Active PFC - summary of all efforts to reduce switching noise

[mike buba]

Hello,

to summarise all the efforts from several of the treads this year (1, 2 and 3), I was finally able to get to higher DC voltages and powers without the 'noise' interfering with the serial communication. Thank you all who helped with advice and recommendations.
I have done the following:

1. Used aluminium plates to connect DC-link capacitors instead of cables
2. Added snubber capacitor and resistor (Csnub and Rsnub)
3. Added output capacitor (Cout)
4. Added SiC Shottky diode in parallel with IGBT chopper inverse diode
5. Used SiC Shottky diode instead of the IGBT chopper forward diode
6. Increased gate resistor values Rgon and Rgoff
7. Increased serial communication baud rate

Final schematics:

Pictures:


The first oscillogram shows input line current (Ch4, 10 mV/A), inductor output current (Ch3, 50 mV/A) and DC voltage (Ch2, 1/200 mV/V).
The second oscillogram is with the AC coupling for DC voltage
Both are recorded for Vout = 650 Vdc and Iout = 4.5 Adc (approx. 2.9 kW)


I still have some noise in the current signal, though. The frequency of that noise corresponds to the chopper switching frequency (20 kHz).
Not sure which, out of all seven, was The One that reduced the noise to an acceptable level and after which the communication didn't break.
Do you have any other suggestions for further improvements or this is good enough?

Thanks

 

[cupoftea]

just had another thought...you can make boost PFCs less noisy by reducing overlapping in the boost inductor...ie, reduce interwinding capacitance.

 

[Rahul_d]

Hello,

to summarise all the efforts from several of the treads this year (1, 2 and 3), I was finally able to get to higher DC voltages and powers without the 'noise' interfering with the serial communication. Thank you all who helped with advice and recommendations.
I have done the following:

1. Used aluminium plates to connect DC-link capacitors instead of cables
2. Added snubber capacitor and resistor (Csnub and Rsnub)
3. Added output capacitor (Cout)
4. Added SiC Shottky diode in parallel with IGBT chopper inverse diode
5. Used SiC Shottky diode instead of the IGBT chopper forward diode
6. Increased gate resistor values Rgon and Rgoff
7. Increased serial communication baud rate

Final schematics:
View attachment 175406
Pictures:
View attachment 175403 View attachment 175404 View attachment 175405

The first oscillogram shows input line current (Ch4, 10 mV/A), inductor output current (Ch3, 50 mV/A) and DC voltage (Ch2, 1/200 mV/V).
The second oscillogram is with the AC coupling for DC voltage
Both are recorded for Vout = 650 Vdc and Iout = 4.5 Adc (approx. 2.9 kW)
View attachment 175401 View attachment 175402

I still have some noise in the current signal, though. The frequency of that noise corresponds to the chopper switching frequency (20 kHz).
Not sure which, out of all seven, was The One that reduced the noise to an acceptable level and after which the communication didn't break.
Do you have any other suggestions for further improvements or this is good enough?

Thanks

Do not see pre filter capacitor after bridge rectifier in your schematic, please add a film capacitor after bridge rectifier to support & filter the high frequency switching currents.
the pre filter capacitor connection loop to inductor & MOSFET source should be as small as possible.

The value of the capacitor will depend on your max output power.

 

[cupoftea]

Also, if you have a low side sense resistor, be wary of putting a too big capacitance grounded to downstream of it...the sense resistor must "see" the current in its decent fullness.

 

[Rahul_d]

Also, if you have a low side sense resistor, be wary of putting a too big capacitance grounded to downstream of it...the sense resistor must "see" the current in its decent fullness.

If you have low side current sense resistor then the pre-filter capacitor should be connected to bridge rectifier negative (not on the source of the MOSFET) (i.e. between Bridge rectifier positive - closer to inductor & bridge rectifier negative)

 

[mike buba]

just had another thought...you can make boost PFCs less noisy by reducing overlapping in the boost inductor...ie, reduce interwinding capacitance.

I am using a pot core inductor; it looks something like this:



Can you please expand a little on this? I wasn't able to find any references or how to do it.

Also, if you have a low side sense resistor, be wary of putting a too big capacitance grounded to downstream of it...the sense resistor must "see" the current in its decent fullness.

Not sure that I understand that as well. I have LEM LA 55-P as the current sensor and LV 25-P/SP5 as the voltage sensor.

--- Updated Apr 12, 2022 ---

Do not see pre filter capacitor after bridge rectifier in your schematic, please add a film capacitor after bridge rectifier to support & filter the high frequency switching currents.
the pre filter capacitor connection loop to inductor & MOSFET source should be as small as possible.

The value of the capacitor will depend on your max output power.

Are there any calculations available on how to determine the capacitance? In the numerical simulations, it seems anything below 1 uF or 0.1 uF should be ok, but I am not seeing a switching noise in my simulations, so it is hard to know for sure and measure current. My application is 3 kW max output at 650 Vdc.
For example this one? Between inductor input and chopper DC minus.

 

[Rahul_d]

I am using a pot core inductor; it looks something like this:

View attachment 175412

Can you please expand a little on this? I wasn't able to find any references or how to do it.


Not sure that I understand that as well. I have LEM LA 55-P as the current sensor and LV 25-P/SP5 as the voltage sensor.

--- Updated Apr 12, 2022 ---

Are there any calculations available on how to determine the capacitance? In the numerical simulations, it seems anything below 1 uF or 0.1 uF should be ok, but I am not seeing a switching noise in my simulations, so it is hard to know for sure and measure current. My application is 3 kW max output at 650 Vdc.
For example this one? Between inductor input and chopper DC minus.

you will need at least 3uF to filter noise & meet differential EMI at the same time.

you can calculate capacitor based on your inductor pk-pk ripple current & required EMI specs to meet.
(also make sure the film capacitor can handle the RMS ripple current.)

--- Updated Apr 12, 2022 ---

I am using a pot core inductor; it looks something like this:

View attachment 175412

Can you please expand a little on this? I wasn't able to find any references or how to do it.


Not sure that I understand that as well. I have LEM LA 55-P as the current sensor and LV 25-P/SP5 as the voltage sensor.

--- Updated Apr 12, 2022 ---

Are there any calculations available on how to determine the capacitance? In the numerical simulations, it seems anything below 1 uF or 0.1 uF should be ok, but I am not seeing a switching noise in my simulations, so it is hard to know for sure and measure current. My application is 3 kW max output at 650 Vdc.
For example this one? Between inductor input and chopper DC minus.

Is this a CCM (Continuous conduction mode) design or BCM (Boundary conduction mode design)?
If CCM I assume you are using high flux pot cores (high flux , cool u , MPP material etc)

If your design is BCM, that will need even higher capacitance at pre filter location due to very high ripple current.

 

[cupoftea]

Hi,
Your PFC is simulated here in the free LTspice sim as attached.......you can run it to find your inductor 20khz ripple current etc etc....have a look at changing the input capacitance etc etc

--- Updated Apr 12, 2022 ---

Can you please expand a little on this? I wasn't able to find any references or how to do it.

...its just that you will be drawign spikes of current through the interwinding capacitance when the fet switches on, so its bound to create noise.......not saying its a disaster but just one of those things,,,,,,,,,,if you'd used a torroid with spaced out windings you may have got less interwind cap'ce...but your pot cores doubtless give you other advantages......

Did you not fancy some interleaved parallel PFC?....Using cheap current sense transformer...eg with UCC28070A?

Did you make the Litz yourself?....beware the insulation on it...some served Litz isnt too good on insulation

 

[mike buba]

Do not see pre filter capacitor after bridge rectifier in your schematic, please add a film capacitor after bridge rectifier to support & filter the high frequency switching currents.
the pre filter capacitor connection loop to inductor & MOSFET source should be as small as possible.

The value of the capacitor will depend on your max output power.

For test purposes, I added 2uF in the circuit. Even without a capacitor, I didn't see any high-frequency switching signal, i.e. 20 kHz ripple current.
I see only pulses at turn-on and turn-off instances.

I increased Rgon and Rgoff by 50% (from 23 Ohm (default) to 33 Ohm; action 6) from my first email), but this doesn't seem to help. All the actions 1) - 6) were done to remove that turn-on and turn-off noise, but it is still there.

 

[cupoftea]

I think that noise i see on your last is just the normal common mode noise you always get "induced" into the scope lead with any SMPS.....you coudl use a high frequency probe if you want....and that will reduce it a bIt....wrap the scope lead round a torroid, etc etc