About the design of the OTA

[yeknight]

When I design an OTA,after I compute the W/L of all MOSFETs,but when I simulate the OTA,I found the MOSFETs not working in the Saturati.How to modify the W/L of the MOSFETs in general,could you give me some experence?

 

[ASHUTOSH RANE]

Hi
u just refer the "Tradeoff and Optimisation in analog CMOS design" by Blinkley .....
u will get y ur calculated valu do not work and also how to optimise the circuit to get good performace ...i am reading it now ...

 

[leo_o2]

Mainly it is not decided simply by W and L. The idea is to reduce Vgs and increase Vds for operating point.

 

[yeknight]

when i sim the circuit,M5 is always work in Linear,what is the problem,how should I adjust?

 

[dgnani]

too bad the image cannot be accessed... yeknight can you verify and re-upload?

 

[yeknight]

reupload the image

 

[erikl]

M5 is always work in Linear,what is the problem, how should I adjust?

Your vin potential defines VDS(M5).
If you want M5 to operate in saturation region,
both {vin+,vin-} > Vdsat(M5) + Vth{M1,M2} .

 

[jimito13]

Your vin potential defines VDS(M5).
If you want M5 to operate in saturation region,
both {vin+,vin-} > Vdsat(M5) + Vth{M1,M2} .

erikl,

i think the equation for the M5 transistor to work in saturation is : Vin+>Vdsat5+VGS1 since :

VDS5>Vdsat5 =>
VD5 - VSS>Vdsat5 => (D5 is S1)
VS1 - 0 >Vdsat5 =>
Vin+ - VGS1>Vdsat5 =>
Vin+>Vdsat5 + VGS1

The same can be written for in- input.

Correct me if i miss something or make any mistake.

 

[dgnani]

hi jimito

you're correct using Vth instead of Vgs is an approximation but a pretty decent one since the difference between the two is the overdrive voltage which has to be small to operate the diff pair in a region of high gm/Id

 

[jimito13]

Yes i was suspecting that,i mean that erikl was implying that M1 works with low overdrive voltage.I just wanted a confirmation and to give a more general equation.

 

[erikl]

... e. was implying that M1 works with low overdrive voltage. I just wanted a confirmation and to give a more general equation.

Of course you're right, jimito,
I just wanted to give a rough approximation. For M1/M2 working in a region of high gm/Id = moderate/weak inversion, their Vgs would be close to (or even less than) their Vth. On the other hand -- due to body effect -- they could well need a bit more Vgs.