The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> the lower frequency bound for the flicker noise
https://designers-guide.org/forum/YaBB.pl?num=1159839876

Message started by circuitsurfer on Oct 2^nd, 2006, 6:44pm

Title: the lower frequency bound for the flicker noise
Post by circuitsurfer on Oct 2^nd, 2006, 6:44pm

Hi guys,

Can anyone familiar with the flicker noise calculation explain how to
set the lower frequency bound for the flicker noise? The theory I know
is that the lower frequency bound should be inversely proportional to
the cicuit ON time. Because of the random walk theory which gives rise to
the 1/f noise frequency spectrum, the longer the circuit ON time is,
the higher the flicker noise energy becomes. For example, people set an
ADC's conversion frequency as the lower bound in calculating the total
flicker noise. This approach is pretty vague and arbitrary to me. Any
rigorous analysis and theory behind this? I am designing an input buffer
for a sigma delta ADC. I know chopping is one way to modulate the
flicker noise and remove it. But before embarking on that path, I want to
convince myself that I can't live with the flicker noise in the buffer.

I will really appreciate any feedback and opinion.

Best regards

Title: Re: the lower frequency bound for the flicker nois
Post by avlsi on Oct 5^th, 2006, 12:51am

Hi,

I am calculating the flicker noise like this.
Flicker is integrated from the unity gain frequency to low frequency.

This low frequency is set by the operation time of the circuit. If you are operating the circuit for 1 year ,convert this time into frequency.

Then use to calculate flicker noise.

Any suggestions are welcome.

Title: Re: the lower frequency bound for the flicker nois
Post by vivkr on Oct 11^th, 2006, 5:15am

Hi,

You should remember that the total noise power within a decade is constant for 1/f noise.
This means that there is only as much noise power in the band from 0.1Hz-1Hz as from
1uHz-10uHz.

Secondly, the lower frequency bound should be chosen on first order as the longest timeconstant
in your system where the ADC will be used. While 1 minute may be too long a lower bound if you
are designing the chip for a hearing aid, even 1 year may be too short if its a question of observing
astronomical objects ;)

Regards
Vivek

Title: Re: the lower frequency bound for the flicker nois
Post by RobG on Oct 16^th, 2006, 11:08am

To get a feel for the probelm, try the lower frequency at 1/(1 month)=380nHz. Then 1/(1 year)=32nHz; then 1/(10 years)=3nHz. What you will find is a relatively slow increase in the total noise. Since this amp buffers the signal, I think you need to look at as long a period as you are going to operate it. That is, if you expect the same A/D output ten years from now for the same input, you need to look at 1/10 years.

If you are close to the spec edge, then I'd be worried about other effects. For example, packaging stress changes over time and can cause more shift. And changes in temperature "look" like noise. These things can be larger than the 1/f component.

rg