The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> PSRR
https://designers-guide.org/forum/YaBB.pl?num=1237902567

Message started by raja.cedt on Mar 24^th, 2009, 6:49am

Title: PSRR
Post by raja.cedt on Mar 24^th, 2009, 6:49am

hi,
How to decide what kind of kind noise profile we have to add in the VDD to get psrr.I used to find in the following manner
Give 100mv supply bump with 1/f rise time and keep some time in that state ,and then finding the steady state change in the output voltage,here f is the dominant frequency in the supply noise spectrum.Can any body explain clearly how to find based on supply noise profile?

Thanks,
Rajasekhar.

Title: Re: PSRR
Post by indysjl on Mar 24^th, 2009, 10:02am

Rajasekhar,

In my experience, PSRR is determined through AC simulation. Sweep (frequency) an AC source (Vin) on top of the power supply, and measure output (Vout) signal, then plot Vout/Vin in dB vs frequency.

Hope this helps,

Saul

Title: Re: PSRR
Post by thechopper on Mar 24^th, 2009, 11:23am

Rajasekhar,

What Saul suggested is correct. PSRR is difined a transfer function of a linear system. Therefore you need to evaluate it at each frequency within your noise bandwith. The system therefore will respond according to the PSRR frequency response given a certain noise profile.
This is characterized with AC simulations as Saul said.

Regards
Tosei

Title: Re: PSRR
Post by Visjnoe on Mar 24^th, 2009, 11:47am

In general, I agree with the above, but after the AC simulation you should also run a transient simulation with supply noise to check the PSRR. What guarantee do you have that the supply noise is only 1mV-10mV...not 250mV?500mV? I've been there.

Good PSRR from AC simulations is a necessary but not sufficient condition.

What tones should you apply? What amplitude? Look at your system, look at the surroundings of the DUT! What clock frequencies are running nearby? Do they have substantial harmonics (most likely the answer is yes)?

I've had numerous debates about PSRR. It's the one spec where most customers just 'put a number' without giving it any extra thought. This might lead to unnecessary iterations for the designers. PSRR specifications should be derived from a system study with input from
IC designers.

Regards

Peter

Title: Re: PSRR
Post by thechopper on Mar 24^th, 2009, 1:23pm

Peter: I agree with what you suggested, but I still think that if small signal noise models such us 1/f are considered for the VCC supply line, then AC analysis should be ok for predicting the amount of such noise going to the output of the system.

However I also got burnt with "noise on VCC" being much larger than what a small signal model accounts for. So definitely is something to define as an input for your system before going too deep into the design. A transient then is the answer for analysis such responses as you stated.

Regards
Tosei

Title: Re: PSRR
Post by indysjl on Mar 24^th, 2009, 3:48pm

Peter/Tosei,

Thank you for supplementing my answer. In my limited design experience I have not performed transient analysis in characterizing PSRR. Do you have any recommended literature references that I might peruse to become more familiar?

Thanks,

Saul

Title: Re: PSRR
Post by raja.cedt on Mar 24^th, 2009, 9:34pm

hi ,
thanks everybody for giving valuable reply ,even i feel the same ,they have to give some other specifications along with PSRR no,...can any one please give clear PSRR specifications for a model supply noise profile? or can you plase direct any data sheet or documents regarding this?

Thanks,
Rajasekhar.