[sdiy] LT1228 Lm13600 crap

[René Schmitz]

Hi Ian and all,

Ian Fritz wrote:
> Thanks for pointing that out.  Looking at the 13700 vs the 13600 
> schematics, it appears that they just erased Q3!  Do you think the 
> current theft could be eliminated by tying the buffer input to the 
> negative rail?  I'm not sure how simplified these schematics are.

Maybe one could just blow out that extra transistor. :-P

> One further question, though.  Isn't Q3 just an additional 
> current-mirror leg?  I would have thought that any error would just be 
> Q3's (small) base current.  Also, shouldn't this be a linear error to 
> first order, assuming beta doesn't change a lot with current.  How does 
> Q3 produce a >5% nonlinearity?

I haven't done the maths for it, however I think its not that simple. 
The question is how the current distributes into Q3 and D1.
Here Early effect and of course Rbe could play into the distribution.

Its also not an exact third mirror leg. They did "tap" current out of 
the "primitive mirror" Q1 and D1. But this means debalancing the wilson 
arrangement arround Q1, Q2 and D1. Its difficult to grasp this 
mathematically, because one doesn't really know the relative geometries.

Cheers,
  René

-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159