OPA814: simulate gain stability for ninv opamp

Hi Dennis,

I would look over our TIPL Stability series (link below). Specifically I would cover the video titled "Breaking loop on differential amplifier." The circuit is not a difference amplifier but you can still break the circuit at the inputs and get the better results for the circuit. I also added a link to an app note that covers numerous topics related to stability. I would skip to page 24 and read that section as this is more information about the same topic of simulating stability by breaking the inputs.

www.ti.com/.../ti-precision-labs-op-amps.html

https://www.ti.com/lit/wp/sboa626/sboa626.pdf?ts=1774367708944

Best Regards,

Ignacio

Hi Ignacio

A follow up question: in the OPA814 datasheet, page 15 figure 7-4, do you know what circuit was used to generate this response?  The datasheet did not declare if the circuit was inverting or noninverting.

dennis brown

Hi Dennis,

These open loop plots are captured in simulation by the designers when finalizing the design. In there simulation software they have a meter where they can measure this and it will give them the gain and phase of the design. In most cases the device is in a buffer and they use this meter to capture this data.

Best Regards,

Ignacio

Hi Ignacio,

I would like to ask follow up questions:

There is the OPA814 reference design for tina.  The circuit is configured like this:

The overall gain here should be Av = (1+ 250/250) or Av = 2, or 6.02dB

When I simulate this under closed loop conditions I don't get 6.02dB but instead the plot shows 2dB.  Can you verify?

While this is a close loop simulation I do see that the phase of the signal is heading down towards -180deg.  Is this a problem? 

In the green box I note that the overall gain is heading towards 0 dB but the phase is also decreasing.  It just seems like this model suggests that this could be unstable as it exists but this is what i am trying to determine. Mainly looking for mitigations to help this circuit.  I do have provisions on my circuit board for a parallel RC at the input and parallel RC in the feedback path from out to inv pin.

Hi Dennis,

I am wondering if you are simulating the ac response using a linear scale on your Y axis because it looks to be about 2. If you double click the y axis you should be able to see the scale of the axis and switch accordingly. Regarding stability, you will not get as much information from the closed loop response. In order to simulate stability, you must break the loop to capture your loop gain. The links I shared previously goes through how to do this. As far as provisions, I would include an isolation resistor at your output as a place holder incase you have any capacitance at your load. 

Best Regards,

Ignacio

Hi Ignacio,

I took the OPA814 as downloaded and modified it to use a typical open loop analysis. 

Provided I have this setup as expected, I am not seeing much of a problem

The phase margin here would be 180 - 62.49 = 117.51

I don't think I have a lot of load capacitance here in my design.

In the above circuit I have 5pF on the load but this does not cause much problem.  Even at 10pF not really seeing anything.

The board layout seems ok to me, I have keepouts around inv, ninv, and output pins as shown below:

As the PCB exists, we can get the oscillations to stop by placing a 50ohm in the feedback loop for a unity gain stage. 

Maybe the problem lies elsewhere, perhaps a power supply start up issue.

Hi Dennis,

The sim model might be overly optimistic on its ability to drive the capacitive load. In the device datasheet we recommended an isolation resistor of 50ohms into a 10pF load. I would assume 5pF would be in a similar range therefore little to no isolation will probably not work in most configurations. However since you have a PCB, we can focus on that to see what is happening. Can you share the schematic for the PCB you are testing. I am unsure what values are being used for the decoupling caps or surrounding components. In your PCB, are you removing any load when trying to debug the oscillation. I would recommend doing so to remove other variables. 

Best Regards,

Ignacio

Hi Ignacio

I think I may have missed the "Figure 6-9. Recommended Isolation Resistor vs Capacitive Load" plot.  We have to do a board spin anyways for other problems the layout tool messed up. This opamp is driving another FDA with appx a 300ohm load.  The stray capacitance should not be that much but I have yet to measure.  I am already in the process of trying to find why the oscillation occurs so I will remove the load.

For the decoupling caps, there are three per rail with 0.01u, 0.1u, and 1u.  All are rated to 10V but the rail is bipoloar 5.5V.  The 10V are not good here and I should use something like 25V parts as the capacitance voltage coefficient may be affecting things here.  At this point I would prefer to talk over email and not share more design info publicly like this.  Please PM me if you are agreeable.  

dennis brown

Hi Ignacio

This request did not show up on my end.

I don't think I require much time to solve this problem.  

dennis brown

Hi Dennis,

No worries I went ahead and sent you an email.

Best regards,

Ignacio