Designer’s Guide Community

subgold

Community Member

Offline

Posts: 97

a question about bandgap reference
Aug 26^th, 2009, 6:06am

let's say i have a typical bandgap voltage reference as shown in the attachment. assume the resistors and the amplifier are noiseless, only the bipolar transistors contribute noise, and their corresponding collector noise current are i1 and i2, for Q1 and Q2 respectively.

then what is the noise appear on the output Vout? what is the noise transfer function for the bipolar noise?

thanks for any help.

Back to top

bg.JPG

IP Logged

andre75

New Member

Offline

Posts: 7
Sunnyvale, CA

Re: a question about bandgap reference
Reply #1 - Aug 26^th, 2009, 3:31pm

Superposition: For each noise contributor, assume an equivalent noise source (eg. vn^2=4kTR for resistors) and calculate the transfer function while setting all other noise sources to zero (voltage zero, current open). Depends what you want to include (e.g. thermal noise of resistors and transistors) you can find a transfer function for each of the sources. The total noise at the output is simply the superposition (addition) of all individual noise sources.
In general I would use Spice, find the largest contributors and optimize those.

Back to top

Andre
http://www.aguntherphotography.com

IP Logged

subgold

Community Member

Offline

Posts: 97

Re: a question about bandgap reference
Reply #2 - Aug 27^th, 2009, 12:39am

andre75 wrote on Aug 26^th, 2009, 3:31pm:

Superposition: For each noise contributor, assume an equivalent noise source (eg. vn^2=4kTR for resistors) and calculate the transfer function while setting all other noise sources to zero (voltage zero, current open). Depends what you want to include (e.g. thermal noise of resistors and transistors) you can find a transfer function for each of the sources. The total noise at the output is simply the superposition (addition) of all individual noise sources.
In general I would use Spice, find the largest contributors and optimize those.

hi andre,

thanks for the reply. however, it is actually the very question that i asked: how to find the transfer function from a certain noise contributor to the output? spice simulation doesn't tell me the transfer function, which i need for modelling before the real design.

i mean i can derive a transfer function, but i am not sure if it is correct. it would be great if you can show me an example.

Back to top

IP Logged

Forum Administrator YaBB Administrator Offline Posts: 145	Re: a question about bandgap reference Reply #3 - Aug 27^th, 2009, 9:00am The xf analysis in Spectre will do this for you. Just identify the output and specify "stimuli=sources" as a parameter to the xf analysis. It will then compute the transfer function from every node to the output. The transfer function is computed assuming that a current is injected into the node. -Ken
Back to top	IP Logged

subgold

Community Member

Offline

Posts: 97

Re: a question about bandgap reference
Reply #4 - Aug 29^th, 2009, 7:33am

Forum Administrator wrote on Aug 27^th, 2009, 9:00am:

The xf analysis in Spectre will do this for you. Just identify the output and specify "stimuli=sources" as a parameter to the xf analysis. It will then compute the transfer function from every node to the output. The transfer function is computed assuming that a current is injected into the node.

-Ken

thanks for the reply, sorry but i still dont get it.
you mean i can get an analytical (NOT nummerical) transfer function from the xf simulation?

Back to top

IP Logged

Forum Administrator YaBB Administrator Offline Posts: 145	Re: a question about bandgap reference Reply #5 - Aug 29^th, 2009, 6:15pm No, it is numerical. It is similar to running an AC analysis. -Ken
Back to top	IP Logged