[sdiy] expo accuracy? or integrator accuracy?, or both?

[Magnus Danielson]

From: René Schmitz <uzs159 at uni-bonn.de>
Subject: Re: [sdiy] expo accuracy? or integrator accuracy?, or both?
Date: Fri, 07 Feb 2003 16:43:49 +0100

> At 16:13 07.02.03 +0100, Magnus Danielson wrote:
>  
> >> I think it is futile to concentrate on a few aspects of VCO design. 
> >> You always will have to look at the whole picture so to speak.
> >
> >This is the point I was trying to make, you have to identify the main flaws,
> >not just a few flaws and attempt to optimize them. Naturally, some flaws can
> >compensate each other, but that may need carefull balancing and rarely they
> >really compensate each other very good, just lowering the error over some
> >range.
> 
> Exactly, thats why I wanted to stress this point again!

Right, OK! ;O)

> At some point of your optimization you might introduce other errors again. 

Indeed!

> (Like what I said about the reset, I experimented with MOSFETs as discharge 
> switches, because they have low Ron. But then I found that the opamp integrator 
> I was using wasn't fast enough for the resulting reset. One would have had to 
> look into leakage again for the MOSFETs, but I didn't get that far... 
> The reset issue is tricky IMO, because the fastest OPAMPs have generally the 
> higher bias currents. The blanket is always too short, so you have to make 
> compromises. )

I especially like the blanket analogy! ;O)

But, it's true and the blanket analogy is a good one. Regardless of the color
of the paper you are taping over the hole in the hull, it's just a paper over
the hole and it will brake for any serious pressure regardless of which
quadrant you'r in.

> Or it might be that as you eliminate one error (which was previously the dominant 
> error) other errors which were previously masked are uncovered. In this situation 
> you are lucky, because you can focus on the next error and eliminate that.

Indeed, but my experience is that you only have one or a few of those clear-cut
cases, but then you run into a whole bunch of errors in about the same size, so
it becomes less obvious.

> I think an "error budged", like proposed in the Chapter on precision Design in 
> The Art Of Electronics is a good method. 

Indeed. It's a very good reading and it gives you a pretty good insight into
the method. Go, read, execute!

Cheers,
Magnus