[sdiy] Filter stability and self oscillation

[Nicholas Gregorich]

Insight into my initial ignorance inline.  ;)

On Monday, May 29, 2006, at 07:13PM, harry bissell <harrybissell at prodigy.net> wrote:

>insightful, perhaps ignorant comments inline...
>
>Nicholas Gregorich wrote:
>> Just finished my second electronics course and feedback (negative) was one of the last subjects we studied.
>>
>> I found it very interesting to learn about negative feedback and finally get some educational insight into what makes a resonant filter...resonate, but some things just aren't adding up and I'd like to request some clarification.
>>
>> 1. Would it be correct to call a filter that self oscillates "unstable"?  Meaning that the -180 degree phase frequency occurs before the unity gain frequency causing the system to be unstable and oscillate at omega -180.
>>   
>beats me. I would call a filter that self-oscillates with amplitude 
>increasing to infinity "unstable" but
>if there is some mechanism to control the amplitude... I'd call it an 
>oscillator :^P

I did forget about the part that a self oscillating filter doesn't increase in amplitude to infinity, rather its either amplitude limited or else the poles are located right on the jomgea axis at full resonance (sounds like an unlikely case to me but I don't know).  As I look at my text again I see in a 2 pole system if the poles are located on the jomega axis they can sustain oscillations but they will not grow in amplitude with time.  Now I guess I have to qualify unstable as having poles in the right hand plane.  I might even venture to say all filters have to be stable or if they are unstable have a limiting circuit included.

>
>> 2. In addition, all two pole systems are stable by nature because their maximum phase lag is -180 degrees but it occurs at an infinite frequency (at which case unity gain would have occurred at some point before).
>>
>> If point 2 is true how do some 2 pole filters self oscillate?  The only example I can think of at the moment is the Korg MS20's self oscillation even though it is reported as a 12dB/oct filter (I've never used one myself).  My only explanation is that it actually not a two pole filter.
>>   
>Who says the feedback is necessarily negative.  Some designs have 
>positive feedback and are almost
>guaranteed to oscillate.
>

Right you are, I assumed all filters use negative feedback.  :(  Positive feedback is not something I have directly studied, only indirectly as in the above case where negative feedback turns positive due to a phase inversion and unity or greater gain (unity in a filter I imagine?).

>The MS-20 (Sallen Key) is ont that is almost guaranteed to oscillate, 
>and needs an amplitude limiter in the feedback.  The 'other' 
>architecture... the 'state-variable' will almost never oscillate... 
>adding feedback
>reduces the resonant peak.  But if you add gain you can force it to 
>oscillate.
>

Interesting about the state variable filter.  It kind of goes against everything I learned this semester, smaller resonance with more feedback.

>H^) harry
>

Thank you for your response.  I was feeling like I had a grasp on some general concepts but now I feel overwhelmed again and see I have much more studying ahead of me.  :/

Nick.

>> That's all I can think of at the moment.  Thanks for any help.
>>
>> Nick.
>>
>>
>>   
>
>
>