Strange Error amplifier of LDO

Question

I am trying to work out what function the components in inverter colors serve, this is a schematic I have derived from a net-list to make it easier to explain my answer

E1 is an Error amplifier inside a LDO Regulator (VCVS)

Net 3 is connected to source of a P-Channel MOSFET with upper and lower clips. (I think this is unusual as I have found that the output of error amplifier is with respect to the input)

Net 9 is the FB (Feedback) node,

Net 12is the reference voltage node,

2.8V output voltage is set using feedback resistive divider. Has dropout voltage of 220mV for 150mA. Nominal input is 3-5.5v.

Is there a name or key word for this type of compensation?

How does this help for stability, PSRR (both) e.g. If I wanted to perform open and closed loop analysis for placement of poles and zeros ? What would this change?

Answer 1

Those are frequency-compensation networks, so the LDO does not oscillate.

I've also seen such networks (implemented with long_channel FETs, etc) on the feedback path from the voltage divider.

===================================

Notice the 10K ohm on the error_amp output. Those 3 RCs turn on, and STAY ON, as the frequency increases.

Fundamentally you have a variable_ratio voltage divider, using 3 shunting networks.

And don't forget the 10Kohm also has to charge and discharge the gate capacitance of the series_regulation FET.

how to analyze it ? basically, you want a CLOSED LOOP with little or no ringing. Hence you could run a closed_loop response 1Hertz to 10,000,000 Hz, with 10 steps per decade, and look for peaking (indicating ringing).

And you could perform a TRANSIENT load_step of 50%, or a Vin step of 1 volt, and check for little or no ringing even at parameter extremes.

But you should also run an open_loop response, so you get to correlate ALL these various methods and learn some of the theory and the practicality.

One way to run the open_loop was what I showed a switch_reg designer many decades ago. He ended up with the world record for the most personally_designed "boxes" in orbit around the earth {best wishes, Pat}. His employer threw up hundreds of scientific satellites over the last 40 years, and I'd like to think his use of this trick helped keep him happily employed til he retired.

The trick --- see that feedback voltage divider? place a very large capacitor on the output. Now use a much smaller capacitor to inject some swept_sin at your choice of node.