Switched-capacitor Bandgap Reference-Switch Sizing

Thread Starter

huntersmile

Joined Jun 10, 2017
34
Hello everyone!

I am now designing a switched-capacitor (SC) bandgap reference, but I am having doubt on the switches size that I should use.

Let say I do not have any issue with settling time and my op-amp is good enough about slew rate and response time.
Regardless of how is the structure of my SC bandgap, I am just asking a general question.

What kind of size I should set for my switches? Is it okay to put the minimum size of W/L?
Let say my process is 0.18um, can I put the switches' size to be all 0.18um/0.18um?

Let say I use all switches to be NMOS.
The reason why I want minimum is to minimize charge injection, because my charge injection is still very bad affecting my node voltages when the switches are turned off. (even with that old-school book of using half-sized NMOS with S+D connected.)
I noticed that when I put that 0.18u/0.18u size to all my switches, my deltaV as impact of switching on/off at my nodes are reduced.

When I say reduced, it is still 10-20mV change (higher or lower) in node voltage when the switches turn off (even with minimum size). If i use higher W, that deltaV could be in hundreds of mV range. Even this 20mV is kind of unacceptable since I am designing bandgap reference.

Would any Master here give me advice please? Is it okay to use minimum size of NMOS to all your switches in your SC circuit design? Because my friend is not convinced that this kind of minimum size NMOS switches are used in the real practice by others.
Anyone know any reference (research paper or something) that people design SC circuits with minimum size NMOS?
So far all the papers that I am reading, they don't talk about using minimum size of switches at the SC circuit design.
I would highly appreciate your kind help and answer!

Thank you!
 

RichardO

Joined May 4, 2013
2,271
I have never heard of this kind of reference. Can you show me your circuit so I can know what you are talking about?
 

RichardO

Joined May 4, 2013
2,271
So, is the idea to apply two different currents to the diode-connected transistor and use the differential voltage as your reference?
 

RichardO

Joined May 4, 2013
2,271
So, any idea for my questions please?
Only more questions.

I am curious about other aspects of the design that may impact the answers to your questions, however.

What are the values of the current sources?
What is the ratio of the current sources?
In what sequence are the switches activated?
What is the ultimate reference voltage desired?
What are you going to do about the temperature dependence of the reference voltage?
 

Thread Starter

huntersmile

Joined Jun 10, 2017
34
Only more questions.

I am curious about other aspects of the design that may impact the answers to your questions, however.

What are the values of the current sources?
What is the ratio of the current sources?
In what sequence are the switches activated?
What is the ultimate reference voltage desired?
What are you going to do about the temperature dependence of the reference voltage?

Awww...
nevermind then if you are not able to answer my questions :)
 
Top