Hi Younes, Jack Alexander's model is a small signal model meant for plot AC Bode plots. The transformer, primary MOSFET and output diodes are modeled by a simple voltage control current source. It is a very simple model and will give you open loop gain plots. Robert
Normally if you work in the industry you mostly likely will have access to a FRA (Frequency Response Analyzer). One rudimental way of checking stability is to do a step change or load change to see how fast the feedback compensate the power supply. If you phase margin is below 30 degrees or lower, you will start to see ringing after a step response. Look on my channel under the Flyback series and watch video 8. In it I show how to adjust the loop and actually do it on a proto board supply. I hope this helps. Thank you for your questions and interest in this subject Best regards, Robert
Hi Kamil, When you are designing a power supply, you typically want the power supply to regulate your output voltage when you have input voltage variations and ever load variations. To regulate the output voltage of a power supply, you have to sample it, feed it back (negative feedback) and compare it against a know voltage or, in this case, a voltage reference. This is done with a op amp. The op amp compares the feedback voltage with against the voltage reference and generate an error voltage that will then be feed to the PWM. If the feedback voltage or I should the gain around the feedback loop are in phase and you have a gain of 1, the power supply will oscillate. It is imperative that you compensate the loop to ensure that you will have a stable power supply thru out its operating input voltage and all its loads condition. When you are tasked to designed a power supply, the customer will offend will specify the stability criteria by specifying the Phase Margin (>45 degree) and the Gain Margin (
Hi Robert, thanks for your prompt response. In few months from now - after design of few power supplies I will feel this topic better. At this point It is good to hear from experienced designer that this is the way how feedback loop must be design for every power supply. Recently I found on internet general explanation of stability issue for systems with feedback so I going to improve my knowledge in this topic. I would like to ask you one more question regarding compensation. Can I be sure that not compensated psu will always oscillate or I only know that oscillation are possible in this unit and some day, under some conditions can happen e.g.. after some specific step change in output current psu can start to oscillate.
Hi Gupta, no you can not use the switching model of a pen to simulate and AC loop response. you would have to invest to an average model which is what I presented in the video lecture. hope this makes sense.
Thank you very much for you valuable information, The question is in CCM we have right half zero, how can we model it using Jack's model and how can we compensate it?
Hello Robert, Watching your video I understand a lot about how to design/simulate a feedback loop, add compensation network around op amp and how to verify loop response. Could you please summarize why we need loop compensation? Is it obligatory to add it in every power supply? What if we leave power stage without compensation? I know how to compensate the loop but I not sure if I am able to explain to someone else why I am doing it. I see good things related to high gain for low freq and low gain at high freq but if you could please summarize this topic.
Very interesting work! I would like to ask how would this apply for several output regulation, ie: 12 in and 2 symetical -+2.5v out 3 amp each per 2.5v rail, so the total gail would be 6amps per volt?, then for the filter capacitor of each rail and RL (0.83 ohm), the compensation should take into account both capacitor with their esr?, only can compensate one side... im little confussed on multiple output Thanks i advance!
Hello Mr. Bolanos, You have very useful videos in your channel. Thanks for your efforts. I would like to ask; The compansation circuit analyzing techniques in this video are usable in the same way for " adjustable constant current supply" in the form of current mode forward converter instead of flyback voltage supply? More clearly to ask, can i feed the output current sample (in V form) back to the amplifer to compare with an adjustable Vref in order to get a user adjustable output current? Do we even need another current sensor in the output while we already have one (Rsense) in the power stage since it does sense the primary current which is actually proportional to the output current ( altough it is only the sample of switch current when conducting in Ton)?
Thank you Mr. Bolanos for prompt reply. So, that means, only one current sensor in the primary of power transformer (that is one feedback loop) would be enough to control output current as we want ? Best Regards.
Yes, the current sense in the primary should be able to set your current limit. However, keep in mind that the Isense in most PWM have a maximum current sense (1V). If you set the Rsense to produce 1V at the source of you mosfets then you are reducing the Vgs across the mosfets. So what most people so is use a current transformer with 1 turn on the primary (connected between the source and ground) and feed the secondary with 50 turns to the Isense resistor. What this do is the Isense=Iprimary/50turn. So then you connect a large resistor at the secondary. Watch the video belowruclips.net/video/LhLKDZk_yRo/видео.html
Hello Robert, thanks for this video! I am a new electrical engineer coming from a physics background and, unfortunately, did not go very much into depth with regard to controls. Do you have any texts that you recommend as well? I haven't finished watching your series, I am about 1/4 the way into this one.
Hi Fabian. This depends on the PWM that you use and the expected voltage in the PWM's COMP pin. Typically the COMP pin's voltage can be around 1V to 3V. The output current is around 0.5A. In my case, I expect a voltage at the COMP pin to be arpund 1V. So the transconductance value is 0.5A/1V = 0.5A/V. I hope this makes sense. Robert
I am so sorry; I did not see your question. R21,R11 and the switch is to perform a transient step load and test the models transient response to a dynamic loading.
Hey Robert, Thats really nice video. I am trying to do the same for my flyback but on Pspice. I am not able to get a veneble supply or variable frequency supply, please let me how do I search that I dont know the exact name of :(
Vaibhav Gupta Hi Gupta, I think you use a voltage source and double click and set the AC to 1 and set the AC analysis to the frequency range that you want to run.
Robert Bolanos, Thank you I was able to simulate it. Just wondering is it possible to run an AC sweep with a actual pwm spice model? Because I tried using it and I was able to run transient analysis, however with AC sweep the same settings are giving me Convergence issue. I am using UC3841, current mode control.
Hi Mr Roberto B. for my projet i must do a boost converter 12/60 V its work good in open loop but now i'm looking de compesator to make the close loop i did a compensator type II isn't work and i want to do a compesator type III but i don't know how to calculet the parametres : CAN YOU HELP ME PLS ? BAY THE WAY : DO YOU SPEAK PORTUGUES OR SPANISH ?
Yes the small signal model should be very similar. I do have a video on how to design a type 3 compensation. It should be the voltage mode compensation. It should be under the loop compensation series of videos
This is a great video - It explains so much about the operation of compensation in op amps
Hi Tim, I am glad you like the video. Robert
Hi Younes,
Jack Alexander's model is a small signal model meant for plot AC Bode plots. The transformer, primary MOSFET and output diodes are modeled by a simple voltage control current source. It is a very simple model and will give you open loop gain plots.
Robert
Normally if you work in the industry you mostly likely will have access to a FRA (Frequency Response Analyzer). One rudimental way of checking stability is to do a step change or load change to see how fast the feedback compensate the power supply. If you phase margin is below 30 degrees or lower, you will start to see ringing after a step response. Look on my channel under the Flyback series and watch video 8. In it I show how to adjust the loop and actually do it on a proto board supply. I hope this helps. Thank you for your questions and interest in this subject
Best regards, Robert
Thanks so much for this feedback loop explanations. I understand it better after this video.
That's a very clear and practical explanation. Thanks a lot for the video!
Thank you the comments Sosa!.. Robert
Hi Kamil,
When you are designing a power supply, you typically want the power supply to regulate your output voltage when you have input voltage variations and ever load variations. To regulate the output voltage of a power supply, you have to sample it, feed it back (negative feedback) and compare it against a know voltage or, in this case, a voltage reference. This is done with a op amp. The op amp compares the feedback voltage with against the voltage reference and generate an error voltage that will then be feed to the PWM. If the feedback voltage or I should the gain around the feedback loop are in phase and you have a gain of 1, the power supply will oscillate. It is imperative that you compensate the loop to ensure that you will have a stable power supply thru out its operating input voltage and all its loads condition. When you are tasked to designed a power supply, the customer will offend will specify the stability criteria by specifying the Phase Margin (>45 degree) and the Gain Margin (
Hi Robert, thanks for your prompt response. In few months from now - after design of few power supplies I will feel this topic better. At this point It is good to hear from experienced designer that this is the way how feedback loop must be design for every power supply.
Recently I found on internet general explanation of stability issue for systems with feedback so I going to improve my knowledge in this topic. I would like to ask you one more question regarding compensation. Can I be sure that not compensated psu will always oscillate or I only know that oscillation are possible in this unit and some day, under some conditions can happen e.g.. after some specific step change in output current psu can start to oscillate.
Hi Gupta, no you can not use the switching model of a pen to simulate and AC loop response. you would have to invest to an average model which is what I presented in the video lecture. hope this makes sense.
Thank you very much for you valuable information, The question is in CCM we have right half zero, how can we model it using Jack's model and how can we compensate it?
Hello Robert,
Watching your video I understand a lot about how to design/simulate a feedback loop, add compensation network around op amp and how to verify loop response.
Could you please summarize why we need loop compensation? Is it obligatory to add it in every power supply? What if we leave power stage without compensation? I know how to compensate the loop but I not sure if I am able to explain to someone else why I am doing it. I see good things related to high gain for low freq and low gain at high freq but if you could please summarize this topic.
great!
Hello Robert,
Where is the freewheel diode for the primary inductance ?
Very interesting work!
I would like to ask how would this apply for several output regulation, ie: 12 in and 2 symetical -+2.5v out 3 amp each per 2.5v rail, so the total gail would be 6amps per volt?, then for the filter capacitor of each rail and RL (0.83 ohm), the compensation should take into account both capacitor with their esr?, only can compensate one side... im little confussed on multiple output
Thanks i advance!
Watch this video,,,,, ruclips.net/video/uU7liNMrD0Y/видео.html
@@RobertBolanos oh much appreciated i will!!
Hello Mr. Bolanos, You have very useful videos in your channel. Thanks for your efforts.
I would like to ask;
The compansation circuit analyzing techniques in this video are usable in the same way for " adjustable constant current supply" in the form of current mode forward converter instead of flyback voltage supply?
More clearly to ask, can i feed the output current sample (in V form) back to the amplifer to compare with an adjustable Vref in order to get a user adjustable output current?
Do we even need another current sensor in the output while we already have one (Rsense) in the power stage since it does sense the primary current which is actually proportional to the output current ( altough it is only the sample of switch current when conducting in Ton)?
Hi Brokensword. Questions and comment can be sent to rbola36518@aol.com
Thank you Mr. Bolanos for prompt reply.
So, that means, only one current sensor in the primary of power transformer (that is one feedback loop) would be enough to control output current as we want ?
Best Regards.
Yes, the current sense in the primary should be able to set your current limit. However, keep in mind that the Isense in most PWM have a maximum current sense (1V). If you set the Rsense to produce 1V at the source of you mosfets then you are reducing the Vgs across the mosfets. So what most people so is use a current transformer with 1 turn on the primary (connected between the source and ground) and feed the secondary with 50 turns to the Isense resistor. What this do is the Isense=Iprimary/50turn. So then you connect a large resistor at the secondary. Watch the video belowruclips.net/video/LhLKDZk_yRo/видео.html
Thank you very much, Mr. Bolanos.
Hello Robert, thanks for this video! I am a new electrical engineer coming from a physics background and, unfortunately, did not go very much into depth with regard to controls. Do you have any texts that you recommend as well? I haven't finished watching your series, I am about 1/4 the way into this one.
Hi Ruben, these are the books I recommend. Hope this helps.ruclips.net/video/uWfOnKQzm5Y/видео.html
Robert Bolanos wow, I should have checked your profile. you are a great resource thanks!
Hi Robert! Why did you set the gain to -0.5 and not -1?
Hi Fabian. This depends on the PWM that you use and the expected voltage in the PWM's COMP pin. Typically the COMP pin's voltage can be around 1V to 3V. The output current is around 0.5A. In my case, I expect a voltage at the COMP pin to be arpund 1V. So the transconductance value is 0.5A/1V = 0.5A/V. I hope this makes sense. Robert
why did we use r21,r22 and sw? What are their effects on the system? I did not use it and got the same results, sir. I am using LTSpice and Microcap12
I am so sorry; I did not see your question. R21,R11 and the switch is to perform a transient step load and test the models transient response to a dynamic loading.
@@RobertBolanos thanks for your answer. You are a very helpful trainer...
Hey Robert,
Thats really nice video. I am trying to do the same for my flyback but on Pspice.
I am not able to get a veneble supply or variable frequency supply, please let me how do I search that I dont know the exact name of :(
Vaibhav Gupta Hi Gupta, I think you use a voltage source and double click and set the AC to 1 and set the AC analysis to the frequency range that you want to run.
Robert Bolanos, Thank you I was able to simulate it.
Just wondering is it possible to run an AC sweep with a actual pwm spice model? Because I tried using it and I was able to run transient analysis, however with AC sweep the same settings are giving me Convergence issue.
I am using UC3841, current mode control.
@@vaibhavgupta0 Hi Gupta, TopSPICE now has a FRA feature where you can run an FRA using you switching model and get a loop plot.
Hi Mr Roberto B. for my projet i must do a boost converter 12/60 V its work good in open loop but now i'm looking de compesator to make the close loop
i did a compensator type II isn't work and i want to do a compesator type III but i don't know how to calculet the parametres : CAN YOU HELP ME PLS ?
BAY THE WAY : DO YOU SPEAK PORTUGUES OR SPANISH ?
Honodjii
hi what own are you using? current mode or voltage mode
voltage mode
Honodjii PDG current mode is simpler to compensate. however, look in my channel, I have a video where I show how to compensate a voltage mode.
in you video you talk about Flyback, is the same like boost converter ?
to make compensator
Yes the small signal model should be very similar. I do have a video on how to design a type 3 compensation. It should be the voltage mode compensation. It should be under the loop compensation series of videos