EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: RoGeorge on October 04, 2022, 10:46:06 am
-
Been playing with LTspice recently trying to design a CC/CV power stage for a signal generator, and been testing all kind of ideas and schematics.
Many of them were oscillating from the very start ;D, others were giving good hopes then they were starting to oscillate under different loading conditions. :horse:
My problem is that there is too much info about the subject, from the most abstract control theory dealing with math only, to the generic advise like "always add decoupling capacitors". I'm somewhere in the middle and without having a good understanding of how to deal with the stability of control loops. For example, I've read about opening the feedback loop and injecting signals, but in practice I don't even know for sure in what point to break the loop. Things like that.
Usually I end up adding random capacitors to the loop, or poking different values for the existing ones then hit Run in LTspice over and over, which is no fun and almost never succeed. :-\
Any good textbooks (or other type of material) to learn about feedback loops design and stability, please?
-
Been playing with LTspice recently trying to design a CC/CV power stage for a signal generator, and been testing all kind of ideas and schematics.
Many of them were oscillating from the very start ;D, others were giving good hopes then they were starting to oscillate under different loading conditions. :horse:
My problem is that there is too much info about the subject, from the most abstract control theory dealing with math only, to the generic advise like "always add decoupling capacitors". I'm somewhere in the middle and without having a good understanding of how to deal with the stability of control loops. For example, I've read about opening the feedback loop and injecting signals, but in practice I don't even know for sure in what point to break the loop. Things like that.
Usually I end up adding random capacitors to the loop, or poking different values for the existing ones then hit Run in LTspice over and over, which is no fun and almost never succeed. :-\
Any good textbooks (or other type of material) to learn about feedback loops design and stability, please?
In spice simulations you don't need decoupling capacitors, unless you have explicitly included parasitic inductance in the model you are simulating. You do require a good understanding of control theory and stability.
First you model your abstract topology in the s-domain (or z-domain if using sampling control systems). After you have understood the conditions under which those models are/aren't stable, you can create a schematic with devices and model how closely it approximates your abstract topology. Finally you add decoupling caps etc to make your circuit board a better approximation to your schematic.
It is difficult to recommend a specific text book, since few will cover the entire field from theoretical control theory through to practical topologies and specific devices. In addition different people find different styles more/less approachable.
-
Generally you have to add the parasitic components to ideal models.
put a small resistor in series with capacitors and inductors.
The spice circuit is composed of ideal components. This will cause unrealistic oscillations.
-
Check out Linear Technology application notes 18 and 47. Both discuss the various causes and types of oscillation.
-
It's a law of nature that, on first switch-on, homebrew amplifiers will always oscillate while homebrew oscillators never will.
-
The spice circuit is composed of ideal components. This will cause unrealistic oscillations.
... as will the non-ideal components and parasitics on circuit boards.
-
Sometimes a type 3 compensator of the error will be needed for a phase margin on the loop gain at high frequency.
The attached images are of a qucs model I made for a 300 watt DC converter with a type 3, showing how the loop was cut to add the sweep generator to produce the bode plots.
The breadboard prototype approximately followed this model and I am building the final version now.
The images are for the voltage control loop with type 3. I used a type 2 for the current limit loop as the plant has lower gain with high load.
https://app.box.com/s/k4yddmriam1mgtoj4c7iee2i3kcnimxe (https://app.box.com/s/k4yddmriam1mgtoj4c7iee2i3kcnimxe)
https://app.box.com/s/w62i35o5pq1vwjr13knqud09jl8w5jpg (https://app.box.com/s/w62i35o5pq1vwjr13knqud09jl8w5jpg)
A reference for the type 3 compensator was Renesas TB417
https://www.renesas.com/us/en/document/oth/tb417-designing-stable-compensation-networks-single-phase-voltage-mode-buck-regulators
-
You could try building an oscillator, then it won't. At least that's my experience.