PID Trainer, issue on setup

[MaxK]

About going back to the drawing board...

Imagine that you are building a racing car.
You built a prototype. To find out if it's good or bad, you need to run it on a twisty test track, but your test driver can only drive in a straight line. After driving a test driver like that, all you know is that your car didn't even turn.
BUT this is not due to the properties of the car, but simply because the test driver does not know how to turn (which was known).
But you didn’t know anything about the car’s ability to take turns.
Question: what exactly will you remake in the car?

 

[geniusintraining]

I know you want to move on, but just sayin'

No no no... absolutely not, I want it to work because I think it would be a great tool for people to use and less than half of the price I make and sell the other one for

I have made a few changes to the program since the first.... well many, see attached the last program and screen shot

Question: what exactly will you remake in the car?

When I wreck in the first turn

I am not building something that is set in concreate but the tools I am building are for the student to get more comfortable with programming and have the ability to figure it out, but I also need to supply a PID trainer that will show them what happens if you only use P or how you can use PD (like you showed me) and how something will react to what your settings are and give them a general idea on how a PID loop works, this also needs to work with any PLC that can use 4-20mA as the I/O

I am willing to make any changes that you guys think are needed but if it gets to involved maybe just scrap the car and build a truck or a boat....

 

[geniusintraining]

Program in PDF and RSS

 

[drbitboy]

I am willing to make any changes that you guys think are needed but if it gets to involved maybe just scrap the car and build a truck or a boat....

This should not be too involved:

• Rung 0001, assign a value of 16383 to the Scaled Max. parameter of the SCP.
• Rung 0002, assign a value of 3604 (22% of 16383) to the SP parameter (I assume that is via the Setup Screen, or PD9:0.SPS)

Then re-run and re-tune the PID.

 

[drbitboy]

Btw, how much did you spend on getting a value from the infrared distance to the analog input?

 

[geniusintraining]

This should not be too involved:

• Rung 0001, assign a value of 16383 to the Scaled Max. parameter of the SCP.
• Rung 0002, assign a value of 3604 (22% of 16383) to the SP parameter (I assume that is via the Setup Screen, or PD9:0.SPS)

Then re-run and re-tune the PID.

I can do that but I will need to change more than just that... That is where I was changing the input from the raw to 0-100, now my PV range is 198-20

 

[drbitboy]

I can do that but I will need to change more than just that... That is where I was changing the input from the raw to 0-100, now my PV range is 198-20

I don't think so. The rescaling and SP change essentially multiplies PV and SP by 163.83. The retuning of K_C should adjust for that rescaling in the PID. Are there PV limits in the PID somwhere? I'll have to look ...

what do you mean by "198-20?"

 

[geniusintraining]

what do you mean by "198-20?"

If the ball is all the way to one side it was (the PV) at 9 and all the way the other side it was at 100 and after I made your changes its now 20-198

 

[drbitboy]

If the ball is all the way to one side it was (the PV) at 9 and all the way the other side it was at 100 and after I made your changes its now 20-198

If you changed the 100 of Scaled Max. on Rung 0001 from 100 to 16383, then at the point when the PV (N7:20) was 9 before, and I:1.0 was around 5385, then after the Scaled Max. change the value of N7:20 should be 1474 at that end of the beam, and when the N7:20 was 100 before (I:1.0 was 29500 or higher), then after the Scaled Max. N7:20 should be 16383 at the other end of the beam.

If you are seeing something else, then I find it hard to see how you changed only Scaled Max. to 16383, and left all of the other SCP parameters, i.e. Input Min., Input Max., and Scaled Min., at their original values of 3000, 29500, and 0, respectively, and got those results.

 

[drbitboy]

What are the values of I:1.0 when the ball is at the ends of the beam?

 

[geniusintraining]

Maybe I changed the wrong thing... I will double check in the AM

 

[drbitboy]

Ah.


I'll bet a large wad of money that you changed the value of Input Max. in the SCP on Rung 0001 from 29500 to 16383, and left the value of Scaled Max. at 100.

Change Input Max back to 29500, and change Scaled Max. from 100 to 16383, so Scaled Max. is the only number changed in that SCP on Rung 0001 compared to the PDF you attached a few posts ago.

 

[drbitboy]

Maybe I changed the wrong thing... I will double check in the AM

+1. LOL. Thanks, that's definitely it. Tomorrow is Saturday, let's take the weekend off and maybe take our wives out to dinner, eh?

Gotta love how a project progresses: pillar to post, like inebriates down the sidewalk.

Meanwhile, @PeterN keeps Orville Redenbacher in business.

 

[Peter Nachtwey]

Meanwhile, @PeterN keeps Orville Redenbacher in business.

Naah, it is Friday. Time for a beer or two or three or ....


I came to the conclusion long ago that a real motion controller is the way to go with fast updates that are synchronous to the PID scan. I read some of the posts above ( yes drbitboy I can read ) and I saw where the position updates are asynchronous to the scan. Updates of 60 Hz or 16 ms are too slow. I am amazed GIT has the signals as clean as they are.


I have pointed this thread out to some of the guys at Delta. We couldn't make anything like what GIT is trying to do for less than about $5K. I am sure that is WAY to expensive for what GIT wants to do. We use these trainers in our training class. We do sell some but they aren't what anyone one this forum would buy.


What is disappointing is that I calculated the controller gains for getting a nice critically damped response and no one is trying them out.


I have also look at other people's ball and beam systems on YouTube. Some are quite good. There are some very good 6DOF systems controlling the position of a ball. It wasn't cheap and obviously the math for a 6DOF system is way more than what GIT wants to deal with.


So keep plugging along.

 

[geniusintraining]

Yep much better... I will work on the tuning this coming week, I will be busy again for the next few days but once I get caught up I will work on the tune and then update

Just to recap your thought was the 0-100 scale did not have enough resolution so this will give the PLC more room to move and be more accurate?