Using the dynamic biasing, we think we can get the LDO to run on about 100 nA of quiescent current in normal operation. That means the LDO could operate for about 3,000 years with two AA batteries (provided you have a 3V to 1.8V switching regulator attached 😄)

Paper: ieeexplore.ieee.org/document/572...

For reference: a Low Dropout Voltage Regulator gives you a clean voltage rail at an adjustable value. They usually have big external caps for stability but that’s wack so we’ll do w/o. Instead we’ll use a differentiator to make a “fast loop” and we’ll crank up the error amp gain. Refer below

Update: I think it went pretty well. The block diagram question oddly tripped me up but everything else seemed almost too easy

Other fun facts about this circuit: It gets monster gain(>100 dB). The GBW is comparatively small. It is more power efficient than a class A amplifier.

Update (3/?): I spent about 8 hours trying to work through this problem without realizing my mistake. Lessons learned? Don’t procrastinate. Ask for help when you get stuck (like really stuck). Pursue ALL possibilities. Always double check connections. Get 8 hours of sleep.

Update (2/?): It turns out I had connected the bulk of the NMOS devices to VDD (high supply) on accident. I’m guessing this was either allowing current to flow from bulk to source or the simulator was giving me artifacts.

Update: the issue was that I couldn’t keep my input transistors (M1T) in saturation. I have a current mirror biasing them. So you can imagine how flummoxed I was when they were in cut-off but the current mirror was happily pulling current. Seemingly, this current was coming from nowhere.

Also fun little memory that I recalled today. The plot shown is in radians so you should divide by 2*pi to get Hz (typical frequency measurement). I know to check for this b/c my prof last semester lambasted us for 10 minutes on that point before an exam (which I proceeded to bomb)