move high inertia sys. with hydraulics thru prop. valve

N

Thread Starter

niran

In my company we use PLCs for machine automation
A task in the machine operation requires the high inertia sys. to be moved with hydraulics thru prop. valve controlled by PLC.

The PLC scan time is about 12ms which causes the
system to override the final position by 10 mm. Is there any solution which will reduce this override up to 5mm?? we have tried to reduce the scan time but met moderate success.

Pls help me, the issue is pretty urgent.

regards,
niran
 
Have you tried reducing the command signal to the proportional valve as the load approaches the target position?
 
P

Paul Dackermann

Is there a possibility fo establishing a "false" target 10mm offset from true? Is the 10mm overrun repeatable?

just a thought.
Regards,
Paul
 
Dear Niran
You need to add a linear encoder to the system. Then you scan the encoder for the position (counting pulses) and you could stop right on target, compensating for any inertia.

good luck
Alberto
 
M
Hello,

You have lots of choices, but they are dependent on your current PLC hardware type:

1. If the overshoot is consistent (always 10mm +- 0.1 mm), then you can simply reduce the setpoint by 10 mm. If the machine's mechanical
characteristics dictate that friction won't be constant then you can't do it this way.

2. Many PLC's have higher speed "counter" cards with dedicated outputs that can respond within milliseconds or microseconds. Keep in mind that the system response time will also include the response time of your actuator.

3. Many PLC's have programmable interrupts, which will cause certain sections of code to be executed immediately based on the status of an input. Along with these, immediate outputs can turn on outputs immediately rather than waiting for the end of a program scan.

4. In some cases, a very inexpensive hardwired solid-state relay can detect an input and set an output almost immediately (microseconds).

I would have to say that the "best" way would be item 2 or 3, but the cheapest and fastest way may be item 1, or 3.

If you let us know what type of PLC you have, I'm certain others can give you an exact method to solve your problem.

Sincerely,

Mark Wells
President
Runfactory Systems Inc.
http://www.runfactory.com
1235 Bay Street, Suite 400
Toronto, Ontario, Canada M5R 3K4
Ph. 416-934-5038
Fax 416-352-5206
 
R

Rafael Jacomino

Perhaps a cheaper and simpler solution would be a second order logarithmic ramp or maybe even just linear negative ramp. I used the later with RexRoth VT amplifier cards and HYD-Prop/VLVs, combined a Modicon Quantum PLC. This PLC has these two math functions built-in so it was rather easy, but if you can build your y = mx + b then you can do it home-gown (you also got to build m but depending on your PLC, it could be fun!). Providing an +/-10V (either polarity addressing A or B solenoid) analog output controlled by both "end" and "before end" proxies switches, we moved a 10 ton molding machine for 12 meters in 3.25 seconds. With no bumps and no jerks! This solution resembles Paul Dackermann suggestion and like he says, just a though...
 
M

Michael Griffin

I'm not sure if I understand your machine properly, but is it possible to use the proportional valve to control the speed of the actuator? If so would it be possible to slow the actuator down prior to reaching the "stop" position? You did not mention how the machine knows where to stop the actuator (linear transducer?, proximity sensors?).
If the problem with reaching the correct stop position is due to the response time of the control system, then slowing down the actuator at a "slow down" position and then "creeping" to the final position should reduce the amount of error.

If you are using proximity sensors to determine your stop position, you may need to add an extra set of sensors to detect the "slow down" position. If you are using a linear transducer, then this of course is not a problem. However, If you are indeed using a linear transducer, then I assume that you have already considered using some form of closed loop control algorithm and found it wanting for some reason.

--

************************
Michael Griffin
London, Ont. Canada
************************
 
D
Niran,

You did not mention if the valve is a cartridge or directional valve.
Scan time may or may not be the problem.

Items to consider.
What is the required flow for max speed and what is the flow rate of the valve @145 PSI (10BAR) drop?
Does the valve have spool position feedback?

I am always amazed how often proportional valves are improperly sized. If the valve is oversized ( and many times they are), you may be able to change the valve to a smaller size and get much better control.
The sizing error comes in when the designer assumes that pressure drop is a bad thing.
Remember that .... flow control=pressure drop...... accross the valve.

There is also a large difference between response time of valves. A DO5 valve will respond faster than a DO8, for example. You will want to make sure the valve has spool position feedback as part of the valve electronics. This is an option with some manufacturers.

Let me know if I can be of assistance

David Kane
[email protected]
Kane Engineering Group Inc.
Motion Control Consultant
 
hi all,
first let me Thank u all for the valuable responses u have given. hereby i am giivng the current scheme we r using.
we are using Mitsubishi FX2N Plcs,The motion is sensed by linear transducer. and the motion is carried in two stages fast and then slow .we r using slow speed in the last stage to control the override,which is the matter of concern.
regards
niran
 
S

ScienceOfficer

niran---

I don't think PLC scan time is your main problem. Most proportional valves (in my experience) have a step response time slower than 50ms, and even high-performance units only get down to the vicinity of your scan time in expensive and highly tweaked situations.

You specifically mention using a proportional valve, not a servo valve. It is possible but not certain that you can't actually do better than you are doing with the valve you have.

This list can help you better if you tell us the PLC model, the general programming strategy (PID?), the make and model of the proportional valve, and what feedback you are using, if any.

Hope this helps!

Larry Lawver
Rexel / Central Florida
 
M

Michael Griffin

1) How slow is your "slow" speed? For example, if "slow" was 100 mm/sec, then with a 100 msec system delay (including transducer, PLC, and valve) you would see (very roughly) 10mm of overshoot.
2) What delays are introduced by the linear transducer, PLC scan rate, and valve response time?
3) Is the transition from "fast" to "stop" long enough to allow the "slow"
speed to have any effect? If the position for transition to "slow" is too close to the stop position, there may not be enough time for the system to react and the "slow" speed never has a chance to take effect. The system has to be able to go from "fast" to "slow" within the time it takes the machine to travel from the "slow" position to the "stop" position.
4) Furthermore, is there any elasticity in the hydraulic system (hoses, accumulators, etc.)?

************************
Michael Griffin
London, Ont. Canada
************************
 
Larry,

I beg to differ, I know of a variety of off-the-shelf servo hydraulic systems with positioning bandwidth of 30 - 40 Hz running a 885usec servo update rate. If you back the servo loop update rate down to 3 - 4 msec, you will have reduced positioning bandwidth of 3 - 5 Hz. The non-deterministic nature of PLC scan times will also force you to tune the system based on the slowest update rate resulting in poor performance at best. Additionally,
PLCs are ill suited for managing the difference in effective gain in a hydraulic system when working on the rod area of the cylinder vs the cap end . . . delivering different forces for the same absolute command out of the
valve. It would be far better to add a motion controller and connect to the PLC via fieldbus or PLCbus and let the motion controller manage the motion app and the PLC manage the state logic app.

Ken Brown
 
S

ScienceOfficer

Ken---

I agree with everything you say, but the given problem involves proportional valves, not servovalves. The best performance I have ever
gotten from proportional valves corresponded with a 20Hz bandwidth, whereas I have gotten 50Hz response from some excellent servovalves.

In my response to niran, I was trying to reduce expectations. Until we know the specs of the proportional valve, though, it will be difficult to speculate about the ultimate positioning performance available.

Meanwhile, your points about implementation are well taken.

Hope this helps!

Larry Lawver
Rexel / Central Florida
 
Larry,

Fair enough . . . I missed the point about it involving a proportional valve. Thanks for the clarification - I'll be more thorough in my reading next time.

Ken
 
J
I think thet you should improve the PID tunning. Put the higest scan that you can (ej. 10 msec) and then tune the PID. The overshoot in 10 mm that you actually has can be reduce, decresing the Proportional gain. In that way, thee system will take more time to reach the final position, but you will reduce the position overshoot.
 
Top