[sdiy] integrator / capacitor leakage

[James Patchell]

I think this problem is being approached from the wrong direction....

Generally, I start out with a set of requirements (a spec), and then find 
out what it will take to meet those requirements.  It doesn't matter how 
you get there as long as you meet those requirements.  Sometimes you are 
surprised at what you can get away with.  The best capacitor in the world 
may cost you big bucks, but if you can meet those requirements with a 25 
cent capacitor, so much the better.  Sometimes we obsess on making 
something really good, when there are better less expensive solutions...

I suspect you are going to have to try several approaches before you find 
the one that will meet your requirements.

         -Jim <why use 1 two by four when 5 will do the same job>

At 10:56 AM 12/3/2005 -0700, Ian Fritz wrote:
>At 09:37 AM 12/3/05, Grant Richter wrote:
>>I think the answer is so obvious, it is hard to see.
>>Any multiplication (amplification) of the integrator output will
>>multiply the errors by the same factor.
>>
>>So, commutative axiom says:
>>
>>(Signal + Error) * 10 = (10 * Signal) + (10 * Error)
>>
>>Since calculus is built on algebra, if you put an integral sign in
>>front of terms, it is still axiomatically correct.
>
>Your argument relies on an assumed linearity of the error. The question is 
>whether that assumption is justified, and if no, how to best minimize the 
>error.
>
>   Ian

         -Jim
***************************************************************
http://www.oldcrows.net/~patchell

***************************************************************