bandgap flicker noise

[amitbhaiji]

Can any one suggest method or some paper to reduce the flicker noise of the Bandgap reference source.

 

[nathan]

Chopper stabilization ;-)

nathan

 

[amitbhaiji]

How to do the chopper stabilization in the bandgap

 

[okguy]

What can be the level order of the ficker noise on top of the bandgap 1.2V ?
(assuming the Figure 11.9 of Razzavi's book
= 3 resistors + 2 PNPs + 1 OpAmp in a standard CMOS technology)

 

[electronrancher]

really? chopper? i would LOVE to see a chopper that did not inject more noise than natural 1/f! 1/f is the random dance of carriers, my friend.. it's tiny. what is reported as flicker is usually measurement inaccuracy, and you know what? that may count!

chopper? even the clock injection from the substrate would be bigger than 1/f!

i understand there is a strong trend towards switched cap. razavi can burn in hell for all i care, although switched cap does have it's place. this paper is minimizing OFFSET thru switched cap, not flicker noise.

OKGUY - my suggestion for you is to reduce flicker would be to increase bias current. swamp out the randomness with strong current flow.. >50uA? >2mA? how much flicker do you consider too much?




PS - the way a chopper minimizes offset is by flipping the + and - terminals of the op amp (or some equivalent) back and forth very quickly. if the op amp has +2mV offset, then half the time it seems like +2v, half the time like -2mV. Suck up the ripple with a cap, and it seems like you have near-0 offset.

 

[rfsystem]

Problem with BG in Std CMOS is the flicker of the PMOS used in the feedback amplifier. It is in the loop of the k*T scaled diode/resistor loop. So the PMOS is already better than the NMOS but the gated lateral PNP would perform better. It is a 5-terminal device. Base/Emitter/Collector/Gate/Substrate. You have to bias the Gate higher than the Emitter. It reduce the filcker also here. But do not use a NMOS mirror at the output. Use resistor instead. Otherwise the mirrors would define your flicker performance. Perorm about 20dB and convert to a conventional PMOS-DifAmp/NMOS-Mirror.

 

[andy2000a]

Hi
I have another problem in flicker noise, as I know , hspice .noise
only simulation thermal noise , about flicker noise , we need spice model support , like KF= ?? but in umc/tsmc 0.35u or .5um process ..
no this item (someone said under 0.25um Fab will provide KF ..)
if we no KF how to know bandgap have large flicker noise ??

thank you

 

[rfsystem]

Sorry,

I does not work for fabless. My fab support detailed modelling. I think for flicker there exist some third party measurements fro KF.

 

[kkoko]

I think that you can icrease size of transistors

 

[xwcwc1234]

With bigger transistor size in input stage , can reduce the flicker noise.