[time-nuts] Re: PICDIV stability

[ghf at hoffmann-hochfrequenz.de]

Am 2022-01-08 10:55, schrieb Attila Kinali:
> On Fri, 7 Jan 2022 12:32:10 +0100
> Attila Kinali <attila at kinali.ch> wrote:
> 
>> stable (e.g. LTC6655 or use a Jung SuperRegulator that uses an 
>> LM329/LM399 as reference).
> 
> Oops.. wrong chip... I meant the LT3042 low noise LDO here, not the 
> reference...
> Sorry about that.
> 
> (Though the LTC6655 would work as LDO as well, just a bit overkill)


Noisewise, the LT6655 is not that good, although it is remarkable
that a bandgap can enter buried Zener territory.

<  
https://www.flickr.com/photos/137684711@N07/28423518000/in/album-72157662535945536/ 
  >

Also the output current is quite limited.
Filtering can do wonders in terms of noise but is hard to do without
degrading absolute voltage accuracy, which is its very purpose.

in zerms of output current and noise, the LT3042/45 is in a different
league, also in comparision to the entire bunch of regulator chips:

<   
https://www.flickr.com/photos/137684711@N07/24070698809/in/album-72157662535945536/ 
   >

If you want REAL output current, the circuit in the data sheet with the 
power NPN
does not lose much of the noise behaviour. (follow the right arrow)

Somewhere at LT (now AD) is a circuit that uses the LT6655 bandgap with 
the
LT3042 output stage. I think it has some filtering.

cheers, Gerhard