Nested feedback/ composite amplifier opamp requirements(unexplained results)

[777]

If you don't have any local parasitic oscillation, then maybe datasheet of the OPA1642 at 21, 22 pages, chapter 8.1.4. can help you.

 

[JohnYang1997]

If you don't have any local parasitic oscillation, then maybe datasheet of the OPA1642 at 21, 22 pages, chapter 8.1.4. can help you.

There is a 1kohm pot in the front, idk if there's anything wrong with source impedance.
If so, it still won't explain. Opa1612 has better distortion figure than opa2156 and it doesn't explain njm2068 performs better. Local oscillation wouldn't allow this good distortion, would it?

 

[777]

In a composite amplifier, the driver must to be slow and output stage very fast. If the driver is fast, then could apear oscillations and you must to over-compensate. You must to see step response of the entire amplifier with different loads, resistive, capacitive, etc. Until you do not know how much stable is your amplifier, you will not know the truth. What is the bandwidth of your entire composite amplifier ? Maybe is too large and must to reduce it.

 

[JohnYang1997]

In a composite amplifier, the driver must to be slow and output stage very fast. If the driver is fast, then could apear oscillations and you must to over-compensate. You must to see step response of the entire amplifier with different loads, resistive, capacitive, etc. Until you do not know how much stable is your amplifier, you will not know the truth.

In simulation, it's stable at 10n capacitive Load. It's designed to have very high phase margin due to the use of T compensation network. I'll try to use square wave and smaller capacitors(220p 560p) to see if there is any behaivour problems. But I highly doubt that because two 220p in series is still high enough to slow the first stage down and the second stage is CFA which is much much faster.

 

[777]

Maybe you must to try a little gain. How look in simulation the bandwidth ? Linearity plot for closed loop gain ? Here is for opa2134 without any gain and LT1028 in the same circuit.

 

[JohnYang1997]

Maybe you must to try a little gain. How look in simulation the bandwidth ? Linearity plot for closed loop gain ? Here is for opa2134 without any gain and LT1028 in the same circuit.

 

[777]

The bandwidth of the closed loop it is too large, about 25Mhz(-3dB) at resistive load, without any capacitive load. The loop look stable overall but it is a lottery over 1Mhz where the parasitic components play an important role. It is mandatory to see a step response with a such large bandwidth. A step response in a simulation and the most important, in a reality. For example AD797 it not simple to corectly implement in reality. Not even 5532 isn't very docile op-am if you want maximum performance from it. The distortion from the high frequency is in directly relation with the stability of the amplifier. Try to reduce the bandwidth below 1Mhz with a perfect step response and then do it again the THD measurements. I don't know, maybe put some gain in the closed loop and then implement an active filter.