wandola
Junior Member level 3
Hello All,
I have some questions about cmos comparator design.
I need to design a very low power comparator. The Vdd = 3.3V, Midband Gain ~55dB, GBW ~500KHz. The current needs to be as small as possible.
I plan to put the differential pair in subthreshold region and the rest of the transistors in strong inversion or moderate inversion.
My architecture is samilar to a conventional 2-stage design. 1st stage is a PMOS current source. Then there is a PMOS differential pair. Below are NMOS load/current mirrors. 2nd stage is following the 1st stage by a wilson current mirrow. It is a cascode structure with 2 PMOS on top.
I have the following questions.
(1). How to determine the gain bandwidth of a comparator? Since comparator is not compensated...
(2). For my 0.18um process, the minimum transistor width is 220nm. So when I try to set the biasing current to about 300nA and I set my PMOS in saturation (Vgs = 500mV, Vth ~ 450mV), I get a W/L ~0.6.
Then I have to set the W/L = 300n/500n.
Is it acceptable in analog circuit design??
(3). For the NMOS active load devices, in order to bias them in saturation, I also have to make W/L < 1 since the current is very small...
So will this be okay??
thanks for the help.
I have some questions about cmos comparator design.
I need to design a very low power comparator. The Vdd = 3.3V, Midband Gain ~55dB, GBW ~500KHz. The current needs to be as small as possible.
I plan to put the differential pair in subthreshold region and the rest of the transistors in strong inversion or moderate inversion.
My architecture is samilar to a conventional 2-stage design. 1st stage is a PMOS current source. Then there is a PMOS differential pair. Below are NMOS load/current mirrors. 2nd stage is following the 1st stage by a wilson current mirrow. It is a cascode structure with 2 PMOS on top.
I have the following questions.
(1). How to determine the gain bandwidth of a comparator? Since comparator is not compensated...
(2). For my 0.18um process, the minimum transistor width is 220nm. So when I try to set the biasing current to about 300nA and I set my PMOS in saturation (Vgs = 500mV, Vth ~ 450mV), I get a W/L ~0.6.
Then I have to set the W/L = 300n/500n.
Is it acceptable in analog circuit design??
(3). For the NMOS active load devices, in order to bias them in saturation, I also have to make W/L < 1 since the current is very small...
So will this be okay??
thanks for the help.