Analog signal filtering question

i have a laser sensor from AB that is measuring block height, it works great but problem i am having is the signal is just all over the place up and down....i started looking on here to see what people have already said about signal flitering and i believe i found my answer 2ce, but i just cant grasp how to set this up in the actual program.... can someone show me an actual lader example (pref. on logix 500)

this is what i used for filtering

FV=FV+C(NV-FV)
Where:
FV= Filtered value
C= constant (range 0-1) The smaller the number the more dampening.
NV= New Value


and basically just took that equation and stuck it in a compute instruction and for the range i just used 1.

now there was another i seen, and i couldnt figure out at all how to set it up......reasion i liked it was it wasnt a simple compute instruction and after setting it up i can actually understand more as to whats going on. kind of like long division as opposed to using a calculator. i like knowing what im doing rather than just having the machine do all the hard work. after i figure it out, then have the computer do the work haha...

heres the other

Total = Current_Sample + Hold_Value
Average = Total / Filter_Factor
Hold_Value = Total - Average
Current_Sample is your raw feedback signal. Average is the filtered value. Filter_Factor is a variable that you manipulate to find the best compromise between response and averaging. The larger the value of Filter Factor, the less impact a single sample has on the average, but the longer it takes to respond to a bona-fide change in the signal. Hold_Value and Total are internal variables that are only used in the calculation.



thanks guys.

can someone show me an actual lader example (pref. on logix 500)

Click to expand...

F32:114 is the filtered value.
F32:105 is the unfiltered value. ( New Value)
0.0001 is a constant. This can be a floating point register which can make it earier to change to get the proper filtering.

I did this and my number won't stop moving up if the signal is 0 doest matter keeps going up where did I go wrong

Well I changed some things around and now unless my constant is 1 my signal although it's at 11000 my computed filter is at 3277

Post your program and someone will take a look.

the program is a disgrace right now, i have been moving things all over the place and now i have the filter on it works good but, now its slow. so i have to just think of a condition as to when to start filtering...thank you again!

if i have anymore problems i will post program

i use an old shcool method to filter with few tags
Each scan, i add a real tag with the instant input value,(Average sum) i add another real one by 1 at the same time(Average count) and each few second, i divide the first by the second to get the averaged result (Result= Average sum/Average count) and i clear both tag during the same scan to start again averaging for the next scan....

I like the mickey's one too, i will give it a try on my next program...

C=1-e^-2*π*Hz*Δt
The amplitude of the noise will be down -3db at the frequency ( Hz )

peter.....what?? haha

Simple

escoplcguy said:

peter.....what?? haha

Click to expand...

C is the C in FV=FV+C(NV-FV)
Now you know how to calculate C as a function of the desired bandwidth.

Peter Nachtwey said:

C=1-e^-2*π*Hz*Δt
The amplitude of the noise will be down -3db at the frequency ( Hz )

Click to expand...

I can understand escoplcguy's response....

posting a formula like that with no explanation of the terms is almost unforgivable.

Now you know how to calculate C as a function of the desired bandwidth.

Click to expand...

only if we know what n and delta-T relate to....

please don't forget that not everyone has the level of understanding/intelligence you are gifted with...