Gain margin - control and automation

[LM741]

hi alll!
just a quick one.

You know when you design a controller - you go and plot the open loop transmission, L(jw) on a nichols plot, where L(jw) = C(jw)H(jw) P(jw)
where P and C are the plant and controller respectively. You then 'manipulate' this plot (i.e. vary your controller) until your design specs are met. My question is once you have a gain margin - what does this actually tell us - say we have gain margin of -5dB.
does this tell us that our controller gain can only increase by 5dB before the system bcomes unstable (assuming the system was already stable).
The thing i don't get is that our controller gain is fixed and doesn't change after we design it! so which gain (the controller or plant or H(hw) ) does the gain margin put a limit on?
think i might be missing something...

thanks!

 

[NCEE]

If the gain margin is 5 dB, this means the open loop gain of the system can only increase by 5 dB before going unstable.
Your controller gain may not change, but a change in the parameters of the plant will throw everything off.
Basically, the higher the gain margin, the less likely a change in systems parameters will affect stability.

 

[LM741]

ok - cool - but these 'parameters' of the plant - what exactly are these and do they keep changing whilst the system is operating? SO the characteristics of the plant CAN change whilst it is operating?
This isn't plant uncertainty is it? - which is why we need a controller to compensate for this uncertainty?
think I've got it - if you could just confirm.
thanks

 

[berkeman]

Anything that varies the transfer function can affect your gain (or phase) margin. Like, the AC mains voltage changing within tolerances in the plant will affect the startup time of AC motors. And where the motor shafts are in their periodic maintenance lubrication schedule will affect startup and operating speed under load.

 

[LM741]

thanks
...