Totem Pole PFC conduction/operation Mode

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
Hello,
I have to construct a SMPS. It should consist out of a Totem Pole PFC (with GaN) and should go from fow Power (approx. 10W) to 1.5kW.
But now I am a bit confused by the operation modes. I understand that there is CCM, CRM, DCR and that they have different advantages.
I found some reference designes for each of the modes but for the Totem Pole PFC with GaN mainly CCM.

But what confuses me is how they stay in the Mode. For the CRM I found that they change the Frequency.
Also are some SMPS designed to go from CCM (at high load) to DCM (at low load). In this case how should one focus on the specifications of the components?

Pls help me - Thanks!
 

Ian0

Joined Aug 7, 2020
9,667
Have you made a PFC front end before? If not, then a 1500W version is probably not the best place to start for a first attempt.
First, you need to know the distinction between CCM (Critical conduction mode) and CCM (continuous conduction mode), Critical conduction mode is a form of discontinuous conduction mode.
In DCM, the current goes from zero to twice the average every switching cycle, so it requires magnetic materials with low hysteresis losses, but the diode switches off when the current is already zero, so switch off losses can almost be ignored.
CCM current only reaches zero at mains zero-crossing, so diode switch-off losses dominate, and a iron powder core can be used because it has larger Bmax, but higher hysteresis loss.
Of course, because dI/dt depends on V and L but not I, CCM changes to DCM as the current reduces. The slope remains the same, but the starting point is lower, so it reaches zero before the end of the switching cycle.
DCM and Critical Current mode designs are easier to get stable.
Then a totem-pole switching stage adds another layer of complexity.
 

Papabravo

Joined Feb 24, 2006
21,157
Understanding how to deal with the issues you raise requires a good dela more reading and effort than is possible with a forum post. You need a graduate level textbook on Switch-Mode Power Supply Design. In fact, it might require multiple books to cover all of it. If you are ready to put in the effort and bear the expense, I have some suggestions.
 

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
Have you made a PFC front end before? If not, then a 1500W version is probably not the best place to start for a first attempt.
First, you need to know the distinction between CCM (Critical conduction mode) and CCM (continuous conduction mode), Critical conduction mode is a form of discontinuous conduction mode.
In DCM, the current goes from zero to twice the average every switching cycle, so it requires magnetic materials with low hysteresis losses, but the diode switches off when the current is already zero, so switch off losses can almost be ignored.
CCM current only reaches zero at mains zero-crossing, so diode switch-off losses dominate, and a iron powder core can be used because it has larger Bmax, but higher hysteresis loss.
Of course, because dI/dt depends on V and L but not I, CCM changes to DCM as the current reduces. The slope remains the same, but the starting point is lower, so it reaches zero before the end of the switching cycle.
DCM and Critical Current mode designs are easier to get stable.
I haven't done a PFC frontend before.
I looked the different modes up and I think I have understood them in theory. They merge into each other with the CRM beeing the boundary. But I still don't get how they choose to only operate in e.g. the CCM. Do they define a min Power? In several publications is stated that the CCM is always on a fixed frequency. For DCM both can be found and for the CRM its always stated that when operated in this mode a variable frequency is beeing used.

Then a totem-pole switching stage adds another layer of complexity.
I know, but as far as I understood the power components can be approximated with the formulas for a traditional boost converter. So the new complexity layer relates mainly to the control.
But the power components have to be designed to meet the control requirements.
 
Last edited:

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
Understanding how to deal with the issues you raise requires a good dela more reading and effort than is possible with a forum post. You need a graduate level textbook on Switch-Mode Power Supply Design. In fact, it might require multiple books to cover all of it. If you are ready to put in the effort and bear the expense, I have some suggestions.
I want to deep dive. I found several several books regarding power electronics and PFCs. Sadly none of them covers a Totem Pole PFC but as stated in my previous post it's heavily based onto the traditional Boost converter, so it should be fine for the beginning.
The main thing I currently try to wrap my head around is when which control is beeing used and when. So I can define the control Mode, on which I want to determine the power components. I think the control, with e.g. softstart, can be left out at first.
I would appreciate some suggestions.
 

Papabravo

Joined Feb 24, 2006
21,157
The basics of continuous conduction mode come first. You need to understand what conditions allow it to persist and what conditions require a mode change to DCM (Discontinuous Conduction Mode). CCM is when the load is not subjected to large changes in impedance. During the switching period the current rises and falls according to the duty cycle. Normally the switch is controlled by a flip-flop that is set at the beginning of each switching period, and the flip-flop is reset when the current in the load reaches a maximum value. The difference between the two (min and max current) is the output current ripple and it can be reduced to very small values with appropriate tradeoffs. This system works best at duty cycles in the neighborhood of 50%. It works less well at extreme duty cycles like 5% and 95%. You need to master the basics of voltage mode and current mode control before venturing further afield. Pick up the textbook you have and bone up on the basics.
 

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
The basics of continuous conduction mode come first. You need to understand what conditions allow it to persist and what conditions require a mode change to DCM (Discontinuous Conduction Mode). CCM is when the load is not subjected to large changes in impedance. During the switching period the current rises and falls according to the duty cycle. Normally the switch is controlled by a flip-flop that is set at the beginning of each switching period, and the flip-flop is reset when the current in the load reaches a maximum value. The difference between the two (min and max current) is the output current ripple and it can be reduced to very small values with appropriate tradeoffs. This system works best at duty cycles in the neighborhood of 50%. It works less well at extreme duty cycles like 5% and 95%. You need to master the basics of voltage mode and current mode control before venturing further afield. Pick up the textbook you have and bone up on the basics.
I have read a lot the last few days. It seems that the current mode control is more apropriate for the Totem Pole PFC. Due to the faster response to load change, the suppression of disturbing stationary and dynamic fluctuations of the input voltage as well as the easier to implement over current protection (which is needed anyway). Therefor, I tend towards a two-loop control circuit consisting out of a current loop (inductor current error) and a voltage loop (output voltage error).

It is suggested that there is not a continuous switching between DCM and CCM. In reality one tries to go through the DCM quickly while booting, or stayes in CCM.
When only working in CCM my PFC needs a min power specification or a base load, to be able to work with a fixed frequency.
DCM has the disadvantage, that it has the highest peak current. It's also not using the circuit efficiently. CrCM is a more common practice, where the operation is controlled to stay at the boundary. For smaller loads some switch from CrCM in the DCM is done with a hysteresis area. But for high power CrCM designs the resulting input ripple current impacts the input EMI filter quite heavily (can be lowered by using interleaving).
It should be noted, that the power stage equations and transfer functions for CrCM are the same as for CCM.

"In conclusion, we can say that for low power applications, the CrCM boost has the advantages in power saving and improving power density. This advantage may extend to medium power ranges, however at some medium power level the low filtering ability and the high peak current starts to become severe
disadvantages. At this point the CCM boost starts being a better choice for high power applications."

The critical inductance for the PFC can be decided by the boundary condition. Which is onefactor of the output current ripple.

Normally the switch is controlled by a flip-flop that is set at the beginning of each switching period, and the flip-flop is reset when the current in the load reaches a maximum value. The difference between the two (min and max current) is the output current ripple and it can be reduced to very small values with appropriate tradeoffs. This system works best at duty cycles in the neighborhood of 50%. It works less well at extreme duty cycles like 5% and 95%.
Do you mean here "the current in the inductor reaches a maximum value"? So it's working in Peak Current Mode Control.
 

Papabravo

Joined Feb 24, 2006
21,157
I have read a lot the last few days. It seems that the current mode control is more apropriate for the Totem Pole PFC. Due to the faster response to load change, the suppression of disturbing stationary and dynamic fluctuations of the input voltage as well as the easier to implement over current protection (which is needed anyway). Therefor, I tend towards a two-loop control circuit consisting out of a current loop (inductor current error) and a voltage loop (output voltage error).

It is suggested that there is not a continuous switching between DCM and CCM. In reality one tries to go through the DCM quickly while booting, or stayes in CCM.
When only working in CCM my PFC needs a min power specification or a base load, to be able to work with a fixed frequency.
DCM has the disadvantage, that it has the highest peak current. It's also not using the circuit efficiently. CrCM is a more common practice, where the operation is controlled to stay at the boundary. For smaller loads some switch from CrCM in the DCM is done with a hysteresis area. But for high power CrCM designs the resulting input ripple current impacts the input EMI filter quite heavily (can be lowered by using interleaving).
It should be noted, that the power stage equations and transfer functions for CrCM are the same as for CCM.

"In conclusion, we can say that for low power applications, the CrCM boost has the advantages in power saving and improving power density. This advantage may extend to medium power ranges, however at some medium power level the low filtering ability and the high peak current starts to become severe
disadvantages. At this point the CCM boost starts being a better choice for high power applications."

The critical inductance for the PFC can be decided by the boundary condition. Which is onefactor of the output current ripple.


Do you mean here "the current in the inductor reaches a maximum value"? So it's working in Peak Current Mode Control.
In CCM, there will be a minimum inductor current and a maximum inductor current, for any fixed value of load impedance. This is NOT the peak current that the supply is capable of producing, but it is the peak current implied by the load, and the desired, nearly constant, output voltage.
 

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
In CCM, there will be a minimum inductor current and a maximum inductor current, for any fixed value of load impedance. This is NOT the peak current that the supply is capable of producing, but it is the peak current implied by the load, and the desired, nearly constant, output voltage.
Do you have a picture of this control logic? I dont fully get it.

The necessary steps I have identified for a basic CCM control are:
1. Creating a desired demand current (in simulation this can be accived by calculating it with Uout/Rload *sin(2Pi*freq*Time))
2. Compare it with the present Inductor current IL (simply meassured)
3. The PWM is conducted by the difference between the desired and the present current
This can be done with a Hysteresis PWM Modulator. But than it has a variable frequency.
 

Papabravo

Joined Feb 24, 2006
21,157
Do you have a picture of this control logic? I dont fully get it.

The necessary steps I have identified for a basic CCM control are:
1. Creating a desired demand current (in simulation this can be accived by calculating it with Uout/Rload *sin(2Pi*freq*Time))
2. Compare it with the present Inductor current IL (simply meassured)
3. The PWM is conducted by the difference between the desired and the present current
This can be done with a Hysteresis PWM Modulator. But than it has a variable frequency.
The pictures, along with the explanatory text, I have are in: Basso, C., Switch-Mode Power Supplies, 2014, pp. 175-184

There may be .pdf articles you can find online that cover some of this material.
https://cbasso.pagesperso-orange.fr/Downloads/Divers/PCIM 2005.pdf

I do have the following simulation using the LTspice Control Library

1646924612740.png
 
Last edited:

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
The pictures, along with the explanatory text, I have are in: Basso, C., Switch-Mode Power Supplies, 2014, pp. 175-184

There may be .pdf articles you can find online that cover some of this material.
https://cbasso.pagesperso-orange.fr/Downloads/Divers/PCIM 2005.pdf

I do have the following simulation using the LTspice Control Library

View attachment 262457
Thank you for your Help.
I was able to make some steps into the right direction. The LTspice Control Library is a great tool.

But unfortunately I am not able to get the simulation running without a diode next to the switch. I think there is probably an issue with the current flowing back into the Inductor.
The following picture shows the circuit and the resulting waveform. Maybe one of you has an idea how to fix this problem.
1648548040153.png

I've modified the control to the following:
1648548097541.png

Note: The only thing I've found and why I think it has to do something with the current flowing back is the following:
As the ICE3PCS01G controller is originally designed for a classic or traditional PFC topology where a MOSFET is switching against a SiC Shottky diode, in order to fit this controller to the CCM Totem Pole topology, a fast comparator is needed to prevent a negative current flow in the PFC choke during DCM operation.
I have attached my LTSpice simulation files. The one called "PCF_Control - different" is the modified version of "PCF_Control". Where the initial Simulation produces the desired Output Voltage and Input Current wave.
 

Attachments

Thread Starter

PowerStarter

Joined Mar 4, 2022
16
My approach with the Comparator was a success. The simulation is now running. I had some minor errors in the initial simulation (e.g. mislabeling, etc.).
 
Top