[sdiy] Expo converter tempco placement

[René Schmitz]

Am 09.07.2014 20:37, schrieb Neil Johnson:
> Justin Owen wrote:
>  >> Can anyone give any insight into the differences between the
>  >> two main locations for tempco placement in expo converter circuits?
>  >>
>  >> Seems like the schems I have are divided fairly evenly
>  >> between those who use a tempco (generally 2K) in the feedback
>  >> loop of the CV scaling op amp that feeds into the base of the
>  >> expo pair and those that use a tempco (generally 1K) as part
>  >> of a voltage divider at the output of the CV scaling op amp.
>  >>
>  >> Secondly, if you were using a tempco int he feedback loop of
>  >> the CV scaling op amp - would there be any problems using
>  >> that tempco in parallel or series with another standard R to
>  >> achieve a certain R value?
>  >>
>  >> Thoughts?
>
> I think there are various reasons.  First that comes to mind is that
> early tempco resistors tended to be wire-wound resistors, with a
> correspondingly high inductance.  You don't want an inductor in the
> feedback circuit of an op-amp unless you want to make an oscillator!
>
> Another reason is simply down to engineering.  If you can get 2k tempcos
> easier/faster/cheaper than 1k then, notwithstanding the inductance
> issue, you'd design the circuit around the available components.

Yes, sure stocking and sourcing components are probably the main 
motivations about the choices here.

I'd like too add, that if you start to design your scaling summer with a 
100k input resistance, then using a 1k in the feedback doesn't leave you 
with enough drive for the ~18mV per octave required at the base of the 
expo pair. You would get 10mV only. So you do have to use 2k (>1.8k) 
with some slack to trim down. Thats why you won't see 1k used in this 
position. (Unless you cut the input resistors down.)

If you move the tempco to a divider after the summer, you have more 
freedom gain wise. Because you can choose the gain of the first summing 
stage, and the reduction on the voltage divider. By which you can 
additionally reduce the offset-voltage of the opamp.

Leaving the inductance out of the picture, because modern tempcos are 
usually not wirewound anymore. (I for one have used schemes involving 
KTY-81, NTCs, PT1000s and thin film tempco resistors instead...) And I 
would guess that the inductance of the old wirewounds wasn't that high. 
You might only have to add some extra feedforward compensation for it to 
keep the opamp stable.

(I could imagine that the original resistors might have been wound in a 
inductance reducing manner, i.e. winding two windings counter wise so 
that the inductance cancels, but I am not sure if that was the case.)

> Finally, there's the blinkered-vision explanation.  If a designer has
> always thought along one track then they'll adapt all problems to fit
> that one solution...

You mean like throwing 2164s at everything :P

Cheers,
  René

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