Hello everyone.

I have found an iPhone app that claims to do PID Control loop tuning. Obviously it's very limited because you can't hook it up to the process, but instead it lets you enter necessary process data. I haven't had the time to test it for real yet, so I don't know how that works.

The question: The app claim to use something called the "Lambda Method" to calculate PID parameters for the controller. A search on the web doesn't bring up much about that. Is that even a real method for Controller tuning?

In case anyone wonders, this is what I'm talking about: http://itunes.apple.com/us/app/looptuner/id436691437

 

Hello Michael,

Yes, "lambda tuning" (method) is very real. A google search brings up more than 9 million results - perhaps too much? One place to start would be to look for this thread:

http://www.control.com/thread/1026175484

in the control.com archive. Please let us know if you have more specific questions.

Regards,
Gerrit.

 

You're right, "lambda tuning" gives far more relevant results than "lambda method".

Anyway, I have now tested the app in question at a clients site and I'm quite pleased with it. Essentially, you draw your bump test on your iPhone, select the closed loop time constant (lambda) and it gives you new parameters for your controller. I successfully stabilized an oscillating tank level with it, so it seems to work!

It was some time ago I had to do control loop tuning. I found some good resources online about Lambda Tuning, but what they don't tell you is:
How widespread is the use of lambda tuning? For what kind of processes have you used it successfully?

 

> How widespread is the use of lambda tuning? For what kind of processes have
> you used it successfully?

MichaelV

I have used Lambda tuning extensively in the power industry for around 10 years and have found consistently good results, far better than say Z-N which can produce instability when systems are highly coupled(eg steam pressure/temperature/electrical output). It makes tuning cascade controllers and coupled systems easier and you can calculate the response to meet performance requirements.

But I don't think it is used enough! I have seen papers in Steel and Pulp and Paper industries.

The App is interesting but suggest you check the controller algorithm type (series, parallel, ideal) for each controller you tune. Also, you <b>can</b> set lambda less than the time const if the lag (dead) time is relatively small, so the app's comment about instability is not always accurate.

 

Thanks for your input. I have had some time to read up on lambda tuning and some opportunities to test it for real. I also got that comment on a loop I tried to tune: "Lambdas close to T may be unstable if there's deadtime." I'm relatively new to this so I thought it was a nice heads up. But you're right, my loop was quite stable at 1.2*T, probably thanks to only 3 second deadtime and low process Gain.

> Also, you <b>can</b> set lambda less than the time
> const if the lag (dead) time is relatively small, so the app's comment
> about instability is not always accurate.