What is the variation expected in Current mirror due to MOSFET mismatch

[Venkatlab]

Hi All,

What is the variation expected in Current mirror(Simple Current Mirror) due to MOSFET mismatch? what are the other parameters that can affect the accuracy?
For example if a current mirror is designed for 10mA. How much is variation expected due to MOSFET mismatch?

 

[crutschow]

Re: variation expected in Current mirror due to MOSFET mismatch

If the MOSFETs are in an IC then it entirely depends upon how closely the devices are matched in the IC fabrication process being used.

If the MOSFETs are separate discrete devices than the match could be quite poor, since it depends upon the difference in threshold voltage (Vth) between the two, and that can be quite large.

 

[Venkatlab]

Re: variation expected in Current mirror due to MOSFET mismatch

If the MOSFETs are in an IC then it entirely depends upon how closely the devices are matched in the IC fabrication process being used.

If the MOSFETs are separate discrete devices than the match could be quite poor, since it depends upon the difference in threshold voltage (Vth) between the two, and that can be quite large.

Hi Zapper,

With Wilson Current mirror(MOSFET) can the accuracy of the current mirror be improved compared to simple current mirror(MOSFET)?

 

[crutschow]

Re: variation expected in Current mirror due to MOSFET mismatch

A Wilson mirror will be less affected by changes in operating voltage but has no improvement due to transistor mismatch errors.

 

[Venkatlab]

Re: variation expected in Current mirror due to MOSFET mismatch

Hi Zapper,

How to find whether these MOSFET is perfectly matched or not?
Part no: Si7956DP

 

[an_82]

Re: variation expected in Current mirror due to MOSFET mismatch

Wilson CM is the best CM rather than others, in term of accuracy. The mismatch can be due to w/l ratios, threshold voltage, channel length modulation, etc.
To find the mismatch you can simply use simulator software, e.g. Hspice, simulate the circuit and then see the .netlist.

 

[Dominik Przyborowski]

Re: variation expected in Current mirror due to MOSFET mismatch

Wilson CM is the best CM rather than others, in term of accuracy.

Simple degenerated cascode CM is better. Everything depends to technology and design constraints.