The inner works of Peak Current Mode (PCM) control

[@vasiliy5578]

Thank you prof. Sam for the interesting content! Your videos are so fundamental like the proper alphabet of electronics. Nice part with chatGPT, clever words.👍

[@sambenyaakov]

Thanks.

[@sambenyaakov]

Thanks

[@paulpaulzadeh6172]

Hi Prof. Interesting result LG measurement what a bout current feedback amlifier case ? Can we use the same method as a voltage feedback amplifier? What if you have combined current feedback, inner loop , and outer voltage feedback?

[@sambenyaakov]

@@paulpaulzadeh6172 You need to decide what to measure. See for example H. Tsafrin and S. Ben-Yaakov, "The dynamic response of PWM DC-DC converters with input filters," [Proceedings] APEC '92 Seventh Annual Applied Power Electronics Conference and Exposition, Boston, MA, USA, 1992, pp. 764-771

[@2meters2]

Thank you Sam. This PCM method seems attractive at first sight, but in my opinion there are a number of issues with it that put constraints on the situations when you can use it : 1) As the wave forms at 2:00 shows, the current detection needs to be very fast and very reliable. If the PWM frequency is 100 kHz, then the peak current detector needs to be at least a factor or 10x faster, so needs to respond within 1 usec or so. If it misses even one cycle, or is a bit too slow, the loop locks up and your converter burns itself out. And since this happens very fast, any sort of second layer of safety is going. to be harder to implement too. Not very 'fail-safe' in my opinion... 2) As the schematic at 3:00 shows, this circuit only works if the 'logic' gnd is the same as lower rail of the converter. This is not always the case, especially not with the recent trend to put isolation devices between the power train and the control logic. 3) Very often, a digital controller (MCU) is used to provide the PWM signals to the switches, and that means we want to use PCM, we need another ADC in the loop, and that one ALSO needs to be very fast (1 usec or so). 4) Finally a note : The Duty-Cycle issue in PCM (diagram at 6:00) that Prof. Middlebrook first reported about in my opinion is simply a loop 'gain' dependency in the feedback loop. After all, with the resistor-in-lower-switch current detector we are only measuring the current when the switch is on. So the loop gain increases linearly with the duty-cycle. Is that a fair statement ? Thanks again for reporting this interesting method of implementing a current loop, but may it be that the application of this method is not very often used any more now that PWM control with an MCU is more common ?

[@peplegal8253]

- PCM needs to be done "analoguely". Don't put it inside the DSP. It's easy to be implemented and gives an intrinsic over-current protection for the Mosfets. - You can "skip" the ultra fast reading of the inductor current (which is costy and prone to noise) by just reading the slower output current signal, using a hall sensor for example...added over the integral of the inductor voltage ( from a secondary winding, for exemple). The resulting signal can make up for the true inductor current...and it's much more clean.

[@2meters2]

@@peplegal8253 Yes. I prefer the Hall sensors to measure the current in the inductor directly. Those are inherently isolated, and the current feedback loop can also be much lower bandwidth (down to PWM frequency) so noise and spikes are much less of a concern. And we can process it in a digital MCU or analog feedback loop alike. So I do realize than the LTC7891 that Sam presented in 16:50 uses a PCM feedback loop, so this method is clearly still used, but with the design constraint and safety issues alike I think those designs are bold and not for the faint of heart :-)

[@sambenyaakov]

Sorry to break to you the bad news. You are in in the minority. Most of today's dcdc designs are based on pcm . Think about all the cellular phone chargers out there. Most use a flyback with PCM.

[@2meters2]

@@sambenyaakov Thank you for correcting my statement. Is it fair to say that the great benefit of PCM is that it can provide a constant voltage output with no switching going at all (super low power use) if there is no load request ? I was thinking in my field of AC/DC power inverters and EV chargers, where high power needs to be moved using an digital controller. For such applications, I have been reluctant to use PCM for the reasons outlined above. But I have to admit that I'm still rather new to this field, and I am learning a lot (especially from your lectures) about which architecture to use for which applications. In fact you have been pretty much my mentor for the past 2 years in this field, and I thank you very much for that.

[@sambenyaakov]

@@2meters2The primary benefits of PCM are cycle by cycle protection and a wider bandwidth. A poor boy compromise that can be easily implemented digitally is the average current mode. This in fact is used (albeit not named as ACM ) in modern EV drives. The inner current loop, and the outer speed loop.

[@idk2412]

Another great video! I was hoping you would also show a practical implementation of how YOU apply slope comp using a generic controller IC, for instance the UC384X

[@sambenyaakov]

This is for homework😊But the fact is that most modern PCM controllers have a built it slope compensation network. There is ltle reason to use UC38x

[@sergiomiranda5495]

Interesante, saludos desde México.

[@sambenyaakov]

Thanks.

[@lllllll396]

Thank you for your videos!

[@sambenyaakov]

Thanks

[@ЁбрагимИпатенкоибнАдхарма]

Oh such an underestimated content. My 100-th like is yours)

[@sambenyaakov]

😊🙏

[@thavakkodi6290]

Sir i have some doubts other than the PWM generation of Peak current control what are the other parameters gets affected due to the short circuit condition in peak current mode control ? like the which parameters gets affected due to that high rate of change of current sir ?

[@eduardinification]

Regardless the AI will be adopted or discarded in the future, the IC manufacturers will get the benefit, provided the huge computation the AI requires.

[@sambenyaakov]

👍Indeed

[@j.vijaymahesh1004]

Is there any calculation to calculate the values of Capacitor and resistors in compensation network

[@sambenyaakov]

See the videos on control design in my KZbin channel.

[@tomheeks2830]

Excellent video, thanks. So this is why you have the ramp pin on a PWM driver. Is it not also used to generate the square wave output?

[@sambenyaakov]

The ramp is part of the oscillator. In voltage mode controllers it is part of the duty cycle generator. In PCM it can be used for ramp compensation.

[@luzhouyang5414]

professor,what's V(EXC) of B4 in your LTspice simulation in your presentation？

[@sambenyaakov]

Please indicate time or slide number.

[@luzhouyang5414]

@@sambenyaakov LTspice PCM model at 9:41,thank you!professor.

[@sambenyaakov]

@@luzhouyang5414 it produces REF which is like the output of the voltage error amplifier. Here it is used as a signal to test tracing. E.g. kzbin.info/www/bejne/mGeapmaPaNFlbNE

[@biswajit681]

Hi sir how are you...long time no videos..

[@sanjikaneki6226]

is there a advantage in having +slope compensation in the current reading vs -slope on the reference? Also what is a good starting point for the slope? -1V /ms ? or this value is determined later with experimentation /simulation ? Regarding this mode of operation can it reach 100% duty cycle? asking since besides the current sensing method i dont see why not ( unless it is a FET driver limitation) ?

[@sambenyaakov]

Not really but you have to keep the signal within the voltage range.

[@paulpaulzadeh6172]

Hi Prof. Interesting results for LG. Wonder for LG for the current feedback amplifier. Can we use the same method as a voltage feedback amplifier? What about if you have one inner current feedback amplifier with one outer voltage feedback amplifier. I understand that your video is DC/DC . But what about the current feedback amplifier with outer voltage feedback? Thanks for answering in advance

[@sambenyaakov]

The method is general. The question is what loop are you are interested in.

[@ameerbaidusi1223]

Thank you!

[@sambenyaakov]

Thanks

[@bobby9568]

More ltspice please 🎉

[@sambenyaakov]

Tying

[@paugasolina5048]

awesome!

[@sambenyaakov]

Thanks!

[@thavakkodi6290]

Sir but the slope compensation method doesn't effective(precise and accurate ) for PCM used for all application.(specifically fr PCM in Current Limiting purpose in Solid state Circuit breaker application),

[@sambenyaakov]

Slope compensation is required in converters, buck boost, flyback, not in switches.

[@thavakkodi6290]

@sambenyaakov sir but i am using buck topology fr current limiting purpose and it also acts like converter during normal operation.

[@sambenyaakov]

@@thavakkodi6290PCM has advantages over voltage mode control and you do not need an external over current protection.

[@thavakkodi6290]

@@sambenyaakov ok sir thx

[@tamaseduard5145]

👍🙏❤️

[@sambenyaakov]

🙏🙂👍

[@giacomosammuri3531]

AI has reached a great level in emulating a human interaction and it is syntactically impeccable...but as some humans It invents or gives generic answers when the argunent is too specific and there Is not a clean database on the matter.

[@giacomosammuri3531]

I tested different AIs in this field and the feedback Is that for the Moment it's not a reliable tool at all!

[@sambenyaakov]

Indeed

[@sambenyaakov]

@giacomosammuri3531 Please don't insult us, the humans. ChatGPT's "intelligence" does not reach that of an ape. It is just Google plus. Just reciting information it gathred - at least in the engineering domain where inference and creativity is the name of the game.