bias circuit
Hi wael_wael,
What do you exactly need? Is it the generation of the gate voltages of the current mirrors P and N and the cascodes P and N?
Yeah, that's a good question. Because that is normally not publish, so it's not so easy to find.
Well, there's not a big mystery behind it. You somewhere would need a reference generation block (like a bandgap).
Now, the simplest is to use a current passing it through mos transistors connected in diode mode and changing their sizing. As a rule of thumb, for the mirrors the references will have the same size as the mirrors and for the cascodes, the diode connected transistors will roughly have the same W and 4*L.
There are slightly more sophisticated structures, but I don't know how to get those drawings right now.
I hope your question was in this direction, otherwise, what a lot of writing for nothing :-b
Added after 12 minutes:
n1cm0c,
There are better ways to sense the CM rather that re-using an already active circuit like the input diff pair, but anyway, there are two big problems with it:
* It s very sensible to mismatch. In fact, if you analyze the circuit, you'll see that the current that goes through the cm control transistor is the difference between the total current of the PMOS current mirrors and the NMOS current mirrors. The mismatch between the two can spoil the gain of the CMFB loop which probably is already quite small
* It has dangerous start up conditions. If you force the inputs of the Y amplifier to gnd, the transistor can be cut-off and not able to push up the cm voltage
I have personally experience problems with this type of circuit. It is not such a bad solution because it uses only one transistor but it is very dangerous. People tend to overlook this part of the circuit and it can make your complete system fail