Motion Axis Jog on 1756-M02AS

Hi everyone, I'm looking to modify some motion controls. There's a hoist that picks up to 8 railway ties, weighing up to 300 lbs each, which then travels to an unloading area and releases them. The current controls uses automatic and manual modes. There is a significant (10% if I recall correctly) dead spot in the valve used to control position, which leads to some correction (back and forth travel) of a few inches at each end of travel. With 2000+ lbs of ties, plus the steel hoist, there's a lot of inertia! The valve ends up oscillating a bit each time, but it works okay in automatic mode. When using manual mode, sometimes the hoist will oscillate forever when the jog buttons are released (forward or back). My new controls will hopefully prevent this by just stopping when the jog buttons are released. The current controls use the jog buttons as Motion Axis Move (towards home or unload positions) instead of Jog, and releasing the buttons mid-travel causes this interminable oscillation. Pressing the jog buttons again will usually cause the hoist to stop oscillating, but it's fairly frustrating for the operators and I think it's quite damaging to the machinery.

I attached the logic for the controls I wish to implement. Currently no movement occurs when the jog buttons are pressed, though I did change the MSO.EN to MSO.DN, after reading a relevant thread here, which I haven't had a chance to test yet, but I don't think that's the problem. Unless the .EN in combination with the ONS is causing the MSO to be enabled only for the current scan?

Yes, I have a request in for a better valve.

Did you come to a solution with this?
We exclusively use Delta controllers for our hydraulic motions as the dedicated controller seems to respond better and the tuning functions are awesome. The interface to the PLC performs great for our point to point moves (250lbs @ 400mm/s with no jitter at move complete with ramped accel/decel curves)
I found that using all of my jog commands with a 10ms debounce timer to prevent accidental triggering as well as a MAS stop command on loss of jog instruction as the jogging instruction will continue even if the rung turns false. Another thing I like to use with the same debounce timer is after holding the jog button for X seconds, retrigger the jog command with a faster speed preset to get "inching" and "high speed" modes which all of the operators like for precice & long distance movements. I tend to prefer axis.DriveEnableStatus instead of the instruction MSO.DN bit as it references the actual axis and not the logic that could be altered in scanning locations of the program.

Robb B said:

Hi everyone, I'm looking to modify some motion controls. There's a hoist that picks up to 8 railway ties, weighing up to 300 lbs each, which then travels to an unloading area and releases them. The current controls uses automatic and manual modes. There is a significant (10% if I recall correctly) dead spot in the valve used to control position, which leads to some correction (back and forth travel) of a few inches at each end of travel.

Click to expand...

The best thing to do is to replace the valve with a valve that has 0 or extremely small dead band. The dead band makes the system A 10% dead band is a killer if you want precise positioning at the set point. If the problem was ONLY the dead band then replacing JUST THE VALVE will do. The M02AS should work well enough but.........

With 2000+ lbs of ties, plus the steel hoist, there's a lot of inertia! The valve ends up oscillating a bit each time, but it works okay in automatic mode.

Click to expand...

Yes, so what is the natural frequency of the system? I bet no one knows. The natural frequency is important because it limits the frequency of acceleration and deceleration. Jogging is often done with a lot of short accelerations and decelerations that often are done at a frequency higher than what the system can achieve.

When using manual mode, sometimes the hoist will oscillate forever when the jog buttons are released (forward or back).

Click to expand...

This is not good. If the system oscillates in open loop or manual mode then the problem is NOT THE CONTROLLER. Is it a new system? I would look at mechanical linkages too.

My new controls will hopefully prevent this by just stopping when the jog buttons are released. The current controls use the jog buttons as Motion Axis Move (towards home or unload positions) instead of Jog, and releasing the buttons mid-travel causes this interminable oscillation.

Click to expand...

It would be good to see a trend of the target position, actual position, and control output at the highest frequency achievable.

Pressing the jog buttons again will usually cause the hoist to stop oscillating, but it's fairly frustrating for the operators and I think it's quite damaging to the machinery.

Click to expand...

This is odd. Are you sure the control output isn't changing when you think the controller is in manual mode. Something must be causing the sustained oscillations if it isn't the controller's output to the valve.

Yes, I have a request in for a better valve.

Click to expand...

Good, but lets see a trend. I can't see why the valve would shift back and forth over 10% if the system is in manual.


jholm90 likes the Delta RMCs but what is even more important is Delta tech support. Especially while planning the system before the hydraulic/mechanical designers screw up the system and leave the automation guy ( you ) to "fix it in software".

Hi Peter, sorry it took so long to respond. I've attached a trend of the hoist movement in automatic. I will try to get one of it oscillating in manual mode. I definitely agree with the deadband issue, and I will try to get a better valve. Managers can be hard to convince though.

What are the PID settings?
Hydraulic systems, especially with spool overlap don't get along with the integrator.


Craig

Tinine said:

What are the PID settings?
Hydraulic systems, especially with spool overlap don't get along with the integrator.


Craig

Click to expand...

There's no PID. Spool overlap is the same as hysteresis (dead band)? In my case, the valve has a 10% dead band at center.

Not wanting to contradict anyone here but I consider the overlap/deadband to be more of a feature in this particular application. Closed ports?

If there is no PID then it's not trying to eliminate error.

I am familiar with the the type of proportional valve with spool feedback to the driver and there is a deadband adjustment pot. Sounds like this might be set too tight(?)

Craig

Robb B said:

There's no PID. Spool overlap is the same as hysteresis (dead band)? In my case, the valve has a 10% dead band at center.

Click to expand...

I am confused? No PID, No feed forwards. Your motion looks pretty good for not PID or FF. The control output does not look smooth. Either you have a very high derivative gain or your feedback has very low resolution. Also, the trend cuts off the position trends so it is hard to see what is happening when in position I wish EVERYONE WOULD DO WHAT Robb did. He sent a trend.

The M02AS has a deadband eliminator. Did you try that? Basically the control output will jump to 11% if it needs a 1% output and the spool overlap is 10%.
The problem is that it still takes time for the spool to shift from -10% to +10% just to correct for a +/1 0.1mm error. There may be a band where the integrator is turned off when close to the command position. The RMCs have that so the integrator doesn't keep the actuator oscillating back and for a tiny amount. The downside is that once actuator is in that band, only the P gain is active. Usually this band is smaller than the in position band so at least the controller get to the in position band.

It has been over 20 years now that I wrote that code or played with a M02AS.

Tinine said:

Not wanting to contradict anyone here but I consider the overlap/deadband to be more of a feature in this particular application. Closed ports?

Click to expand...

Not on a true servo system. Valve dead band is non-linear.

If there is no PID then it's not trying to eliminate error.

Click to expand...

I am confused too. It is tracking well with no PID. Maybe Robb is using feed forward only but then the control output wouldn't have all the spikes.

I am familiar with the the type of proportional valve with spool feedback to the driver and there is a deadband adjustment pot. Sounds like this might be set too tight(?)

Click to expand...

Yes, some valves have it and some don't. I added this feature in ours and Rockwell's controller. The problem is that no matter what, it still takes time for the valve spool to shift from -10% to +10% when around the commanded position. Also, feed forwards are not as effective unless they take into account the dead band. Some valves have dead bands of 40%! and customers try to make them work. The valves with large dead bands should only be used in application that are always metering in on the same port of the valve. This avoids moving the spool large distances across the dead band.

Agreed but it sounds like there is no commanded position. I think you already nailed it with move acc/dec vs jog acc/dec.

Don't really want a docile jog, right?

Robb B said:

There's no PID. Spool overlap is the same as hysteresis (dead band)? In my case, the valve has a 10% dead band at center.

Click to expand...

I should clarify: There's no PID in the logic. There may well be one internally to the motion controller or motion commands, but nothing in the ladder logic.

Peter Nachtwey said:

I am confused? No PID, No feed forwards. Your motion looks pretty good for not PID or FF. The control output does not look smooth. Either you have a very high derivative gain or your feedback has very low resolution. Also, the trend cuts off the position trends so it is hard to see what is happening when in position I wish EVERYONE WOULD DO WHAT Robb did. He sent a trend.

The M02AS has a deadband eliminator. Did you try that? Basically the control output will jump to 11% if it needs a 1% output and the spool overlap is 10%.
The problem is that it still takes time for the spool to shift from -10% to +10% just to correct for a +/1 0.1mm error. There may be a band where the integrator is turned off when close to the command position. The RMCs have that so the integrator doesn't keep the actuator oscillating back and for a tiny amount. The downside is that once actuator is in that band, only the P gain is active. Usually this band is smaller than the in position band so at least the controller get to the in position band.

It has been over 20 years now that I wrote that code or played with a M02AS.

Click to expand...

Found the PID part internally.

I looked through the configuration pages but I don't see deadband anywhere, unless it's called something else. Back to reading I guess. The position trends aren't cut off, but they reach the min/max of the scale. I can do each on separate charts if you'd like clarity. I find it easier to see what everything is doing at a given moment if they're all on the same chart.

Ah-ha...that be the culprit. Something set too aggressive.

Maybe some pics or I.D. of the motion controller?

Craig

Disregard....I didn't see the above on my phone

I would start by reducing the proportional

Not sure you even need differential.