
kwaltzer (Customer) asked a question.
I have a Gefran WPG-A-M-0400-E linear position sensor running into Productivity 3000 using Quabbin control and signal cable, shielded, 3 conductors, 20 AWG, 7-stranded, tinned copper wire. I am using it to control a couple things.
1) A fixed set position point that will tell the operator to add material. Ex. when deck position goes below 85mm, prompts operator to add material for next cycle. This works just fine.
2) Using the change in position over time to control part of the cycle. this is where i am having some issue. Once ram pressure has met a set amount the cycle starts, at this time I have the scaled linear output in a Change of Value Block with a sample rate of 1000msec. This Change of Value output is compared to a set amount (ex. 0.05mm) with a compare contact, once it is equal to or less than for longer than X-time (ex. 3 sec) it will start a "packout" time, which when done will end this part of the cycle.
Problem I am having is the linear sensor seems to have a "pulse" in which it will show a jump in position even though there is no physical movement, and it seems to jump almost at a specific time interval. (Please see attached picture of graph). So during the compare of the change in value part of the program just before it goes to the "the packout" phase it will jump an the controls think that it hasnt been below the compared set point long enough for the timer to start, thus the cycle wont end because the jump keeps happening in the same time interval. LS1= Linear Scale
Point COV = Each point on the graph is the Change in Value function output
The red arrows show the jump in the Change in value keeping the cycle from moving on.
Here is the sensor "calibration" for the linear scaling function.
in min = 0 point on the scale
in max = is the max point on scale
out min = what the scale means to us. 0 means 0mm
out max = 500mm
input is live from the scale before scaling
scaled is live after the scaling
The data type for the scale into the card is Analog Input integer 32 bit
LS1 scaled is a Float 32 bit.
Should the Analog Input be a float also???
I have also tried both types of cursors for this sensor (floating and ride on type) and didnt make a difference.
Change in Value block. Flasher 1 is used to plot the Change in Value point on the graph.
the compare contact rung and how it works to control the packout start.
If the scale didnt jump like it does this set up works great. I can kind of finagle the setting points to get it working but its not ideal as set point changes would have to be made through out the day and the idea behind this is it would just be a start and go while the controls adjust cycle automatically. This is running on some smaller machines with success but it still has the jump but not enough to affect anything.
Thanks for your input!
That is exactly it. We utilize a variable vane pump with unloading valves. Currently we use the unloading valves to double as a regulator for the system pressure, and everyone of those spikes is the system charging back up to set pressure. It just so happened that on this particular press with this material it is showing a large fluctuation in position when the system pressure rebuilds.
This algorithm is a new feature i came up with for these machines and that amount of rebound was not expected.
I do still get fluctuation on the scale readout when I know there is no movement but its about 0.03mm (live in the scaled readout) but that is manageable with the Change of Value pulse rate at 500-1000msec.
Moving forward I order up some pressure regulators to change the system pressure around to maintain even pressure throughout the cycle.