TL081H: Simulating Loop Gain with basic TL081H opamp based Linear Power Supply

I'm unable to add *.tsc files to my response. Is there another way to do it?

Peter

LPS Pos Regulator - Loop Gain Eval.zip

Hi Alex,

Thank you for your response to my question.

From your modifications done to the Sim files, I created four Linear Power Supply simulation circuits.

Two for Positive Voltage. A Low-Drop-Out version and a “standard” version

Two for negative Voltage. A Low-Drop-Out version and a “standard” version.

You made an offer to determine the optimal R & C values for these circuits. I’d appreciate you doing that. Firstly, I’d like to see what sort of combinations are appropriate. Additionally, I’m sure that others would like to see what you do as well.

Both (non LDO) Regulators appear to have approx 60-degree phase margin and look good.

On the LDO versions, the phase swing was severe. I added Miller Capacitance across the PASS transistor which smoothed out the phase considerably.

Many a time, while refining these circuits, I’d get an “Access Violation” message. I could never determine what it meant.

Your comments and the corrected sim files have been very valuable for me.

For my (and others) education, why do you prefer to draw the simulation circuit “backwards”?

Finally,

I got tremendous value out of Tim Green’s “Single Opamp Slew Rate Limiter” – which included both design information and sim files.

Possibly you might consider tidying up these sim files and packaging it with some design information. Then add it toTI’s Opamp Stability and Compensation resources for others to benefit from.

Best Regards,

Peter Baxter  

Hi Peter,

I'm glad to hear that my comments have been helpful. I'll need a bit more time to answer your questions about choosing the values of the compensation components, as the frequency responses here are fairly complex, so I'm considering splitting the feedback network into a multiple 1/Beta curves. Also, I just wanted to say that these updated TINA files you provided in the ZIP file are very well put together and easy to work with, which is greatly appreciated. 

Regarding your other questions/comments:

• I mirrored the amplifier in the TINA files to better match this schematic in the literature I linked in my first response. Mostly just a stylistic choice on my part.
• The "Access Violation" error message is likely a bug within TINA, and unfortunately I'm not sure what causes this error. Usually restarting TINA helps make this error go away in my own experience.

I'll follow up within 48 hours. In the meantime, feel free to ask any follow up questions.

Lastly, 

Hi Peter,

I don't have an answer for you just yet as I'm still discussing with my team. Please allow an additional business day for us to work on this, and I appreciate your patience.

Kind Regards,

Alex Curtis

Hi Alex,

Please take your time - as I am in no hurry. I have plenty else to do.

Best Regards,

Peter Baxter

Hi Peter,

An update on our progress so far. We broke the circuit down into a few different layers to make it simpler to see how each element impacts the overall response. First, we looked at the effect of just the outer loop with the amplifier removed entirely. This shows the effect of the miller cap.

The Miller cap shifts the first pole to approximately 1.36kHz.

Positivie_LDO_Example_bipolar_only_FIXED.TSC

Then, looking at the inner part of the loop with amplifier:

Positivie_LDO_Example_amp_only.TSC

Then combining both loops together:

Positivie_LDO_Example_both.TSC

Below is the overall transient response.

Positivie_LDO_Example_both_transient.TSC

A few takeaways so far:

1. The high frequency results are very complex and ultimately aren't very useful to stabilize the circuit given that a rolloff at a low frequency is usually more desirable. 
2. We switched to the newer TL07H model instead of the TL07 model to reflect the new device's performance.
3. This circuit is difficult to analyze, and although the result is stable, I'll need to go through the AC analysis of each portion in more detail to give a better picture of which interactions are the most significant, particularly when it comes to how the output impedance of the amplifier is interacting with the miller cap and load resistor. Update: see the corrected plot of the response of the outer loop with the Miller cap.
4. Because of the presence of gain in the feedback loop, phase margin will only tell you so part of the story, so there might be some Nyquist plots headed your way soon.

Best Regards,

Alex Curtis

Hi Alex,

My Apologies for not getting back to you. I got distracted by other technical work.

You left a cryptic comment at the end of your last response which actually had me thinking I'd get the final bit of wisdom/knowledge - in a following email.

"so there might be some Nyquist plots headed your way soon."

Yes, I was hoping that it would come through as I am learning from what you have given me. You used a far higher Cf value (56n), than I was expecting.

If you can do the Nyquist plots - I'd be very interested.

Regards,

Peter Baxter.

Hi Peter,

I appreciate the follow up, I had meant to return to this earlier but had some other priorities to take care of since my last post.

I've spent a few more hours on the stability simulation of the full circuit, and checking the stability with Nyquist would be more involved than I initially thought. That being said, I might be able to find a way to generate the plots by pulling the AC responses into MATLAB, but I would need more time to go down that route. Ultimately the magnitude responses aren't too strange (for example, the magnitude response of the loop only crosses 0dB once), so they may not be too valuable here, but I can explore this more if you're interested.

The Nyquist plots aside, I think the bode plots are still useful for illustrating how the inner compensation values change the overall response. Regarding why the cap value was set to 56nF, this value depends on where the zero frequency needs to occur to bump up the phase.

I set the inner components as Control Objects (Analysis -> Select Control Object) to step through different component values to illustrate this. 

Below, I zoomed in on the 1k-10MHz range to illustrate how the phase response changes due to the compensation.

Zooming in further to where everything crosses 0dB, these are the phase margins:

The best phase margin of these combinations was with R3 = 200Ohm and C8 = 50nF. You can adjust further using the simulation below:  

Positive_LDO_Example_both_V2.TSC

Best Regards,

Alex Curtis