OPA192: Bandwidth calculation

Hello Kuramochi, 

Yes, it is very confusing, I tried to discuss here - 

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

Hi Kuramochi-san,

In addition to Michael's great article, here is another way to look at the unity gain Bandwidth subject. Please see the attach article on the topics, which is in Japanese. Unfortunately, I do not have a copy in English. 

OPA192 BW is measured at 9.5MHz and shown phase margin of 65.8 degree at unity gain, see attached Tina simulation with small signal injection Loop-Gain Method. 

OPA192 Loop Gain 02142020.TSC

Small Signal Injection Loop Gain in Japanese.pdf

Best,
Raymond

Thanks Raymond, 

Yes I mention that other technique in the references of the article below - it is very thorough but a little more cumbersome - 

I did find a OPA192 Aol TINA file, the way I do it you have to add the Ccm+Cdiff at the feedback node - this set up is unity gain so that cap looks like a load to the open loop output Z. Anyway, sims to 65deg phase margin with LG=0dB at 9.4MHz. Here, Vlg meter is rotated to report PM directly. I also use those odd LC values to help reduce numerical chatter we see on some models. 

Getting the phase margin is the 1st step, what has been missing is the bandwidth extension when phase margin <90deg. I worked this out about 2years ago now (and it seems to work pretty well) but have never published that derivation. 

So say this is 1.57X at 65deg phase margin, that would predict an F-3dB = 9.4MHz*1.57 = 14.8MHz - pretty darn close to your 14.6MHz unity gain simulation. Maybe that 10kohm load I have is the difference. 

Anyway, this originally showed up in part 5, 

And here is the TINA file for the LG sim, (essentially, the gain bandwidth product idea is too simplified to predict low gain F-3dB). 

OPA192 Aol test.TSC

Hi Kuramochi-san,

I am going to close this inquiry. If you have any other questions, please let us know.  

Best,

Raymond