Hello, I want to design a wide-swing current mirror. So I use this architecture as shown in figure. But with wide-swing, Vg1=Vdd-Vt-Vov, Vg2=(Vdd-Vov)-Vt-Vov, which are all function of Vdd, but Vbg is indepenct of Vdd. So this architecture is wrong. Is there any other idea about this even replace this kind of current mirror?
Re: Using Voltage Reference to do a wide-swing current mirro
I think that the opamp output connect a nmos which create a constant reference current, then use this current to create cascode pmos bias(such as the left picture ), which the swing can be maximum.Hope it is useful for you.
This helps me so. But I wonder the purpose of R1 and R2 because N1 and N2 may be pushed into linear region at low Vdd and large current. I think R1 and R2 can force the current between N1 and N2 to be matched more than without them, right?
hi,
this is a V/I the voltage on the Rext is equal to Vbg (provided by the used buffer in a -ve feedback loop)
since the Voltage on Rext is constant then the current passing through it is constant too and equal to Vbg/Rext
so u converted the Vbg into constant current this constant current is passed to the pmos and mirrored for further usage.
regards,
A.Safwat