Noise Filters

[kamenges]

Honeytree-

That's the same as Tom's implementation except you have a way of defining 'k' in terms of scan time and time constant. Your method also yields values very close to my posting. You just define 'k' a little differently.

Keith

 

[Peter Nachtwey]

I acts effectively like a single higher order filter. Cascading two first order filters will produce the effect of a single second order filter. This filter will have a steeper roll-off as frequency increases than the first order filter does.

You forgot phase lag. That is why you can't add too much filtering to the input on a PID. The filter adds phased delay which gets the open loop phase response to 180 degrees much quicker.

Where I still need to do some looking is in converting time constants to frequencies. A first order filter has a time constant that defines the amount of time required for the output to reach a specific level based on a step input. Higher order filters are expressed in terms of frequency and damping ratio. How do I move between the two representations?

Keith

frequency calculations are normally done in radians. Frequency in radians is the inverse of time constants.

exp(-alpha*t) = exp(-t/Tau)

if alpha = 1/Tau.

Notice that the term ( -alpha*t) and ( -t/Tau ) are unitless.
Convert frequency in radians to Hz by dividing by 2*PI.

Check out this:
ftp://ftp.deltacompsys.com/public/mcd/

I got listed as one of the contributors to the SCI.ENGR.CONTROL newsgroup because someone liked my high pass besel or butterworth filter.

BTW, there are all sort of filters. A PID is a filter. One should be more specific when asking about filter code.

 

[Tom Jenkins]

Rats!

Tom, the description in your ladder and the equation in the compute don't match. The description appears correct.

You are absolutely right, Keith. Here is a revised file, and as a penance I added the timer to set intervals between filtering.

 

[kamenges]

Originally posted by Peter Nachtwey:

You forgot phase lag. That is why you can't add too much filtering to the input on a PID. The filter adds phased delay which gets the open loop phase response to 180 degrees much quicker.

You caught me. You would think I would have listed that too, since I posted this in post #5:

Originally posted by kamenges:

This will give a simple first order response (like a capaciter charging). You can string several first order filters together to get higher order filters with sharper cutoffs, but that has it's own issues. As the filter order increases so does the delay for a given filter scan time.

Thanks for the explanation of time constant versus frequency. That makes sense. Now I just need to think a little about how I can use that information.

Originally posted by Peter Nachtwey:

BTW, there are all sort of filters. A PID is a filter. One should be more specific when asking about filter code.

That is true. It is a filter with one pole and one zero and adjustable gain, I believe. However, the pole seems to be in kind of a nasty spot as far as filters go. Which brings me to an off-topic question. Is the integrator in a PID filter a necessary evil in most systems to satify time domain requirements? It seems to me that adding a pole in most systems is the last thing you want to do, especially a pole at the origin.

Keith

 

[Honeytree]

Conclusion?

I did just a little tweek to the last pdf I posted.
Works fine. Using the last scantime keeps it sweet and simple.

Thanks again for all of your help.

 

[Peter Nachtwey]

Good question

That is true. It is a filter with one pole and one zero and adjustable gain, I believe. However, the pole seems to be in kind of a nasty spot as far as filters go. Which brings me to an off-topic question. Is the integrator in a PID filter a necessary evil in most systems to satify time domain requirements? It seems to me that adding a pole in most systems is the last thing you want to do, especially a pole at the origin.

Keith

An integrator is required for type 0 systems like temperature and velocity systems unless one uses a feed forward or bias.

Position systems don't need integrators if getting to position is not required.

Integrators do add phase delay and slow down response.

Hi from Tumbarumba AU!

 

[kamenges]

Originally posted by Peter Nachtwey:

Hi from Tumbarumba AU!

We ain't in Kansas anymore, Toto! That's a little ways from home for you. I hope it's for fun.

Keith

 

[BobB]

Hi from Tumbarumba AU!

Doing some trout fishing by chance? Just finished a water features job at Brungle which is near Tumut - close to you. For a trout fishing teaching farm.

If the pub has Carlton "black" on tap it is excellent. So is James Squire brown ale, Coopers pale ale etc etc. Bottle stuff try Boag's St. George - citrus flavoured hops - much better than Corona.

 

[Peter Nachtwey]

I am working

We ain't in Kansas anymore, Toto! That's a little ways from home for you. I hope it's for fun.

Keith

I am at the saw mill that has all those Ethernet nodes. Alan Case works sometimes in the dry or planer mill. I am working in the green mill. I have met Alan and may again tonight.

BobB, my plan is fly out of here to Sydney and look around a few days before going home. I like Carlton Draught. Haven't had the others yet.

Hopefully they will let me go home. At this time I am helping a saw mill integrator. After I am done the saw mill will hire me to train personel on motion controllers.