How to size transistors for cascoded current mirror?

[surreyian]

cascoded current mirror

hi,

designing cascoded currrent mirror, how do we size the transistors?
i usually see (for NMOS current mirror) they usually have a large L at the bottom, why is it so?
having a large L will cause the vgs to increase, arent we trying to get a smaller vgs?

and, in cascoded mirror, the top transistor is the one affecting the mirroring effect?

 

[safwatonline]

Re: cascoded current mirror

hello,
being a newbie in analog design dont take my words as FACTS, anyway i think that large L is to improve mirroring (matching) as the current is directly prop. to (1+λVds) and u cannot make sure that Vds is the same for the two MOS , so we tend to decrease the effect of this term by decreasing λ (i.e. increasing L), also as L increase the o/p resistance increase which is also wanted " ideal current sources infinte R".
bye.
a.safwat

 

[Humungus]

Re: cascoded current mirror

When designing cascoded current mirrors, the important thing is the output swing. Once you have determined the minimum allowable output voltage, size your transistors as to have an equivalent saturation voltage that satisfies your needs.

Actually, you can bias the mirror transistors to be in weak inversion as the W/L ratio increases faster than gm/Id. Also, use non-minimum L (~3-4 times minimum L will be OK) to reduce the posible mismatch of the Vds of the mirror transistors due to the Vgs mismatch of the cascode ones.

 

[obi1]

cascoded current mirror

if L rise , Ro rise and matching is better, but you consume more silicon beacause you must grow W.

 

[holddreams]

Re: cascoded current mirror

Page 257 to page 259,Ken Martin's book:

for the wide-swing cascode current mirror,the L of the bottom transistors are less than the top.

I have seen some design circuits, the L of the bottom transistors are larger than that of the top transistors.
I get confused.

 

[laglead]

Re: cascoded current mirror

hi,

designing cascoded currrent mirror, how do we size the transistors?
i usually see (for NMOS current mirror) they usually have a large L at the bottom, why is it so?
having a large L will cause the vgs to increase, arent we trying to get a smaller vgs?

and, in cascoded mirror, the top transistor is the one affecting the mirroring effect?

Large L transistor has small gm which can improve matching and has low flick noise.
Small L has big gm which can boost the current mirror's output resistance.

 

[safwatonline]

cascoded current mirror

hello laglead,
i didnt get what u meant by decreasing L to get high gm
in order to get higher Rout , i think that increasing the L will increase the Rout and if u are talking about cascode then to increase the gm u may use large L along with large W.
regards,
a.safwat

 

[laglead]

Re: cascoded current mirror

hello safwat,
Of course increasing L while keeping aspect ratio constant can get high Rout.
But it is area consuming in most cases. Generally the smallest length in a specific process such as 0.18 or 0.13um is used in upper N transitor in the cascode current mirror. Then large w of this transistor can increase gm. The upper transistor then can boost the lower long-length transistor Rout by gm*r0 times.
Choice of small L of the upper transistor is mainly due to the concern of area.

 

[montage3000]

cascoded current mirror

at bottom and top the large size to reduce offset

 

[rajharvijay]

Re: cascoded current mirror

hey guys this is the procedure of design of cascode current mirror...
i hope you would like it....
please email me back if you have any thing to say....


and ya thanks always welcommed...

bye

 

[turtlewang]

Re: cascoded current mirror

hey guys this is the procedure of design of cascode current mirror...
i hope you would like it....
please email me back if you have any thing to say....


and ya thanks always welcommed...

bye

thanks very much!

 

[numberto]

That's Great!
Thanks

 

[hujiaomianhao]

hey guys this is the procedure of design of cascode current mirror...
i hope you would like it....
please email me back if you have any thing to say....


and ya thanks always welcommed...

bye
Attached Files
eee523proj1_1779.pdf (233.5 KB)



thanks for your sharing