LabVIEW
PID tuning array instance
You might need a reverse action.
For instance if a dT has a positive error, the temperature should be increased. If a dP has a positive error, a release value should be decreased.
IIRC, you can reverse output min and output max to get that effect.
I think theoretically, negating P, I and D should do the same. There might however be checks in NI's PID implementation that check P, I or D <0...
Search LabVIEW like a graph!
@AndrewIons01 wrote:I just tried that and got half of them going to the setpoint and half of them were not moving towards it. All four need to go the setpoint.
Then check what causes this.
See if the loops that need inverse outputs have their process values reversed, or find any other relation, find the cause, then fix it.
We don't see any of your values, so we can't comment on that.
I don't know if you know, but you can put probes on wires while the code is running. That gives you run time information (that we don't have).
Hi,
wiebe@CARYA wrote:
You might need a reverse action.
For instance if a dT has a positive error, the temperature should be increased. If a dP has a positive error, a release value should be decreased.
IIRC, you can reverse output min and output max to get that effect.
IMHO I would not swap the output min/max: you set a (logical) range for the PID output value.
To set a reverse action you "just" need to change your PID gains! And to make it even more easy: you just need to change the sign of the P gain!
It helps to learn about the "basic principle" of the PID gain parameters:
- P gain describes the (linear) relation between error input and PID output value, like "when there is a temperature error of 10 K then I need to change the heater power by 25 %". P = heater power change / temperature error = 25%/10K = 2.5 (%/K)… (This is a very simple P gain estimation, even simpler than Ziegler-Nichols.)
- I gain describes the time needed to get the same effect on PID output value like the P gain, when a constant error input is received…
- D gain describes the time to react upon a change of error, but usually it is ok to start with a PI controller…
- In NI's PID implementation an I/D gain of 0 means "no I/D action"…
@AndrewIons01 wrote:I will check to see what causes this and let you know the result. Is there a way I can modify the code to enable you to run it if I send it to you?
In short, you should send something that doesn't have any dependencies to stuff we don't have.
You can replace everything that returns data with some sort of simulated values. These could be constants or controls with default values set to reasonable values...
I guess you can log values that enter and exit a VI by enabling database access on the VI (right click the VI, "Enable Database Access"). Then you can send the 'database' and the VI, and we should be able to play back the results. This option is staring in everyone's face, but it's hardly ever used (anymore). Never used it...
There is at least 1 VI missing.
More importantly however, we're missing the camera setup.
Even if we had all code, we wouldn't be able to run it.
If you suspect a VI is giving you trouble, all we need is that VI with realistic values. Everything else will just be distraction.
