OPA4376-Q1: How to measure the bandwidth of OPA4376AQPWRQ1 with peripheral circuit

Hi Pin,

I would connect a sine generator at the left of R428 and a scope at the output of OPAmp. Insert a 100R resistor between the output of OPAmp and the scope probe, though, to prevent instability :-)

Kai

Also Pin, 

Make sure you use a very low output swing for test - the SSBW is small signal, often folks run into slew limiting which gives a false reading 

A swept sine wave generator will certainly work, most of the time for datasheet plots, we use network analyzers where I/O impedance matching is a concern. I believe I showed a detailed test setup in the OPA837 datasheet. 

Hi Pin,

I think that your measurement circuit looks like the one below, where Gain=3. According to the datasheet, Aol gain is 5.5MHz. If we use Gain Bandwidth Product rule, the closed loop BW is predicted at approx. 5.5MHz/3=1.83MHz. The Aol BW of 5.5MHz is a typical value, and the closed loop BW of the circuit should be close to 1.83MHz. 

/cfs-file/__key/communityserver-discussions-components-files/14/OPA376-E2E-AC-Analsysis-01252021.TSC

Q: Do you have any instruction that can share with my customer and me to study?

In addition to Kai and Michael's comments, you need to use network analyzer to inject small signal as shown in the simulation above. The actual BW test parameters are shown on top of datasheet page where GBW of 5.5MHz is defined. For the small signal injection technique, 100Ohm and 1MEG/10pf probe were not required. The 100Ohm is needed if you are trying to estimate the BW with the following method. 

Normally you can use the generator and scope to estimate the BW of a circuit. You can input a small signal (take into account BW and slew rate of op amp at ~2MHz), say approx. 1mV in amplitude, and sweep across the frequency range and looking for -3dB point (0.7079*Vin) at an op amp's output. Once -3dB point is found, the obtained figure * Gain(3) will be approx. to the closed loop BW of the circuit.  

In this case, the above method will not be able to estimate the BW, since there is a Q presented near Op Amp's unity gain BW (it looks like that there is a second pole near the unity gain BW ). You can tell the phase margin of the circuit is approx. 63 degree. There will be an error if you estimate the BW of the circuit using the amplitude method (searching for -3dB point), because the Q will skew the actual -3dB point (push -3dB measurement point to higher frequency). 

If you have additional questions, please let us know. 

Best,

Raymond

Well Raymond, since you had drifted off into this bandwidth extension idea, I thought I would test it against a development I had done about 3years ago now and put into a stability article, Here is a LG sim for the circuit you showed (looks odd, but valid for gain of 1, had to put the input total C parasitic as a load) - this shows a LG=0dB Fxover at 6.23MHz with 56deg phase margin. 

So this is showing a 6.23Mhz LG=0dB xover with 56deg phase margin, great, now going into this curve (which comes from a couple of 2nd order equations, and works perfectly if the loop is 2nd order, none are so still approximate but closer). So we take the Fxover*1.6 to predict a F-3dB = 9.97MHz - pretty close to what you got closed loop. We used to wave our hands in High Speed and call this a low phase margin bandwidth extension over the predicted GBP result. GBP analysis was always very simplified assuming a 1pole loop - those rarely exist but this figure helps a little for real world loops. From this article,  

Hi Michael,

Yes, I remembered that I read the great article sometimes ago. Thanks for presenting it. 

Here is another BW measurement technique with an oscilloscope. Traditionally, the injection transformer and power analyzer setup are very bulky and expensive. With Picotest injection transformer, a regular scope is able to perform the unity gain BW measurement and analysis without too much difficulties (equivalent to small signal injection method as simulated above).  

The video is shown an example of small signal injection control loop response in DC converter; and it should work the same way if it is used in an op amp feedback control loop for frequency response analysis. 

https://www.youtube.com/watch?v=ALbcfztsnAo

https://www.picotest.com/products_injectors.html

Best,

Raymond

Incidentally, I found that file generating all of this (had not opened for a couple of years, the exact multiplier is here for 56.45deg phase margin, 

So the expected F-3dB would be 6.23MHz*1.61 = 10.03MHz, even closer to your closed loop sim of 

I then reran the gain of 1 sim looking at the output pin where this analysis applies, I actually get 9.87MHz F-3dB or 1.5% error - yea for phase margins <90deg (almost always) this works a little better than GBP