Sample Time of the PID ISA

[Saffa]

Sorry for the delay in reply, but I do actually have my own projects to complete too!

To check for noise on your process value, you could always connect an oscilloscope across your input terminals? Hopefully you've got one! i don't have what other tools you have available, I normally just set up a trend on our HMI package so I can see what's happening. Are you using Versa Pro to program it, if so what version? i might be able to take a look.

How does the machine behave if you simply "ramp" the control value up by hand? Do you see non-linear behaviour at 84-90% output? If so, then your PID loop is going to have exactly the same problems if the machine is exibiting some form of resonance or something like that at particular speed / frequency / whatever.

 

[jcp]

Hi;
I used LM90 to develop logic.
Thanks a lot to guide me. I explain some points may be for that reason its work like this;
1. I check the PV with Oscilloscop there is no noise and also check CV no noise found.
2. THe problem which i think is that SP and PV is not matched for example
I use set point as
SP=62*115.20= 7142.4 Hz and then increment one by user press button foe example User press one push SP change to 63*115.20=7257.6
now in GE 90-30 the PID required only integer value.
While PV is directly feed to the PID %AI.

If i scale the setpoint and PV to counts or change in %age is same gains work?

I attached an image and you see that SP=7373/64% while
PV=7357/63%
If i calculate manually then
SP=7373*100/11520=63.1% and
PV=7357*100/11520=63.8%
why SP shows me 64%.
Is error calculate on the percentage or on the real input values? If i select SP-PV=7373-7357 or 64-63
How can i correct this?
Regards

 

[jcp]

Hi;
Friends you see why this happenning?
Regards

 

[Saffa]

Hi JCP,

I've done a bit of reading in the 90-30 series manual:

http://support.ge-ip.com/support/re...tent/live/DOCUMENT/0/DO704/en_US/gfk0467m.pdf

And (as I suspected) the PID is calculating error based on the raw, 16 bit values not the percentages shown on screen. These are purely to make things more human readable. I'd look at the counts myself.

If your PV seems to have an offset from the SP, try tweaking up the integral gain a little.

If those "1%" setpoint changes are still causing instability, a technique I've used before at a water plant to prevent big SP changes causing problems is to have the operator's setpoint be pointing at another memory address. When that value differs from the PID setpoint, I either slowly ramp up or ramp down the PID setoint (using a timer and increment function) until they are equal.

Other than that I don't think I can really help much more, since I avoid our GE 90-30 like the plague and when I really, really, realllly have to do anything in it, we're running Versa Pro not that funky logic master software!

 

[jcp]

Hi;
Thanks i follow the instruction and now system is almost stable. But a point come where it create a problem like when speed reaches 88.5 to 91 the overshoot occur while i also slow down the sp .
I think some resonance occur. But no noise found.
How can i handle the resonance effect in the logic or from the PID.
Regards

 

[Saffa]

Hi JCP,

That could be resonance of some kind, best way to check (to make sure it's not your PID causing it) is to put the PID control in manual and try running at those speeds. if it's exibiting oscillation even with no compensating control action then yes, it's possibly resonance.

If that particular frequency band (and it looks like a reasonably small band) experiences resonance, then you'll probably have to find a way to "skip" over that frequency. Not sure how best to do this in your application, you might need to do a couple of compares of your CV and SP to this band of frequencies, and either manipulate one or both to avoid running at that frequency. Hopefully that's not an issue for your process?