| Electronics > Projects, Designs, and Technical Stuff |
| 3rd order Sallen Key Unity Gain LPF - stability question |
| << < (3/6) > >> |
| Wimberleytech:
--- Quote from: JDW on April 25, 2020, 01:17:33 am --- Next, if one contends the calc is incorrect with regard to Oscillation, one could also contend the calc is not sufficiently accurate to reply upon for general design. --- End quote --- It appears that the bode plot from the calc site agrees with the LTSpice simulation |
| JDW:
--- Quote from: Wimberleytech on April 25, 2020, 01:31:37 am ---Three things: 1) analyze the circuit by hand and prove to yourself that you cannot get the required phase shift to achieve oscillation. This is some work because you will also have to take into account the phase shift associated with the opamp. 2) Simulate it. Simulators are very helpful to run different cases to see if they oscillate. 3) Run corner case examples with your breadboard. I am posting a simulation file for you. It is always best to confirm performance in the large-signal domain (transient response). --- End quote --- Thank you for your willingness to help, but I must repeat that I have zero experience with LPF design, as evidenced by the fact I used that online calc to create my existing design in the first place. As such, I do not know how to use your *.asc file. I also am unclear on the part of doing manual calculations for "phase shift" on the design I have presented. And lastly, I am not clear on your meaning of "run corner case examples." I fully realize that folks such as myself are extremely frustrating to you gentlemen who are knowledgeable on the subject. Some may rightfully contend that I have no business even considering use of such a filter when I know so little about the fundamentals of filter design, but the reason I posted here is for some serious hand-holding on the subject. Just treat me as a little kid who knows next to nothing on the subject. That will be a learning experience for me that will get me on the right track. My humble thanks for your time and kind consideration. |
| Wimberleytech:
--- Quote from: JDW on April 25, 2020, 01:43:48 am --- --- Quote from: Wimberleytech on April 25, 2020, 01:31:37 am ---Three things: 1) analyze the circuit by hand and prove to yourself that you cannot get the required phase shift to achieve oscillation. This is some work because you will also have to take into account the phase shift associated with the opamp. 2) Simulate it. Simulators are very helpful to run different cases to see if they oscillate. 3) Run corner case examples with your breadboard. I am posting a simulation file for you. It is always best to confirm performance in the large-signal domain (transient response). --- End quote --- Thank you for your willingness to help, but I must repeat that I have zero experience with LPF design, as evidenced by the fact I used that online calc to create my existing design in the first place. As such, I do not know how to use your *.asc file. I also am unclear on the part of doing manual calculations for "phase shift" on the design I have presented. And lastly, I am not clear on your meaning of "run corner case examples." I fully realize that folks such as myself are extremely frustrating to you gentlemen who are knowledgeable on the subject. Some may rightfully contend that I have no business even considering use of such a filter when I know so little about the fundamentals of filter design, but the reason I posted here is for some serious hand-holding on the subject. Just treat me as a little kid who knows next to nothing on the subject. That will be a learning experience for me that will get me on the right track. My humble thanks for your time and kind consideration. --- End quote --- Sure I will be patient! The .asc file is an LTSpice file. Most everyone that helps out on this forum is expert at using LTSpice. It is a free tool and extremely powerful. You should learn to use it. Alternatively, if you have access to another circuit simulator, use it instead. I spent 40+ years designing integrated circuits. When we say "corner cases" what we mean is run the various extreme cases for parameter variations. This includes temperature. In your case, that would be playing around with 5% (or 10%) variations in component values to see how the performance varies. Not sure what the worst case corners would be for the SK filter. One approach is to do Monte Carlo simulations and let the simulator run a statistical analysis for you. The SK filter is very robust...esp with opamp in unity gain. It is not going to oscillate unless you have a problem with the opamp itself. |
| Wimberleytech:
For a circuit to oscillate, it must feed back a signal to the input in phase with the signal that generated it. Kind of a clunky way to say it...a better way is the Barkhausen criterion: The magnitude of the gain in the loop is 1 and the phase shift is 0°. So, how do test this SK configuration. Knowing where to break the loop for this analysis can be tricky, but for the SK it is easy...break the loop at the output of the opamp. Then input a signal at the dangling end of C1 and analyze/calculate the transfer function to the now open output of the opamp. If you can get a gain of 1 at 0° then you have an oscillator. If you get a gain more than 1, generally you will still have an oscillator as the amplitude increases, nonlinearities will bring the gain down to one at some point. You can do this analysis in LTSpice and save yourself some algebra! But at some time in your life, you need to plow through this just once. |
| Jay_Diddy_B:
Hi, (Wimberleytech - congratulations on reaching 1000 posts !!!) I will give you the answers without too much of an explanation: Frequency Domain result Time domain - step input This modelling shows a well behaved 3 pole low pass filter. Stability Analysis Stability Model Results The stability model results shows that there is 6dB of gain margin. This means that gain of the amplifier has to change from unity to x2 for oscillation. Gain Stepping The opamp has been replaced with an amplifier so that I show the effect of increasing the gain: Gain step results This model shows that the gain has to be increased to 2, a 6dB increase for oscillation. Regards, Jay_Diddy_B |
| Navigation |
| Message Index |
| Next page |
| Previous page |