OPA2354: GBW and -3dB small signal bandwidth difference

Well Harson, this topic comes up explicitly or implicitly quite often. Essentially, the GBP idea is a super simplification that might have been valid for way overcompensated (phase margin>80deg) type parts - circa 1970's. It never has worked worth a darn at low gains for higher speed parts. 

Unity gain has its own issues (feedthru particularly). 

In essence, if your LG crossover is happening with <60deg phase margin, you actually get about a 1.6X multiplication from that Fxover to the F-3dB (with a varying amount of peaking along with that). Also, the reported GBP in the data sheet is often quite different than the true GBP (Aol =40dB frequency * 100X) in the TINA model. 

For the OPA354 model, my parametric design tables show it has a TINA GBP of 120MHz (that might be a downlevel model at this point - nope, that 2011 one I have is the most current). 

Let's look at the OPA354 2011 model (what is still in the current web folder)

At one time or the other I have tested all these models for what they do, here is the Aol curve showing 40dB Aol at 1.21MHz or a 121MHz true GBP

Well darn, the cut and paste is not working again - I will continue with no TINA sim inserts. 

At Aol = 0dB, the phase margin is 63deg at a a 105Mhz xover. According to the (F-3dB/Fxover) vs phase margin plot in this article, that should give us an estimated closed loop F-3dB of 1.57*105Mhz = 165Mhz. 

https://www.planetanalog.com/why-is-amplifier-gbp-so-confusing-insight-12/#

Then running a TINA gain of 1 with that spec 1kohm load gives us a SSBW of 165Mhz - exactly matching theory - how about that? 

If there are measured plots in the PDS showing higher, that 250MHz number at gain of 1 is likely specious. I ran into that characterizing the CLC110 buffer - essentially you have direct feedthru at gain of 1 making it look like you have wider bandwidth - that shows up on a step response (if your equipment is fast enough) as an initial quick spike than the rise time goes on to match the actual SSBW numbers. 

Too bad the cut and paste is not working, but you get the idea. 

Incidentally, I went on to run a closed loop gain of +10 with the specified 604ohm feedback, shows exactly the 12.1MHz expected from a 121Mhz GBP with the >80deg phase margin at the higher gain. 

Still no insert image working, but that is what the TINA sim shows, 

Hi Harson,

To paraphrase Michael, the second pole of the device is close enough to reduce the phase margin at unity gain, so the first order approximation would no longer apply.