Stability Analysis of Power Supplies

[@prydin]

In 12 minutes, you explained a couple of weeks of EE college control loop theory. Very well done!

[@sdgb5363]

Very nicely done! Truly made me realize why control theory is a powerful tool in power supply design!

[@esmfamil3929]

thanks for your nice explanation

[@AjayGupta-knows]

Really Great video and content! Thanks!

[@OMICRONLabTutorials]

Thanks!

[@hScience100]

Hello! What is the value of injection resistor that you used ? What mean 10:1 for de ch1 and ch2 in the parameter ? For my work a used 100 ohm And on the parameter i guess o need to compensat this value I did not well uderstand if i use 100:1 or 1:100 for the two channels or just one . Thanks

[@OMICRONLabTutorials]

Dear Honodjii PDG, thanks a lot for your comment. The value of the injection resistor is not that critical. We normally use a value around 10Ω. The value must be small in comparison to the feedback resistor such that it does not change the DC operating point too much. On the other hand, the smaller the injection resistor is, the smaller the injected voltage will be. But normally we start with 10Ω. You don't need to compensate for that value at all since the Bode 100 measures both voltages on either side of the resistor. So you don't need to compensate for the injection resistor. It will just influence the absolute size of the injected signal. Best regards, your OMICRON Lab team.

[@robr8554]

Fantastic video

[@johnwhittaker346]

Excellent Video! I am going to get a Bode 100.

[@OMICRONLabTutorials]

Well, that makes us really happy. :-) Thanks for your post!

[@songzhengchen451]

Hi, thanks for explaning. it is really clear about the setup and the test point in the video. But i am using a primary-side regulator flyback converter. And the feedback loop uses a voltage divider before the recitifier diode and go into the Vs pin of the IC. Do you have any suggestion on the injecting and measure point for that topology? thanks you so much.

[@OMICRONLabTutorials]

Dear Songzheng Chen, We do not think that is makes sense to measure the feedback loop in a primary side regulator because you cannot adjust any poles / zeros in the controller. What does make sense is to measure the output impedance of your supply. When you measure the output impedance, then you will see how good your supply is and if there are any critical frequencies that can cause ringing. If you haven’t seen it, maybe you can check out: www.omicron-lab.com/applications/detail/news/output-impedance-of-power-supplies If you have any further questions, please feel free to contact us anytime at: [e-mail]. best regards, your OMICRON Lab team

[@thebrakshow7415]

You mention to take care using an E-Load because if can interfere with the DUT. What do you mean by this? Do you have any examples of issues with E-Load on any of your videos? Also, what is the resistive load you are using. I'm having a hard time finding a bench top purely resistive load. Currently I'm using cheap rheostats dials.

[@OMICRONLabTutorials]

Depending on the control mode of the electronic load, it also incorporates feedback and therefore it is a dynamic system as well. This can distort the measurement of the loop. We also use resistive sliding loads or power resistors mounted to a heat sink for loading.

[@yangkuang5252]

great sharing, I have brought a bode 100, and want to know if i inject a big signal, like you showed 8dBm, what kind of stauration will happen?

[@OMICRONLabTutorials]

This will depend on your circuit. You can get e.g. slew-rate limitations in the error amplifer if you use too much signal level.

[@robertovolpini9359]

hi thanks for this explanation, do you have some suggestions in order to make measures on 360Vdc output voltage of power supply? I saw that 10x probes are not good because the high noise? thanks for your suggestions BR/Roberto

[@OMICRONLabTutorials]

Hi Roberto, 10x probes do normally work but you should not use standard scope-probes with our Bode 100 for high voltages. The Bode 100 inputs are AC-coupled with high impedance, so normal scope probes don't divide DC. Please use the PML-111O 10:1 passive probe. Alternatively you can use active high-voltage differential probes. But these normally do have more noise. For more details, please contact us via [e-mail]

[@robertovolpini9359]

@@OMICRONLabTutorials thanks for your reply

[@norris1218]

Does this work with standard 10M 10:1 oscilloscope probes? I forgot to buy the PML probes and now it will take forever for my company to get them in.

[@OMICRONLabTutorials]

Dear Norris, If you have voltages

[@amleshkumar6255]

Thanks for the explanation. To remove noise at Low frequency, Shape Signal is used. 1) what about if noise is present at high frequency? 2) How to correlate/implement this shape Signal with real product? 3) To remove noise, is there any other way? Such as compensation circuit or any other way? Thanks Amlesh

[@OMICRONLabTutorials]

Dear Amlesh Kumar, thanks for your comment! Our answer to your questions: 1) Noise at high frequencies can be reduced similarly to low frequencies. This means using the shaped level function, reducing the receiver bandwidth as well as the attenuators at channel 1 and 2. 2) More information about the shaped level function can be found in our application notes (e.g. www.omicron-lab.com/applications/detail/news/dcdc-converter-stability-measurement/; page 12 onwards). 3) To further remove noise in the measurement, you could also do sweep to sweep averaging. Adding filters is not recommended because they will influence the measurement. If you have any further questions, please feel free to contact us via [e-mail]. best regards, your OMICRON Lab team

[@cuchulainkailen]

6:48 You state 80 deg. phase margin. The Y axis implies raw phase, so then PM would be 80 - (-180) = 260 deg. It's 80 because (at 10:33) of the inherent -180 rotation due to injection point. The display shouldn't define Y axis as phase, but "phase margin".

[@OMICRONLabTutorials]

Since the equipment does not know what kind of external circuitry is measured, the axis just shows phase. And Phase Margin actually only exists at the crossover-frequency. Calling the axis "Phase Margin" would also not be really correct. I think, the user does need some application knowledge to interpret bode-plot measurements.

[@marcinmazur2990]

Great content!

[@OMICRONLabTutorials]

Happy to hear that!

[@paulpaulzadeh6172]

best explanation

[@OMICRONLabTutorials]

Thanks for your feedback!

[@scottdurant8599]

what Rheostat are you using?

[@OMICRONLabTutorials]

Hi, we are using a Contrex rheostat.

[@simonbour56p]

Great video!

[@OMICRONLabTutorials]

Thanks!

[@simonbour56p]

@@OMICRONLabTutorials It would be amazing if you did a video on shunt reference/opto-coupled feedback for power converters.

[@OMICRONLabTutorials]

@@simonbour56p Thanks for the proposal. What would be your interest? The transfer function / stability or how to design it?

[@simonbour56p]

@@OMICRONLabTutorials Actually both. Designing for a specific transfer function and how to measure it (validate it) isolated from the power supply itself.

[@OMICRONLabTutorials]

@@simonbour56p , we do have an application note how to measure the transfer function of an opto-coupler: www.omicron-lab.com/applications/detail/news/frequency-response-of-optocouplers/, for the design, please refer to kzbin.info/www/bejne/e6KbmXqAZ79_l8U and the other videos from Biricha. Hope this helps!

[@gauravmg]

Does anyone know what kind of resistive load is being used in this application?

[@OMICRONLabTutorials]

That's a slider resistor. I think it is also called Rheostat.

[@hScience100]

T(s) = - G(s)*H(s) and then non reference signal mean Vref(s) =0 ?

[@OMICRONLabTutorials]

Yes, it was assumed that the reference does not change in that case. If you have multiple inputs and outputs, normally they are analyzed one-by-one. In this case the disturbance on vref was considered to be zero that's why vref is not mentioned. However, Vref does not actually influence the loop gain as it is not directly part of the loop.

[@hScience100]

@@OMICRONLabTutorials Using Bode 100 is like we make study the small singal variation. Vref is a constant and its small variation is zero.

[@OMICRONLabTutorials]

@@hScience100 You are right. But in this case we don't change Vref but the measured feedback signal Vfb by injecting a voltage signal. Then the feedback loop reacts to that change in Vfb and the loop gain can be measured.