Smoothing Analog Inputs

OkiePC · Jul 30, 2018

Mickey said:
You can, but not necessary. It will just update every scan.

Just adjust the filter constant to get your desired results.

I usually use a bit from the free running clock and a 1-Shot preceding the math so that I am sure to have a regular controlled filter when doing this in a Micrologix. This is especially important if I have things going on that might make the scan time irregular.

VAN · Jul 30, 2018

Mickey said:
You can, but not necessary. It will just update every scan.

Just adjust the filter constant to get your desired results.

Right, but it'll give an extra parameter to slow the filter down a bit. I worked with a couple of boiling solutions and adding a timer allowed you to smooth the filter even better.

This gives a smooth line (trend) instead of constant up and down.

Cookz615 · Aug 3, 2018

janner_10 said:
I used this guys Time Average AOI, like the MAVE but usable in ladder. Works very well.

http://blog.briangallimore.com/2014/08/29/time-average-in-ladder-logic-controllogix-aoi/

Excellent find! This worked out perfectly for some analog inputs. Glad I stumbled up on this thread now. Thanks!

V0N_hydro · Aug 3, 2018

VAN said:
Right, but it'll give an extra parameter to slow the filter down a bit. I worked with a couple of boiling solutions and adding a timer allowed you to smooth the filter even better.

This gives a smooth line (trend) instead of constant up and down.

that doesn't make sense to me. By adding a timer that makes the filter run less often it is just decreasing the sampling rate. If the filter doesn't appropriately account for time then it would slow down the filter, but if the filter knows how long it has been since the last time it has run and is calculating change over time correctly then decreasing the sampling rate but keeping the filter time constant the same should make the output more bumpy not smoother.

If you increase the filter time constant the output should change slower and smoother. Generally it is better to sample as fast as possible, although it might not be necessary for signals which change slowly.

VAN · Aug 5, 2018

V0N_hydro said:
that doesn't make sense to me. By adding a timer that makes the filter run less often it is just decreasing the sampling rate. If the filter doesn't appropriately account for time then it would slow down the filter, but if the filter knows how long it has been since the last time it has run and is calculating change over time correctly then decreasing the sampling rate but keeping the filter time constant the same should make the output more bumpy not smoother.

If you increase the filter time constant the output should change slower and smoother. Generally it is better to sample as fast as possible, although it might not be necessary for signals which change slowly.

Right it slows down the sample rate. If the sample rate is 0.001 of a second or even up to like 0.01, it just smooths it out better than going off of the scan rate. Or you can make the timer 0, but I usually like to build this out like an AOI and I would rather have the extra parameter just encase.

V0N_hydro · Aug 7, 2018

sampling at a higher frequency should result in a smoother output than filtering at a lower frequency. If that is not true then the filter is not calculating or using delta-t correctly.

kdthornton · Oct 1, 2018

Found this thread and it worked perfect for what I needed to filter some flow signals feeding a PID loop. I needed to use it in FBD so I made an AOI block and figured I would share it.

Pretty simple, unzip, import AOI. Rename it if you like and enjoy. Thanks Mickey for the code.

Peter Nachtwey · Oct 1, 2018

It is best to smooth the analog output from the PID, not the inputs. Smoothing the inputs make the PID hard to tune.

Smoothing Analog Inputs

OkiePC

Lifetime Supporting Member

VAN

Member

Cookz615

Member

V0N_hydro

Member

VAN

Member

V0N_hydro

Member

kdthornton

Member

Attachments

Peter Nachtwey

Member

Similar Topics