LMP7721: Opamp simulation with output load

So guys, this is another case where the method shown is isolating the interaction between the open loop zout and the feedback network, if you do it the way I describe in this article it becomes 50deg phase margin - not 62deg. probably still ok, but quite different - if I am setting this up right

attached plot and TINA file for this, the article also shows how to sim open loop Zol

LMP7721 LG sim march14.docx

LMP7721 Zt LG updated March14.TSC

Hi Michael,

I used the brand new LMP7721 GWL simulation model for the stability analysis. It is listed available from the LMP7721 web page > Tools and Software > Models. It is dated 27 Feb 2019.

That revision was done to bring the model up to date and exhibit the correct Aol and Zo characteristics.

Regards, Thomas

Precision Amplifiers Applications Engineering

Thanks Tom for rhe latest model, with that I was able to run some modeling testing against old and new to extract what is going on - wow!!! Incidentally, this is coming from my 30years of intermittent modeling work on higher speed op amps and FDA;s. 

This discussion is mainly for Kai and you Tom, the 2008 NSM model had NO input capacitance - that means any transimpedance work done earlier using that model was going to be a bit off. The data sheet is not super clear but on page 18 says it has 11pF input C - that has to have come from the designer I expect.

The updated model appears to have 15pF Ccm and 5pF Cdiff (way different than the11pF number on page 18). We did quite a bit of remodeling and testing a bunch of the HSP amps some years back and I recall one conclusion for most parts was the Cdiff is very small being series combination of device parasitic C. The attached TINA file eventually shows how to adjust the new model to 1pF Cdiff and 10pF Ccm - if I am doing this correctly, this would be preferred for transimpedance design work. 

One lesson (again) is to be very sure what is in the model before going too far, not hard to test, what is harder is reconciling to PDS measured plots representing nominal silicon to gain confidence in the model.I would say the new model is probably closer to silicon in Aol and Zol, but a bit off on input C.Those are the key elements in LG phase margin. 

Now, I may go and test the different LG phase margin approaches and see if there is any difference - often, with just a resistive Zol there is not much difference, but - if the Zol is reactive, the newer approach I put into that article can shift quite a bit - and I think that is more accurate to have the Zol see the feedback network and inverting input C. 

Testing the old against the new LMP7721 TINA models.docx

And then the final TINA file with adjusted input C's. 

LMP7721 new model adjusted Cin.TSC

LG phase margin testing with updated LMP7721 model with correct 11pF input C.docxSo I here I tested the LG each way, almost no difference with that low 20ohm Zol. And then the closed loop which looks real good.