laser frequency stabilization using pid controller

[Yasmine90]

The setpoint is in Volts, since the processed variable is in Volts, I thought it has to have the same unit. 

I put 10 ms as a time constant to the lock-in, it seems that GPIB is not that low, I mean, the value of the output of the lock-in passes instantly via GPIB but for the laser it's a bit different, it's clearly slower.

[Yasmine90]

Hi,

Can you tell me please how can I adjust the parameters of a pid on labview without knowing the transfer function of the system?

[Yasmine90]

Hi,

Can you tell me please How can I adjust the parameters of a PID on labview without knowing the transfer function of the system?

[johnsold]

There are techniques for tuning PID controllers which do not require detailed knowledge fo the transfer function. A Forum participant with a lot of PID experience, nathand, recommends "a tuning method such as Cohen-Coon or Ziegler-Nichols." Search the Forum for "PID tuning" for many posts on the subject.

The time it takes the laser controller to respond to the GPIB update will set a minimum loop time for your feedback loop. If that time can vary substantially from one iteration to the next or if that time is just too long, it may be nearly impossible to get good stabilization. Sometimes feedforward ssytems can be used to compensate for long delays but you usually need a pretty good idea of the transfer function to design them.

How quickly does you laser drift away from the desired wavelength if the input current is held constant?

Lynn

[Yasmine90]

I know the change of the wavelength of the laser depending on the drift of the response of the lock-in.

I put the time constant of the lock-in to 10 ms.

It's clear that the value of the current that the pid outputs on LV is a bit shifted compared to the current displayed on the laser controller itself but in any case I don't need a stabilization every second, the laser doesn't drift that much, the drift of the laser takes time so even if LV sends the correction via GPIB every minute it's OK. 

Now my problem is that I don't know how to fix the parameters of the pid, if the assembly of the system is correct of course.

I'm used to stabilize devices on Matlab with PID. I send a step voltage to the device and I see its output and from that I know how to operate, but now I don't have any idea.

I went to the pid forum but it didn't help a lot.

I was wondering that maybe you you can help me better since you've already worked with such instruments and faced such problem. 🙂

[Yasmine90]

Actually, if I put nothing but an appropriate Kp and if I let the experiment running for a whole day the laser's current doesn't drift much, but when i caluculate the drift of the wavelength, it's much more than expected and not really what I want. 😕

[Yasmine90]

Hello,

I would like to know wich timing tool I can use to set the time step of a while loop and how can I write in a file in the same loop with a longer time without affecting the execution time of the loop?

Thank you in advance. 🙂

[johnsold]

Did you ever get the laser adequately stabilized? Your last posts on that topic seemed to suggest that wavelength drift was larger than the current drift. If that is true, then there is something else causing drift, possibly laser temperature.

As for your timing questions, it would help to see the code you are using. Generally you can use Wait (ms) or the timing of a data acquistion process to set the loop time. 

Two methods to write to file at slower rates than the loop rate are:

1. Put the file write inside a case structure. Only call that case when the desired write time has elapsed, the desired amount of data has accumulated, or the specified number of iterations between saves has occurred.

2. Use a Producer/Consumer architecture to allow largely independent timing of acqusitions and saves.

Lynn

[Yasmine90]

Hello,

Thank you for your response.

Yes actually, I haven't suceeded in stablizing the laser completely and adequately, especially that I couldn't fix the exact parameters, since the empirical methods don't work with this kind of system 😕 plus other constraints.

Concerning the temporization, I will try the first method, it seems to be easier.

Thank you.

[johnsold]

Is the output of the photodiode proportional to the wavelength of the laser? How do you measure the laser wavelength drift? 

How much drift do you get with a constant input to the laser controller? What is the resolution of the controller?

I am trying to get some idea of which component is the source of the remaining drift.

Lynn