Yea, you have your DC bias at ground for a ground negative supply, the output does not operate with no headroom - here is a little DC bias added
Here is back at ground with the output saturated into the supply rail, doesn't look too good
You can't ask the output to approach ground, these RR outputs all require some headroom - often, the input can go to ground but in unity gain that is not the limit,
The simulated gain of 1 SSBW often does not match "measured" data sheet plots for several reasons,
The red curve is what we are looking at here, about 6MHz SSBW gain of +1. But, this has 10kohm in parallel with 100pF load defined - that CLoad is often used to extend the bandwidth by reducing the phase margin,
If I rerun your circuit with that load, I get this, that is really off, model must have very poor Aol match to the device, but actually, this is supposed to be a 2MHz GBP part, at noise gain of 3, that measured plot looks way off. Should be <1MHz. Might have put in a curve from a different part. not much change here, about 500kHz F-3dB.
If I run a LG on this ng of 3 circuit, I get this, 500kHz LG=0dB xover with 90deg phase margin? Should be about 500kHz closed loop as the closed loop sim with some DC bias shows. Yes, the measured figure 8 might be from a different part.
Hello user6349971,
When I simulate the OPA188 open-loop gain (Aol) vs frequency using a little different circuit arrangement than Michael used the response agrees closely with the datasheet Figure 7, Open-Loop Gain and Phase vs Frequency graph. The 2 MHz gain-bandwidth product frequency is obtained and the phase margin looks very close.
The Aol simulation circuit was set up with the VCM set to a mid-scale level of +2.5 V, to place the OPA188 input and output at one-half the +5 V supply voltage. The OPA188 output was loaded in the simulation circuit with the 10 kilohm resistor and 100 pF cap to ground as used for the Figure 7 graph.
The OPA188 simulation model is accurate and it includes the correct open-loop output impedance (Zo) vs frequency which is needed for accurate ac simulations. You will need to make sure the OPA188 sees a good circuit representation for the buck converter load if you want good results.
Regards, Thomas
Precision Amplifiers Applications Engineering
Actually, the ideal SSBW extension from LG=0dB crossing for phase margin <60deg is about 1.6X. That, plus a LG sim technique that does not isolate the open loop output impedance from the feedback network, is in this article,
So it would appear this OPA188 plot might be in error in summary.
