplicity for my customer. If it is all in the PLC where they already have the tools and some basic knowledge it is much easier for them to troubleshoot and understand. I am sure the customer that my HACD project is going to has no HACD background. This will make things much more difficult for them when it breaks or doesn’t work.

I guess the other thing I don’t like is all the handshaking and co-ordination that must go on between the PLC and external controller. I also despise manufacturers’ finger-pointing when something doesn’t work right. Is it the PLC or external controller that is the problem? I wasted a day on my last project due to getting an HACD with old firmware where the devicenet didn’t function properly. I believe it keeps the project simpler and cleaner if you leave as much in the PLC as possible. Now when performance requirements are in excess of what the PLC can easily provide, by all means don’t hesitate to do something else.

As far a tuning, I’m sure dedicated controllers are superior. But, with an Ethernet connection and the trending that RSLogix offers, a lot can be done there, too.

As an OEM - Unless I have relative experience with a similar application, how do I know what is the right tool for the job? That is why I started this thread.

 

[agarb]

I have movies that show the effects of active damping
ftp://ftp.deltacompsys.com/public/pneu/Preparing%20for%20IAAPA.MP4
This is a pneumatic system which is very difficult to control at a smooth rate.


I will check into it. I think we do. I know I can make it happen. Brad can also do a quick tour just to get agarb started.

Pretty neat video. What type of valve are you using for the pneumatics?

Who is Brad? (Bruce?)

 

[Peter Nachtwey]

Pretty neat video. What type of valve are you using for the pneumatics?

Who is Brad? (Bruce?)

The pneumatic valve is a HR Textron valve. They are a little expensive. Although we can control pneumatics we don't push it. One can do better with the new linear actuators in most cases. We use pneumatics as a testing platform since it is easy do demonstrate the slip stick and the active damping. If one can control pneumatics then hydraulics is easy. We can get similar but not quite as good performance with Norgren valves and others which cost about 1/3 the Textron valve.

Brad is our in house sales person and covers the Northwest and does the Ethernet training and demos.
Bruce is our North East sales person and covers Ohio.

BTW, there is a RMC150 accessible on the Internet. I will PM the IP and URL of the RMC150 to those like CharlesM that are interested. Remember, not everyone can use the RMC150 at once.

 

[kamenges]

Originally posted by agarb:

Keith, I believe your "take it into a vacuum" comment was referring the mention I made to Peter about private discussion of my future application.

No, that wasn't my concern. I see you (agarb) as a little bit of an anomoly. You are actually looking at the process and seeing what you truly need to make it work. It actually takes a little work and knowledge on your part to do this. And knowledge is NEVER bad. Not many people do this. Quite often, the position take is 'It moves, must need a motion controller'. This is not necessarily true, as you have seen. My post was in response to Peter's post about demonstrating that a dedicated motion controller will outperform a general-purpose plc doing motion. This seems intuitively obvious to me. The 'vacuum' I was talking about was taking this single piece of information without looking at the context of the application. Do I really need a pneumatic nailer if I am only driving one nail? Some of the nailgun salesmen will say I do. I tend to disagree.

I think allscott's bus/sports car analogy is a bit off. The plc/motion controller question is more like comparing a Porsche to a Toyota Corrola. They will both work in the low to mid performance area. There are obviously some things a Porsche can do that a Corrola can't. And if I am only buying one I will probably buy a Porsche. I know it will do anything I need to do no matter what the performance required.
But now, what if I am buying a fleet of cars? It is fiscally irresponsible of me to buy the Porsches if the Corrola will do the job. My purchase isn't just a one-time thing. It is a 100-time or 1000-time thing. So even if it takes me a little more to make the lower performance item work it makes sense to do it.

There are some guys around, Peter being one of them, that can analyse and model a system accutarely enough to be able to tell you if a plc is good enough or a motion controller is required. For mere mortals such as myself, if I had the time I would try it with the plc first and add the motion controller if needed. If I didn't have the time I would go with the motion controller right away and see if a plc can do it later.

In general if you are doing multi-axis coordinated motion you are most likely better off with a motion controller, especially if your relative position tolerance is very tight and your motions are not slow. This is where the added tools of a dedicated motion controller really help.

Keith

 

[Peter Nachtwey]

Do I really need a pneumatic nailer if I am only driving one nail? Some of the nailgun salesmen will say I do. I tend to disagree.

I think allscott's bus/sports car analogy is a bit off. The plc/motion controller question is more like comparing a Porsche to a Toyota Corrola. They will both work in the low to mid performance area.

I think the nail gun analogy is off too. One nail can be driven quickly either way.

There are obviously some things a Porsche can do that a Corrola can't. And if I am only buying one I will probably buy a Porsche. I know it will do anything I need to do no matter what the performance required.

Again I think this analogy is off. Two people can start at the same place in a Corrola and a Porsche and legal get to the same place at the same time.

The motion controller saves time and is insurance because it is loaded with diagnostics that PLCs can't match. I had one customer call it his get out of jail free card because the motion controller allowed him to finish the project and go home.

But now, what if I am buying a fleet of cars? It is fiscally irresponsible of me to buy the Porsches if the Corrola will do the job. My purchase isn't just a one-time thing. It is a 100-time or 1000-time thing. So even if it takes me a little more to make the lower performance item work it makes sense to do it.

This brings up a good point. OEMs that use 100s or thousands don't user commercially available motion controllers. They bring it in house and make their own controller with just the features they need to bring the cost down.
We have hit this wall many times. We could make motion controllers for a lot less. We can just strip out the diagnostics and the setup software and just mount a 4x20 LCD on the controller. This would be fine for a OEM application where a simple motion controller is required for dump trucks or lifting wheel chair platforms. In these cases the motion controller would always use the same parameters and once one controller was setup one could copy the parameters for the next project. Extensive procedures can be written for debugging and trouble shooting because these controllers would be sold by the hundreds each year.

We would lose our shirts if we sold a controller like this to the average application. All profits would be gone on the first tech support call and there would be no graphs to provide a clue as to what is wrong. Tech support is hard enough even with the graphs and all the tools.

The reality is that we have no customers that sell 100s of controllers. The closest thing is some large sawmill OEMs. These OEMs are always pushing the performance envelope. They also want documentation, diagnostics, auto tuning and polish, that no PLC PID based motion system can match. Down time is expensive. Good diagnostics is the key to getting up and running. This can save the cost of the controller. These OEMs also like having access to someone that knows the math and physics of motion.

Most of our customer buy less than 10 units a year. It isn't wise to write your own PID motion any more than it makes sense to write your own HMI in VB. It is better to buy WonderWare or something like it. You can't justify the time and support. That doesn't mean it can't be done, but it is kind of like re-inventing a wheel and not a very round one. Even if you can do it there is the PLC law about creeping features and usually these causes the code to be too complicated.

There are some guys around, Peter being one of them, that can analyse and model a system accurately enough to be able to tell you if a plc is good enough or a motion controller is required.

I don't count. There is only one of me and my time is best spent doing other things than installing motion controllers. One of them is making motion control easy for others to install and maintain. It has to be so easy that our salesmen, student business major ( cave man ) and distributors can do it. If a salesman can do it, an engineer, good PLC programmer or one of our application/tech support people engineers should be able to do it too.

I hate to say "motion control for dummies". I would prefer to say that we require as little mental energy as possible so there is plenty left for the rest of the application. Some applications out there are very sophisticated.