UCC28180: Regarding the distorted supply side current waveform.

Hello SMPL, 

I agree, the current waveforms are definitely distorted, and in an unusual way.  
I see that the current waveform is shown on a 5V/div scale.  How is the this current being probed, and where is the probe located? 
What is the conversion from V to A, please? 

The AC voltage is shown at 200V/div, and the peaks look like ~300V (not sure exactly)   230Vac input has peaks to 325V, so either you are losing almost 25V between the AC source and the PFC input or actual input voltage is less than 230Vac.  This seems like a digression from the issue, but I want to be sure of what the peak current should be for a known input voltage and accounting for conversion losses.  

In the schematic diagram, your VSENSE divider values program Vout = 443V.  Is this intentional?  If not, what Vout do you require? 
What is the full required input voltage range for your PFC (such as 170Vac to 265Vac, or other range) and maximum Pout required? 

Does your schematic accurately reflect the actual values of all the components shown, or have any of them been changed on the board being tested?
There are several improvements to your PFC circuit that I can identify, but none of them pertain to THDi so I will refrain from discussing them until after the main problem is solved. 

Can you please describe the type of boost inductor used here?  Is it made with a gapped-ferrite core or with powdered metal?  If powdered, what is the L vs. I curve, please?
Is it possible to insert a current probe to measure the inductor current directly?  

I'm looking forward to your replies.  I know I asked many questions.  For better support, please don't forget to answer all of them.

Regards,
Ulrich

Dear Ulrich,

Thank you for your quick reply.

Let me first Clarify you about your queries.

1. The Current waveform is measured using a Current Probe of 10x Attenuation. We connected a EMI Filter before PFC. The Probe was located at the Phase Line at input of EMI Filter.

2. The input AC Supply was provided from a Auto Transformer. I am attaching new measurements today where measurements are done at 230V.

3. The Vout is required is 400V. We do not wish to boost upto 432V. Based on measured Vout our Engineer calibrated the divider.

4. We have plan to use this design in different applications 

     400W : Line 85-270V

     800W :Line 85-270V

    1600W : Line 170-270V.

Currently we are focusing on 1600W Design. We have been using our PFC using UC2854B IC, and we get satisfactory results.  Now we wish to reduce component count and cost. The PCB design is almost similar. We are getting very good THD in that Board. The board is being used in our production. So I am unable to share the gerber in forum but I can share it to your email if required.

5. We are using Powder Iron Core Inductor. 

Manufacturer : POCO

Core PN:NPS250125

Before Coating:OD×ID×HT=62×32.6×25 mm

After Coating:OD×ID×HT=63.09×31.37×26.26 mm

AL Value:400 nH

Path Length:14.370 cm

Ae:3.675 cm×cm

Core Volume:52.81CC

Inductor is designed with 29T with 34SWGx84 Wires Parallel. Inductance = 336uH

Below is the DC bias graph

Please find the attached file with More Measurements on 800W and 1600W Design. We have measured Inductor Current directly.

Please note we changed out shunt value to 1 miliOhm to improve current Waveforms.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

The measurement results are attached in the following File.

UC28180.docx

Updated Schematics with the current test setup

PFC Board_RND_0.00_PS

PFC Board_RND_0.00 SCH

Looking for your valuable feedback.

SMPL Team

Dear Ulrich,

The formula you have used for Calculating H field should not contain mu in it. 

The manufacturer software also calculates the inductance around 194 uH at 14A Peak. The dc bias graph also predicts similar. Also the current amplifier is disconnected and we are taking current feedback directly from shunt. (R8,R7 ,R13 is not populated and R9 = 0R and R14 = 220R)

Core PN:NPS250125
Core Number:1
Frequency:100 kHz
RMS Current:10 A
Peak Current:14.14 A
Ripple Current:2.828A
Wire:1.8mm×2Pcs
Turns:29
Availabe Core Window Area::19.0934%
Current Density:1.96 A/mm^2
DCR:9.09mO
================================================
L@0A:336.4µH
L@10.00A:230.88µH (H=25.36Oe)
L@14.14A:194.00µH (H=35.86Oe)
================================================
Core Loss:8.848194 W
Copper Loss:0.909W

Total Loss:9.757

Hello SMPL,

Uli is out of office so he will respond by the end of the week when he returns. In the meantime, can you please clarify your question and exactly what you need?

Regards,

Jonathan Wong

Hello SMPL team, 

You are correct, my formula for H should not have mu in it. I consulted a faulty reference. My mistake.  

After looking at the boost inductor current waveforms again, it looks like there is a shift from DCM to CCM and vice-versa at the locations where the AC current abruptly increases and decreases.  

Please examine the ICOMP signal when the DCM/CCM transitions occur.  ICOMP is the filtered output of the current amplifier and it feeds into the PWM comparator.  If the ICOMP wave-shape does not follow the input voltage shape, then that will translate into high distortion.

Your original schematic shows C28 = 2700pF.  Make sure that cap is working properly, without significant reduction from DC-bias effect.

In the previous WORD document, I noticed that there is a connection between R9 and R6 that does not appear on the schematic diagram. 
Please check that out, if it is intentional or a mistake. 

Regards,
Ulrich

Ulrich,

We found that is a PCB Mistake. We will cut the track. However since Rf and Cf are open in our case, only 2700pF Connected. We will measure ICOMP voltage and share it with you.

Since this is a feedback pin, is it OK to probe it? Or we should buffer it and check?

Hello SMPL Team, 

Yes, it is okay to probe the ICOMP pin directly. 
I recommend to use the "Tip & Barrel" method to avoid picking up switching noise which can obscure the signal.  

Regards,
Ulrich

Hello SMPL Team, 

It has been over 3 weeks since your last posting. 
Have you solved your problem?  
If so, please let me know so I can close this thread. 

Regards,
Ulrich

Dear Ulrich,

After all the suggestion and implementations, the current distortion is still present.

Regards,

SMPL Team

Hello SMPL Team, 

I was waiting for you to provide ICOMP waveforms that correspond to the different ac current distortions at different line and load conditions.  
Your post of May 6th indicated that you would measure ICOMP and share the waveforms.  

Please use 1ms/div sweep and 0.5V/div for ICOMP, 100V/div for rectified input RECT+, 2A/div for AC input current. 
Place the zero-references on the same horizontal graticule line.  Focus on the zero-crossing region.

Regards,
Ulrich

Hello,

We have tried to measure waveforms on the ICOMP pin but have faced a few problems:

1. Tried to measure the waveform directly with a probe in the "Tip & Barrel" method (as suggested).

    Result: The input (AC) line fuse (rated at 6A) burned at an output voltage of 50Vdc (without load).   

2. Tried to measure the waveform by using an opamp (in Unity gain buffer mode with 10Kohm resistor in series).

    Result: PFC didn't boost the voltage.

3. Tried to measure the waveform by using an opamp (in Unity gain buffer mode without any series resistor).

    Result: The input (AC) line fuse (rated at 6A) burned at an output voltage of 50Vdc (without load).

Supply Voltage: 230Vac, 50Hz

Required Output Voltage: 400Vdc

Hello SMPL Team, 

The symptoms you describe as a result of probing the controller strongly indicate that your AC source is not isolated.  

The oscilloscope probe ground is connected to the 'scope chassis ground which is directly connected to earth-ground through the line cord.
If you are powering your PFC directly from the building AC outlet or through a non-isolated variable transformer (such as Variac), then the 'scope GND will short out the AC input through the PFC diode bridge and, in the best case, blow the fuse.  Often the diode bridge is also damaged.    

You cannot safely probe the primary side of any converter (PFC or otherwise) unless you are using an isolating probe OR the AC source is isolated.
In my experience, HV isolating probes distort low level signals, so I always use normal X1 and X10 probes for debugging and use an isolated AC source to power the circuit.  Whatever node the probe GND is connected to will assume earth-GND potential.  

AC power can be sourced from an electronic AC supply (with input to output isolation) or using a  50Hz 1:1 isolation transformer between the wall outlet and the PFC input.  You must have the isolation. 
And when probing several signals at once, the probe GNDs must all be connected to the same reference node.  You can't have two different reference GNDs or damage will result.  

Regards,
Ulrich

Hello Ulrich,

Thanks for your advice to keep the isolation. 

Here, I have attached the waveforms for 400Watt, 800Watt and 1600Watt.

Channel 1: Input Voltage (230VAC, 50hz).

Channel 2: Input Current.

Channel 3: Voltage measured at ICOMP pin.

1. For 400Watt:

2. For 800Watt:

3. For 1600Watt:

Regards,

SMPL Team

Hello SMPL Team, 

Thank you for the ICOMP waveforms.  They are not usual, in the sense that normally, the ICOMP signal more closely resembles a rectified sine current with valleys near the zero-crossings.  

Instead, these ICOMPs seem to have a DC "floor" level of about 1V or higher.  
The "floor" seems to correspond to areas where the inductor current may be operating in DCM instead of CCM. 
Lighter load has more DCM and higher "floor" voltage.

For an experiment, please connect two inductors in series to see if the higher inductance lowers the "floor" voltage. 
I do expect higher L to reduce the amount of DCM operation for the same output power levels (400, 800, 1600W).  

While you have the 2nd inductor installed, may I also request a screen capture of a single switching cycle of inductor current taken at the peak of 230Vac input with 1600W output, please?   

Regards,
Ulrich

Hello Ulrich,

We have DOUBLED the inductor value but still facing the same issue.

Here, I have attached the waveforms for 400Watt, 800Watt and 1600Watt.

Channel 1: Input Voltage (230VAC, 50hz).

Channel 2: Input Current.

Channel 3: Voltage measured at ICOMP pin.

1. For 400Watt:

2. For 800Watt:

3. For 1600Watt:

Regards,

SMPL Team

Hello SMPL Team, 

There is a very slight reduction in the "DC floor" level in each screen shot, but I'm not sure why or if it is significant. 

But I do think that the cause of the DC level factors heavily into the poor THDi. 
I don't have an answer yet and have to think more about this. 

Can you please provide an updated copy of the schematic diagram, as your circuit is implemented today? 
I would like to be sure that I am working with the latest and correct information. 

Regards,
Ulrich 

Hello Ulrich,

Sorry for the late reply.

We are using the same Schematic, provided before.

[Schematic: PFC BOARD RND 0.00 SCH (2).pdf]

This time we have changed the inductor value only.

Regards,

SMPL Team

Hello SMPL Team, 

Thank you for resending the schematic file.  I am glad that nothing has changed (except for the inductance). 

At this time, I can think of only one possibility for the "DC floor" voltage on ICOMP: I suspect some negative input offset voltage on ISENSE. 
The schematic diagram indicates that several components on ISENSE are not connected (NC) and should not cause any issue.

But I request that you please verify that your board is actually depopulated according to the schematic.
In fact, I request that you please remove all of the parts that are encircled, even if you are certain that R13 is not loaded.

Also, please remove D4 from your board as well, and verify that C29 is 1nF and R14 is 220 ohm, not some unwanted values. 

I wish to eliminate the possibility that anything is affecting the ISENSE signal from R1 (10mR) except the R14-C29 filter. 
And please verify that the "GND_PFC" net on the controller page of the schematic is directly connected to the "PFC_GND" net on the power stage page of the schematic.  

I probably should have though to ask this a long time ago:  Have you tried replacing the UCC28180 IC with a new one? 
I ask in case there is some permanent damage done to the IC due to ESD or board-debug probing error or something.  

Regards,
Ulrich