controlling laser diode with temperature feedback

[AndyClegg]

It sounds like you have good understanding of the input/output characteristic - and so you should consider using that within feedforward action, combined with some feedback control since it won't be perfect. This may make it easier to achieve the required performance, than if you tried to do everything with just feedback control. As Lynn asks, how well this works depends how repeatable the input/output characteristic is.

[grega]

The change of the temperature (when the generator is on) depends on the generator power, which can be from 100 to 1000 W. In that way the probe can glow in a mater of a second to "never". I will post some graphs tomorrow.

Yes that's exaclty what I thought, Lynn. That is why I will use the probe in the place inside the tube where the plasma is not so strong (if I move away from the point of plasma creation, the density of the particles decreases, therefore the probe temperature also decreases). Basically, with the probe, I will detect only small changes in temperature variation of the catalytic material (small changes in density). I thought I can make some kind of protection if the plasma is too strong...but I gues that's impossible, since the IR sensor response is only about 100 ms. I think I am trying to automate things, that cannot be automated...at least not with that kind of hardware options. So I guess the user will have to take care when the plasma will be turned on, that the power of the generator won't be too high.

Maybe the best thing would be to turn off the laser when the "plasma generator ON" signal is sent, and then gradually increase the laser power, after the plasma is ignited, to achieve the desired value on the IR sensor.

[johnsold]

Grega,

From what you are now describing it appears that you are not yet to the point where you can begin to design a control system. First you really need a good model of your "plant" (the term used by control systems engineers) to describe the behavior of the system you want to control.

You have a non-linear, poorly reproducible laser current to probe temperature relationship. You also have a very high rate, and probably non-linear, input from the  plasma generator. On top of that your sensor responds much more slowly than the temperature of the probe you want to control.  If this is an accurate assessment of your situation, then you may not have a good solution.

You might want to consider getting someone skilled at control systems to look at your project. I think this kind of situation may require on-site expertise. An experienced person can sometimes see things which you might not even think to describe or find relevant some minor aspect of the behavior.   At the very least investigate faster temperature sensors.

Lynn

[camerond]

It sounds like one characteristic you need is a feedback control which can drop the laser power quickly to prevent damage, but raise it only slowly (slightly slower than your sensor response) to prevent runaway oscillation, giving a sawtooth-like control curve. That way, you should be able to sneak up on the optimal laser power without blasting your experiment away, but faster than a manual start. Programming in a response which is proportional to your distance from optimum should speed things up further, even if your control is noisy.

Cameron

[grega]

Hey guys,

thanks for all the info. I've managed to speed up things a bit. To 20 ms, so that should be ok. Cameron I am already taking your advice as I am programming the regulation. Will get back when I get some useful data WITH working regulation.

Greetings, Grega