PLC scan rate and hydraulic pressure

[mbuis]

I had posted an earlier thread about a scale w/parameters issue. Turns out it may not be the issue. Since, I do not have much PLC experience, nor do any of my colleagues, we contracted a local expert. He was extremely helpful and very knowledgeable.

However, he ran into the same problem we were originally having. He did simplify what we had and made some improvements, though.

Here is the layout:
We are trying to cycle a load from a low load (1000 lbs, for instance) to a higher load (15,000 lbs, for example.) We need to stay within plus/minus 2% of these upper and lower values in order for the test to be valid. We have a load cell which has a load rating of 0 to 50,000 lbs. It has a voltage rating of plus/minus 5 volts. Positive volts meaning it is pushing from 0 to 50,000 lbs and negative pulling. We have a MicroLogix 1100 which will read plus/minus 10 volts. Range from -32603 to 32603.

Here is the problem:
We would vary the input for the higher load trying "to dial it in" and we would reach a point where all of a sudden we are well over (or even under) the 2% variance we are allowed. The load was not stable. If we slowed the pump down, it worked great. But, this is way too slow for this to be a practical application. The guy even speed up the scan rate to the maximum rate (by selecting unfiltered) and it worked better, but still had too much variance.

He was convinced that since the hydraulic pressure is nearly instantaneous, the PLC was not able to keep up with the pressure changes. The original controller had a scan rate of 8 to 10 times faster than the PLC is capable of. Unfortunately, this controller is outdated and cannot be replaced.

After, seeing the responses we were getting today I believe the guy is correct. However, I was wondering if anyone had any thoughts on this?

I even found this article which seems to support this hypothesis:
http://www.controldesign.com/articles/2008/105.html

 

[TConnolly]

How are you ramping up the force? With the additional info you just provided my knee jerk reaction is that a PLC is way too slow for what you are describing.

I suggest you talk to the guys at Delta. Peter, the company president, is a regular here.

 

[James Mcquade]

Here's what i would do.

Program your output to be turned on by bit x. Bit x in turn enables the analog output with minimal output - say 10,000 lbs force. Bit x also runs a timer - 50 ms that resets itself. when the timer runs out, increase the analog output number by 500 counts. eventually, you will get your 15,000 lbs and you may even be able to get to within 1% of 15,000.

i did this on a system and it worked great.

the first thing in the morning, i had a real low number.
i then built up to the pressure i wanted say 12,500 lbs and stored that integer number. as each cycle progresses, the oil heats up and flows faster, so you will have to change the number every cycle for the 12,500 lb minimum. if you test your your logic, you may be able to get the 15,000 +/- 2% first thing.

just a few things that i did on a similar system.

regards,
james

 

[CharlesM]

What kind of scan times are you getting? Most hyd applications require 2ms or less.

What is your ramp times?

 

[mbuis]

Thank you all for the replies. I had to dig around to answer some of the questions. We have an 1762-IF4 Analog Input Module. I think the scan rate is 130 ms. This is an unfiltered signal. We are not using a "ramping" function. At least not how I believe people are asking about. What happens is the PLC turns on the hydraulic pump and the cylinder starts to apply pressure. We handle the maximum and minimum loads with math functions. As in as long as the load is less than the maximum load, keep the pump on. When the load is equal to or greater than the maximum load turn the pump off. Then when the load is less than or equal to the minimum load turn the pump back on. Repeat.

James, your reply is interesting, but I am wondering about cycle times. This test can run for 150,000 cycles (or more). The faster we can get through the test, the more tests we can run. However, your suggestion seems worthy of exploration. I just wish I had more PLC experience. Guess over the next few months I will get my wish though. LOL.

 

[mbuis]

Alaric, I am new to the company that currently employs me. But, I believe Delta has a quote in here to replace an outdated Rexroth set up which is in our lab. And the gentleman that we brought in to help us out, had high regards for Delta. So, I may become familiar with Delta inthe coming year.

 

[CharlesM]

I think you are heading in the right direction. If the Bosch control has been working well in the past then we can assume the hydraulics are good and the load cell signal conditioner is fast. I think your PLC is way to slow. A Delta RMC controller should do what you need. If you have not already you should check out the videos from thier web site. They have a testing machine similar to what you are talking about.

http://www.deltamotion.com/applications/videos/appvideos.php

 

[RocketTester]

mbuis,

If I read correctly, it sounds like you are controlling the cylinder force by just turning on and off the pump. That is unique. Most people control cylinder pressure by using a servo-valve to proportionally control the hydraulic pressure applied to both sides of the piston. They are similar to the spool valve on top of the power steering gear on older Pitman arm cars, except the spool is moved by an electromagnet coil.

The controller outputs a current to the coil, typically +/-20 mA, and usually there are 2 coils you can wire in series or parallel. Look on Moog's website for details. The simplest type is a "servo-valve", which basically converts input mA to cylinder speed (as I understand). That probably requires feedback control on the load cell force. Next higher is a "proportional servo-valve" which basically converts input mA to cylinder position. You might be able to use that in "open loop" mode, where your controller outputs mA and you get a proportional (maybe non-linear) cylinder force, but fairly predictable. If so, you might operate as James suggests, by outputing a pre-determined mA profile, with a slow "trim" feedback. Top line are the "digital servos" with a motor positioning the servo spool. They are insensitive to changes in the hydraulics, especially flow-rate.

Re controller, in your current "pump on/off" control mode, the 130 ms cycle time might not be a problem since the motor probably has much more lag than that. However, typical hydraulic controls require <2 ms cycle time. We use Delta Computers RMC75E and RMC100E and Beckhoff Automation EtherCAT. Delta's are dedicated PID controllers (0.5 ms response, I recall) and Beckhoff is a general purpose PLC (0.2 ms response). Another fast PLC is B&R Automation (2 ms response, I recall). Another fast PID box is the Gefram 2301 used in plastic extruders we got years ago. More choices today. Parker has a PID box. One of the least expensive is Galil's Rio. A-B Contrologix is usually 10 ms min response. Most PLC's don't output +/-20 mA, so you need a "V-to-mA converter" - Delta VC2100, VC2124, Axiomatic IC-DR-16. A PID box makes most sense for a few control loops, a PLC for many control loops. Delta and Galil are in-between a PID box and PLC, allowing some configuration and/or programming but not standard PLC code. For you, I think adding an external PID controller would solve all your problems. Your PLC would then output a +/-10 V "force setpoint" signal to the controller.

 

[Peter Nachtwey]

The RMC75E will interface to your existing AB Ethernet. What the RMC can do that the others can't is record the test position, velocity, and forces and do calculations like peak detect at high rates. This way you KNOW if your test is valid. The PLC can then upload the data at its rate. The RMC75E has 64 MB of ram so it can save a lot of data. A PID box doesn't do much good if the PLC can't read or record the test. How do you verify that the test was valid?

Better yet is to connect the RMC75E directly to a HMI or PC. If you use a PC our setup software may be all you need. We also have interfacing software that allows one to up load and down load code using excel or VB or C# and other languages. If you want to spend the big bucks you can buy Lab View. We have a Lab View certified VI that provide an easy way to communicate between Lab View and the RMC. The RMC can do all the fast sequencing. No PLC is required unless there is a need for a lot of I/O.

Pumps generally respond too slowly. We do have a few customers that control pressure using servo drives or VFDs to drive a pump and directly control pressure/force. That works but the motor must be able to supply a lot of torque while holding force and that requires a lot of current.

The most common way to do this is to use a pump, accumulator and servo quality valves. Many valves have amplifiers on board so they will have +/- 10 volt inputs and convert that to the current needed to drive the solenoids that move the spool. The key is to get a valve with a linear and zero over lap spool.

Load cell amplifiers are often a source of problems. Too many are too slow. You should get one with bandwidth of at least 4 KHz.

 

[shooter]

as you have many cycles do not use any PLC related to control the force or pressure.
make one buffer low pressure and one buffer high pressure and switch between both with a simple valve.
just correct the pressures to your readings the first few cycles and run the test. Just have the pressures in the buffertanks monitored and controlled. for the high you will need a pump and for the low one an pressurecontroller.
the swap can be done with a simple servo valve running from one side to the other in a controlled speed. So only one moving object (and the pump).
no need to be fast cycled.
however it is possible with some servovalves not direct with pumps as they have long time to slow down and speed up, and the capacity is a big cylinder in the pump. so no way to get it faster.

another way is to use a mechanical system with some pressure valves.

 

[TConnolly]

I'm going to take all the fun out of building your own, but might I suggest a ready made solution whether new or used. One of these can run through a 150,000 cycle test in just a few hours.

 

[Peter Nachtwey]

I'm going to take all the fun out of building your own, but might I suggest a ready made solution whether new or used. One of these can run through a 150,000 cycle test in just a few hours.

Yes, they are our competition in the testing market but they can provide the complete package at a price. We only supply the controller. If you buy a MTS system then you know the hydraulics will be done right. We have seen too many hydraulic designers under estimate how difficult it is to do a sinusoidal test system after the frequency goes beyond 10 Hz. The peak velocity goes up with the frequency and the peak acceleration goes up with the frequency squared! The diameter of the cylinder goes up with the frequency if the cylinder going to be able to accelerate and decelerate the mass. The area is proportional to the diameter squared so the flow required goes up proportional to the frequency squared. The costs go up dramatically as the frequency goes up.

 

[KalleOlsen]

Yes, they are our competition in the testing market but they can provide the complete package at a price. We only supply the controller. If you buy a MTS system then you know the hydraulics will be done right. We have seen too many hydraulic designers under estimate how difficult it is to do a sinusoidal test system after the frequency goes beyond 10 Hz. The peak velocity goes up with the frequency and the peak acceleration goes up with the frequency squared! The diameter of the cylinder goes up with the frequency if the cylinder going to be able to accelerate and decelerate the mass. The area is proportional to the diameter squared so the flow required goes up proportional to the frequency squared. The costs go up dramatically as the frequency goes up.

Just a question from an ignorant:
It sounds like you need a lot of power to do this kind of exersise. Is it possible to build some kind of "oscillatng" test rig for the purpose - to save power? A hydraulic equivalent of a spring and a weight.

Kalle

 

[Peter Nachtwey]

Just a question from an ignorant:
It sounds like you need a lot of power to do this kind of exersise. Is it possible to build some kind of "oscillatng" test rig for the purpose - to save power? A hydraulic equivalent of a spring and a weight.

Kalle

This is simple physics and it makes no difference if it is a servo motor or a hydraulic system

Power = Force * velocity
velocity is proportional to ω
Acceleration is proportional to ω²
Force is proportional to mass x acceleration or mass x ω²
so power is proportional to mass x ω² x ω or mass*ω³

You can see that increasing the frequency, ω, increases the power quickly.
Up until know I have said proportional because I have left out the amplitude.
The peak velocity is Amp x ω
The peak acceleration is Amp x ω²
so power = mass x Amp x ω² x Amp x ω

If limited by power then the amplitude can be reduced. However, as ω goes up so must the natural frequency of the system to be able to accelerate and decelerate at the higher rates. There are a lot of places where designers go wrong.

The wise ones either know what they are doing or know these systems and are tricky so the call me. The other option is to pay the big bucks and get a MTS system that takes all these factors into account. The ignorant ones don't know there is a problem until it is too late. These mistakes are very expensive as they will not work as desired. I have seen cases where manifold, valves, accumulators and cylinders all have to be replaced.

One more thing. A PID will work on some hydraulics systems where the natural frequency and damping factor are relatively high. When either the natural frequency or damping factor are low then one must use the second derivative gain. Most controllers don't have a second derivative gain. The only other controllers I have seen, other than our RMC, with a second derivative gain are the old MTS TDC 200 controllers. I must give them credit because they had that figured out long before I did. However, the TDC 200 didn't really have the CPU power or feedback resolution back then to implement the second derivative gain correctly.

I'm going to take all the fun out of building your own

I question the 'fun' part. If mbuis didn't have a few of us telling that what he is doing won't work he may have banged his head against his problem for months. I have seen it happen.

Just to make a point. mbuis, how fast are the analog inputs on the PLC? I bet they don't have a 4 Khz response. What about the analog output? I bet the analog output doesn't have a very fast response either.

 

[KalleOlsen]

Peter, thanks for the maths repetition lesson!

But I am still an ignorant and wonder where one find the power losses. Power out = power in. Does it mostly go to heat the work piece and oil?

(My hydraulics experiences stops with my '94 New Holland tractor with back lift arms and front loader.)

Kalle