Need Temposonic Rod Scaling Help 4

[patmando]

Have a 130" temposonic rod to keep track of press position.
Using Siemens PLC 555. Rod is an MTS Flexible Sensor, 0-20ma @ +24vdc. MTS 201554 magnet. Not too sure how the scaling functions work. Need some advise please.
thanks

 

[ScottyUK]

Based on brief and frustrating experience of a Temposonics sensor, I would have to say chuck it in the skip and buy a Balluff equivalent. I don't work for Balluff. I am a big fan of their Micropulse transducers where I need precision and reliability. I think they are physically interchangeable with the Temposonics unit.

Do you have a part number for the one you are having problems with?


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[PNachtwey]

If the rod sends a 0-20ma signal for 130' then isn't the scaling done in the PLC?
You need to provide the A to D counts and position at two extremes of the rod.
Then it is a simple matter of calculating m and b for y=mx+b. Isn't there a scaling block in the TI PLC?

BTW, I deal with Balluff and Temposonic rods all the time. They are roughly the same. I also get involved with a lot of presses. PLCs work some times but the motion control is crude. It looks like this press can open wide which means that it will want to move quickly to close. This is another trick. The velocity at which one makes contact with the work piece is critical. One want to match the kinetic energy of the press with the work required to form the piece as closely as possible. Too much kinetic energy results in pressure spikes and too little requires the press to build up pressure slowly ( slow production ) to form the piece and even then the piece may not be formed right. Since the kinetic energy is proportional to the velocity squared one can see the good speed control is critical. The point is that it is difficult to get could speed control with the resolution possible from an A to D card or the asynchronous nature of the MDT rod updates and the PLC scan. One should really get a controller that can directly interface to a MDT rod either using start/stop or an SSI feedback.

 

[ScottyUK]

Hello PNachtwey,

What is your experience of the digital interface Balluff (or Temposonics) transducers? Sounds fairly good for a system requiring a fast loop response. I looked at them a while ago for fitting to a steam turbine but I wasn't prepared to put the processing electronics in the hostile environment of the turbine-generator if I could avoid it and the distance to the turbine controller was long so it was 4-20mA on that job. I am interested in the SSI type for future applications though. I have been using the US version of the BTL-5 explosion proof transducer where it has proved almost impregnable to mechanical damage, and resistant to process fluids and environmental contaminants. Pricey, but this industry can't afford downtime.


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[PNachtwey]

ScottyUK, my company makes hydraulic motion controllers. Actually general motion controllers but we specialize in hydraulic motion control. In the US and Canada the lumber industry does just about all linear positioning with hydraulic cylinders with the MDT rods inserted in the back of the cylinder and the magnet mounted on the piston. Since our company makes over 50% of the industrial hydraulic motion controllers in the US and Canada we have a very good picture of how differences between the different rods.

I prefer the start/stop interface or the SSI. Start/stop is the basic or native interface. The rods will naturally do start/stop. I like this interface because I know there is nothing between the motion controller and the data coming back. SSI provides more resolution at a slightly higher cost. What I like about SSI from a practical point of view is that there is no need for a gradient. All 5 micron rods are the same in theory so one can change on out for another without changing the scale factor. One must be careful to screw in replacement rods the same number of turns as the old rods so the offset remains the same. Otherwise a new offset must be determined.

I dislike analog rods for motion control. That includes presses. Hint. The problem is that analog signals are noisy and drift but the worse part is that the motion controller or PLC doesn't know when the data was valid. It is very important to know not only the position but the time at which the position is valid. This rules out analog rods because the start pulses are initiated by the rod itself and not the motion controller. I can go into this in more detail. The simplest interface is the start/stop interface. The motion controller sends the start pulse every millisecond or half millisecond. We use a FPGA, field programmable gate arrays, to do this because interrupts have jitter that causes velocity errors. There can be no software involved with the timing. We use a 240 Mhz counter that will provide about 0.00045 inches or 12 micron resolution. We use observers to provide excellent velocity and acceleration feedback data. The SSI rods convert the time between the start and stop pulses to a digital number that represents the position. Initially I had problems with both Temposonics and Balluff because they would claim 5 micro resolution but the rod update was asynchronous to the controller. This means if the actuator is moving 1 m/s it is moving 1mm/msec. I would have a 5 micro position that was valid anywhere within the millisecond which essentially meant I had 1mm resolution at high speed. This has all be fixed now but Temposonic and Balluff don't synchronize the positions to the controller in quite the same way so there are differences in the quality of feedback. Our motion controller can graph the data from the rod a plot it. My favorite test is the drop test. I point the rod up and drop the magnet. I should see a smooth acceleration due to gravity.

I/we have worked with both Temposonic and Balluff for many years. We try to stay neutral. Both companies have their advantages and disadvantages. I have see both screw up big time and other times do very well.

If you have more questions just ask here or see

http://www.deltamotion.com.

 

[itsmoked]

Nice lesson P. Thanks.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[patmando]

Thank you for all your advice. There are scaling instructions in TI, but for the most part here they prefer to use math instruction blocks. The main problem is that the scaled value fluctuates +-20 thousands. Say my press target is .835 inches when fully closed, when it goes past the target by 20 thousands, we have a safety function to shut down press hydraulics. So how can I code it so it does not fluctuate so much, being if my feedback signal is good and no noise in my system.??

 

[PNachtwey]

What do you think your source of error is?

One must be prepared to adjust the offset. A press will "warm up" and the offset will change as function of the metal expanding.

 

[ScottyUK]

I'm not a press expert at all, but a couple of thoughts to consider:

How fast is the approach to the target, and how is the loop tuned? A critically damped loop will see some overshoot. Could this be a servo tuning problem rather than a problem with the feedback?

Also consider that reliably resolving 20 thou in a 130" travel is approaching the limit for a 16 bit DAC. How good is the DAC fitted to the Siemens PLC I/O channel?


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[patmando]

Thank you for your inputs, the press we have is for pressing particle board. It's a 20 hole press, steam for heat which can reach over 300 degrees. There is an existing rod in place, but not of that length. The code for this one is in a parallel rung of code. Using the same type of logic instructions. Not in PID. The scaled output for the existing one does not fluctuate. Speed in the up cycle is not an issue I believe, only in the final pressing cycle is there an issue to reach the disired thickness. The signal goes into a TI505 Analog 8 channel module, wired for current. Maybe I need to do a PID loop.? any more thoughts?
I appreciate your replies.

 

[PNachtwey]

You are aware that each rod will have its own scale factor and offset values because each rod is different? Even if you used SSI rods the offset would be different.

If so then I don't think this is a scaling problem. I think it is a mounting problem. Is the problem rod mounted differently than the good rod? Have you tried using a dial indicator. .020 inches is a lot.

A PID will not cure a feed back problem.

Are both rods using the same analog input card?

It is time for you to start telling us about all the things that are different between the rods. Mounting, location, power supplies, wiring, anything. You should be thinking about what causes a difference, any difference.

So far you have not said this is a control issue.

[OT]

"A critically damped loop will see some overshoot."

No it shouldn't! Under damped systems over shoot. Critically damped system aproach the set point exponentially. However this requires the proper form of PID and that usually isn't available on a PLC. The problem is caused by the zeros introduced by the P and D gains in the forward path. Another form of PID is the I-PD where the P and D terms are only in the feed back path and therefore dont add zeros. I-PDs can be tuned so they don't over shoot. What one give up is bandwidth.

Ask about this on another thread in the control section. I have graphs and Bode plots that show the differences between PID, PI-D and I-PD
[/OT]

Peter Nachtwey

 

[patmando]

PNachtwey,, thankyou for your replies.
I do not believe I have a control issue or even feed back at that. I think it is more of an linearity performance problem. At 125 inch rod set at 1 recirc is about the best I am going to do. So I think at 0.02% times the length of the rod is going to give me 0.025, which about the swing I am seeing.
Thanks again. You were a big help.