Position Control Fun

I work in an auto parts plant that has an electroplating line. This plating line has seven hoists that lift racks of parts out of tanks, move along rails, and lower the racks into the tanks. The racks are supported by copper bars that rest in v-blocks on the tanks. The twelve year old controls use a PLC5, VFDs, proximity switches on the hoists and a row of flags mounted along the rails. There are four proximity switches on each hoist – forward decelerate, at position, reverse decelerate and a home prox that acts like a zero reference for the position counter program. These proximity switches and flags are a source of significant downtime and several of our sister plants have replaced the hoist position hardware with laser rangefinders. I have project approval to upgrade to ControlLogix and purchase laser range finders.

My limited experience with motion or position control has involved projects in which the PLC sent basic commands to servo controllers that then took care of everything else. I don’t see an opportunity here to incorporate rotary encoders or servo motors because there is no hard mechanical connection between the drive components and the rail position. The hoists ride on metal wheels on a metal rail where some slippage is inherent.

Therefore it looks like I have to program a position control system that uses a linear displacement measurement and standard VFDs. The good news is that speed is a minor issue and the required position accuracy is around +/- 0.25”, pretty loose. However, I have observed the plating hoists in our other plants and I was not impressed. Lots of PID loops and they don’t work very well. They tend to overshoot or undershoot and then spend a couple of seconds hunting for the target position.

I am hoping to get some general guidance here. Any suggestions?

the Laser positioning will function well. with the DC braking function of most VFD's the ability to stop the hoist accurately is fairly easy.

PID loops should not be needed - you will need to program a slow down area.

Using Laser / Analogue Input it is scalable to the accuracy you need.
simply compare for position points.

You may want to talk to Rockwell again. They have a specific logirithm for just this aplication that runs in the Control Logix platform. They will charge you for this but it will work and make things simpler. If you would prefer a turnkey system please PM me, I know some guys that build exactly what your talking about all the time and it is fast and does not cause cable sway.

What your guys at the other plants are probably forgetting is to look at the drive. (think torque feedback)You need to know what kind of load your putting on the motors as you travel so you can figure out your stopping curve and accel curves. Just using distants and rate is not enough.

If you really want to roll your own.. instead of looking at PID, google "forward feed" this will give you an idea of what you need to do.

What Ian said is true as long as your load is constant and you are not overramping your motor. But cable sawy can become an issue real fast.

KerryP said:

I am hoping to get some general guidance here. Any suggestions?

Click to expand...

Now many VFDs? If there are two or less then this would do nicely.
http://www.deltamotion.com/products/motion/rmc70/
Use a Balluff or Temposonic MDT rod for linear position feed back. Hopefully your PLC5 has Ethernet because that would make communications easy using MSG blocks. A serial port could be used but the PLC5 doesn't have a lot of serial ports and they are slow.

I would keep everything as much the same as possble as far as the existing logic. All you really need to do is "Emulate" the original limit switch signals with "virtual signals" you can create with some comparators based off your lasers. If it was the hardware that was the problem before and not the program, this makes it a pretty quick and simple job.

I have to agree with the gentelman from new york in this case. unless you just want to do the extra work. but i think a micro logix would be enough plc for this unless you just want to spend the extra $$$$

The problem with having discrete signals for accel/decel is that everything works OK if the hoist/crane hits them at full speed. If for some reason the hoist stops within the "slow down" zone, the "slow" signal may not be enough to get it going. IMHO it's better to have an "analog" switch that scales the speed command based on distance to the end of travel point.

ttichy said:

The problem with having discrete signals for accel/decel is that everything works OK if the hoist/crane hits them at full speed. If for some reason the hoist stops within the "slow down" zone, the "slow" signal may not be enough to get it going. IMHO it's better to have an "analog" switch that scales the speed command based on distance to the end of travel point.

Click to expand...

Depends on how the switches are flagged. If the sensors are made such that they stay on for the entire accel and decel range, rather than just simply being triggers, then you avoid the avoid the issue you mention.

What kind of accuracy and repeatability do you need? What is the environment like where the sensor will be installed? What is the length of the span that will be required of your laser distance sensors? Do they interface to the C-Logix digitally?

When I did controls with multiple flags and switches, I still "modelled" position of the final driven element with simple math in the PLC based on motor speed feedback and error checked it in the PLC. You can figure out position windows even for sloppy control to fault and stop when the amount of slack in the system increases too much and you can avoid "losing" the position of the hoist under 99% of circumstances. I have a sorting system using this principle with no feedback, using only commanded motor speed scaled into measurements of "expected" feet per minute...built on simple math and fifteen minutes with a hand tach...poor man's encoderless tracking with the best available data...

You can also make it easy to program faults when a sensor triggers way outside of the modeled position. Invert your LIM blocks with A/B PLC code and latch a bit to alarm and continue or fault when there is noise, a flaky sensor, or other bogus input. You can also use flag bits in the PLC to allow the position to be corrected by any of the intermediate position switches (if they're unique inputs...strongly suggested), so you never have to home the thing unless you power cycle it and move it while PLC is off or unable to measure the expected delta of position.

I did this on a simple ballscrew device that previously had to be homed twice a shift. They were close to the limit on scanning a prox on a toothed wheel with a fixed speed motor so it would get "off" by about an inch after a five or six changeovers when it moved to a new position which was about every 6 hours.

I did this on another one with real good drive speed feedback and it would ramp right to the center of each switch on command...very sweet...the setup techs loved it.

I count 7 hoists implying 14 drives

Have you looked at the Powerflex 755 ?
They look like they can close the loop on anything.

I think 7 vfds and 7 fwd/rev motor starters.

Thanks for the feedback. I am the only controls engineer in this plant and some days I feel like no one has any idea what I am talking about. I really appreciate this community.

iant - I failed to mention the travel motors also have mechanical disc brakes.

ClayB - Thanks. I will be connected to the drives with either RIO or EthernetIP and will be able to pull the torque value out of the drive. Excellent idea. There are no swinging cables, the lift mechanism has 4" square tube vertical elements and a roller trolley that rides the vertical tubes.

DamianInRochester - Thanks for confirming my initial approach. At first, to prove the rangefinders, I want to keep the code basically as it is and use the analog position data to generate "virtual flags".

bce123 - The ControlLogix will eventually run the entire plating operation, hoists, rectifiers, pumps, fans, filters, etc. The PLC5 is jam packed and running out of memory.

ttichy - The exiisting flags are merely triggers of about 2" width. The decel proxes crossing the flags tell us it is time to slow down.

OkiePC - Our initial choice is TR Electronic's LE200. It is accurate to 3mm over 125m. Our longest single hoist travel is about 90 feet. My initial, conservative design is for +/- about 0.25" and I am pretty sure +/- 0.5" will perform well. With Ethernet IP, I will be getting numeric position data at speeds beyond those required. I will be able to place these lasers away from the process nastiness, unlike the flags and proximity switches. Thanks for the second post also.

MichaelG - Exactly. However, for a lot of reasons, we will continue to use proximity swtiches for the up/down motion.

bce123 again - Why starters? I am getting used to using drives without starters, just safety relays.

Thanks again everyone.

Kerryp, thanks for replying to everyone.
by stating:
bce123 again - Why starters? I am getting used to using drives without starters, just safety relays.
I was assuming starters for the lifts and vfd for the moves. but drives work too
thanks, steve