PFC Circuit issue using 28019

[sabu31]

Dear All,

I am experimenting on a PFC circuit using UCC28019. The system turns on after AC input of 125V . And output is around 400V. However, the input current is not correcting. THe load used is 280 Ohms. I am attaching the waveform.
The input current is in Red. The output voltage is in Green. The input voltage is in yellow.

Though I have designed for 2kW.
The current sense resistor used is 0.05Ohms. Is it current sense hitting upper limit.
Please let me know what should be done to troubleshoot the system.

 

[mtwieg]

Dear All,

I am experimenting on a PFC circuit using UCC28019. The system turns on after AC input of 125V . And output is around 400V. However, the input current is not correcting. THe load used is 280 Ohms. I am attaching the waveform.
The input current is in Red. The output voltage is in Green. The input voltage is in yellow.

Though I have designed for 2kW.
The current sense resistor used is 0.05Ohms. Is it current sense hitting upper limit.
Please let me know what should be done to troubleshoot the system.

Seems like poor loop compensation. Can you post the waveforms on the VCOMP and ICOMP pins?

 

[cupoftea]

please show your cct including compensation components for current and voltage loop......also, show Cout value...and Cin value. Then we can tell....it looks like the current loop is too fast.......slow down the outer voltage loop, then work on the inner currnt loop.....and be sure any overcurrent trip is not messing you up.

Also, what is inductor val?

without such, i doubt you can be helped.

 

[sabu31]

please show your cct including compensation components for current and voltage loop......also, show Cout value...and Cin value. Then we can tell....it looks like the current loop is too fast.......slow down the outer voltage loop, then work on the inner currnt loop.....and be sure any overcurrent trip is not messing you up.

Also, what is inductor val?

without such, i doubt you can be helped.

Initially the value of Cin was 100uF. Now its changed to 0.47uF. The updated waveforms is attached.

The PFC schematic is attached. The inductor used is 500uH. THough in design its mentioned minimum value should be 800uH. Presently the load is 500W.

 

[sabu31]

Hi all,

I increased inductor to 900uH. I am attaching the waveforms which includes inductor current

The inductor current is seemingly discontinuous (Ch2 waveform). I tried increasing the adding the previous inductor also . There is no improvement.

 

[mtwieg]

You didn't mention what each trace is in those scope captures.

 

[sabu31]

You didn't mention what each trace is in those scope captures.

Hi mtwieg,

The waveforms for the above snapshot are as follows

Yellow (Ch1): Input Voltage
Blue (Ch2): Inductor Current
Green (Ch3): Input AC current
Red (Ch4): DC output voltage

--- Updated Aug 3, 2023 ---

Initially the value of Cin was 100uF. Now its changed to 0.47uF. The updated waveforms is attached.

The PFC schematic is attached. The inductor used is 500uH. THough in design its mentioned minimum value should be 800uH. Presently the load is 500W.

In this the
Red Waveform is Input Current
Yellow is the Input Voltage
Green is Output Voltage

 

[mtwieg]

Seems like poor loop compensation. Can you post the waveforms on the VCOMP and ICOMP pins?

 

[cupoftea]

What you can do is put your max power dummy load on…..then put a diode to the vcomp pin , anode to vcomp…the cathode of the diode put to an opamp whose output voltage you vary…eg use it as a buffer……then start with opamp vout as low as possible…..then start up and apply mains……then wind up your vcomp pin voltage…slowly(!)……till you get your wanted vout………then from that point, play with the current compensation capacitor till you get a nice sinusoidal input currnt.

Be sure your sense resistor is not too small

Remember...you must connect it to the dummy load first, and keep the load on.

you will have to put the regulation voltage (actually just under it) into the vout divider input first, otherwise it wont startup....but dont use the divider...if voltage on vout divider input is out of range it will shut down...so read datasheet to find.

 

[Easy peasy]

it is likely that noise on the current pick up due to layout is the likely root cause of your ills

please post some photo's of the physical build

also please add 10 ohms to the gate drive to limit circuit noise

 

[sabu31]

Hi Mtweig,

After changing of capacitors/resistor and also the output capacitors. I am able to get the slightly improved input current albeit with noise. However, the voltage regulation is not working.
Presently, I am using variac to softly increase voltage.
I am attaching the figures with Input voltage at 230V and around 200V. Also I am attaching the schematic with updated parameters. I am not able to put probes on the Vcomp, Icomp pin as even connecting the ground probe of the passive probe leads control/system output going haywire.

The yellow signal (input Voltage),Blue signal (inductor current), Green Signal (Input current), Red Signal (Output Voltage). The values obtained from excel tool provided by TI is not helping.

(a) What should be done to improve regulation.
(b)Whats the procedure for tuning control.
(c)Can this issue be replicated in SPICE (TINA-TIA) or LTSPICE so that I can understand more about it

Thanks

 

[cupoftea]

Attached is your PFC done in open loop in free LTspice at max load, so you can see it.
It 240vac in, 2kw out. 65khz.
Inductor is 130uH.
One secret of PFC, is if you want it easier to stabilise, then dont make it so high in ccm....reduce the inductor value so you have more ripple current in it.....as long as your core losses dont get to high.

I trust that you know that a PFC shoudl be designed as a boost converter with mains peak voltage = vin.
Your UCC28019 is a cheapo PFC controller..it doesnt sense the instantaneous mains input bus....but infers it from the rise in current that happens in the inductor.....UCC28019 doesnt like DCM...so you tend to use it with a high value inductor...which isnt always the greatest thing..........LT1248 nice...but costly.....some of the texas are good....eg UCC3819 etc.

UCC28019 also is a bad 'un in that it has no peak current shutoff.

At 2kW, you coudl go for the ucc28070 types that do current transformer sensing....and with some inductor current "shape construction"

I take it you have some hvdc bus capacitance, and from the AC side aswell....i think you said you had...not much, but still need enough.

 

[cupoftea]

But i think that ucc28019 needs you to be in ccm otherwise it doesnt properly work...so maybe if you are in dcm at light load, the ucc28019 wont be able to do good PF correction.....after all it doesnt do "bus sense"

 

[Easy peasy]

This gives us no information whatsoever:

" After changing of capacitors/resistor and also the output capacitors "

listing the changes would be helpful to your cause

all boosters will enter DCM at light load - this is the physics of the situation - for this chip to work properly CCM is required at moderate to heavy loads - i.e. a larger boost choke as the freq is only 65kHz.

It appears the feedback is too fast, it is easy to use a very slow volt feedback on the bench as you control the Vin and the load - slowing the voltage feedback will allow a more sinusoidal input current.

Also you appear to have tons of switching noise - related to layout and over aggressive gate drive.

--- Updated Aug 6, 2023 ---

Further, the shown snubbers are not that great, try 220 ohm, 2W instead of 10k, and also add across the output diodes. Soldered right across each device - not miles away.

Also upping the Rg to 22 ohm for test is a good idea, with reverse diodes 40V 1A schottky and 4E7 for faster turn off.

--- Updated Aug 6, 2023 ---

also change, or add to C10, to 2.2uF 630VDC polypropylene, and have it close to the power circuit

--- Updated Aug 6, 2023 ---

similarly for the input put 2.2uF on the LHS of the sense R , up to the LHS of the choke - this supplies the pulse currents and smooths out the input ripple ...

--- Updated Aug 6, 2023 ---

also place 470nF MLCC 1206 50V right next to the IC for HF decoupling, as close to the Vcc and gnd pins as you can.

--- Updated Aug 6, 2023 ---

Try making R23 = 10k, lowering the gain of the volt loop.

--- Updated Aug 6, 2023 ---

Note the layout suggested in the data sheet:

--- Updated Aug 6, 2023 ---

and the values used in the suggested ckt:

--- Updated Aug 6, 2023 ---

Note that you will NEVER get good power factor with the 100uF cap in place .!.!.!.!

--- Updated Aug 6, 2023 ---

Also your output caps are too small for 2kW, 400V, you need around 470uF - 750uF in total

this is important for the feedback loop.

--- Updated Aug 6, 2023 ---

So - change C9 to 2.2uF 630VDC MKP and then you will be a lot further along ....

--- Updated Aug 6, 2023 ---

also:

https://www.ti.com/lit/ug/sluu272/sluu272.pdf?ts=1691280914850&ref_url=https%253A%252F%252Fwww.google.com%252F

for assistance

 

[Easy peasy]

Also - if you want to probe the Vcomp pins etc, use a 22k leaded resistor, SFR16, with 3mm of lead on the end to go to the pin and the other end goes to a x10 scope probe, this stops noise injection into the pin from your probe, and allows you to see what is going on.

 

[cupoftea]

, yes, and heres a doc on pcb layout too.

Yes all boosters enter dcm at light load indeed....and as you know, with lt1248 you can be dcm to a pretty highish point on the mains 1/2 cycle and itll still be ok....but ucc28019 wont like that so much.....ucc28019 will be harder to compensate.........and at 2kw...dont you think you need some peak current shut-off...as given by lt1248 and co.

Also, why not move across to the LT1248?.....you have the sim model of it above.......and it will allow you easier time in compensation, as you can use a smaller val inductor with lt1248...because lt1248 is well tolerant of dcm....in fact, the lt1248 is a similar one to many many of the other texas ones....the ones with v(bus) sensing.

 

[sabu31]

, yes, and heres a doc on pcb layout too.

Yes all boosters enter dcm at light load indeed....and as you know, with lt1248 you can be dcm to a pretty highish point on the mains 1/2 cycle and itll still be ok....but ucc28019 wont like that so much.....ucc28019 will be harder to compensate.........and at 2kw...dont you think you need some peak current shut-off...as given by lt1248 and co.

Also, why not move across to the LT1248?.....you have the sim model of it above.......and it will allow you easier time in compensation, as you can use a smaller val inductor with lt1248...because lt1248 is well tolerant of dcm....in fact, the lt1248 is a similar one to many many of the other texas ones....the ones with v(bus) sensing.

Thanks for the reply. I was thinking of LT1248. But there is no design note like TI appnote. I may have to be expert in TI TINA.

--- Updated Aug 7, 2023 ---

This gives us no information whatsoever:

" After changing of capacitors/resistor and also the output capacitors "

listing the changes would be helpful to your cause

all boosters will enter DCM at light load - this is the physics of the situation - for this chip to work properly CCM is required at moderate to heavy loads - i.e. a larger boost choke as the freq is only 65kHz.

It appears the feedback is too fast, it is easy to use a very slow volt feedback on the bench as you control the Vin and the load - slowing the voltage feedback will allow a more sinusoidal input current.

Also you appear to have tons of switching noise - related to layout and over aggressive gate drive.

--- Updated Aug 6, 2023 ---

Further, the shown snubbers are not that great, try 220 ohm, 2W instead of 10k, and also add across the output diodes. Soldered right across each device - not miles away.

Also upping the Rg to 22 ohm for test is a good idea, with reverse diodes 40V 1A schottky and 4E7 for faster turn off.

--- Updated Aug 6, 2023 ---

also change, or add to C10, to 2.2uF 630VDC polypropylene, and have it close to the power circuit

--- Updated Aug 6, 2023 ---

similarly for the input put 2.2uF on the LHS of the sense R , up to the LHS of the choke - this supplies the pulse currents and smooths out the input ripple ...

--- Updated Aug 6, 2023 ---

also place 470nF MLCC 1206 50V right next to the IC for HF decoupling, as close to the Vcc and gnd pins as you can.

--- Updated Aug 6, 2023 ---

Try making R23 = 10k, lowering the gain of the volt loop.

--- Updated Aug 6, 2023 ---

Note the layout suggested in the data sheet:

View attachment 184257

--- Updated Aug 6, 2023 ---

and the values used in the suggested ckt:

View attachment 184258

--- Updated Aug 6, 2023 ---

Note that you will NEVER get good power factor with the 100uF cap in place .!.!.!.!

View attachment 184259

--- Updated Aug 6, 2023 ---

Also your output caps are too small for 2kW, 400V, you need around 470uF - 750uF in total

this is important for the feedback loop.

--- Updated Aug 6, 2023 ---

So - change C9 to 2.2uF 630VDC MKP and then you will be a lot further along ....

--- Updated Aug 6, 2023 ---

also:

https://www.ti.com/lit/ug/sluu272/sluu272.pdf?ts=1691280914850&ref_url=https%253A%252F%252Fwww.google.com%252F

for assistance

Thanks Easy peasy for the detailed reply.

As you mentioned, the input capacitors were changed to 0.33uF.
The output capacitor is still low and I need to change it.

After the change in compensator values as suggested by TI support. I was able to get it running.

Is the HF noise in input current due to probe bandwidth (70khz, Keysight clamp probe).

I will change parameters suggested by you and report the changes in the performance.

The updated compensator values are

Cvsense=C24	1000pf
Cicomp=C19	1000pf
Cvcomp=C25	10uF
Rvcomp=R23	13k
Cvcomp=C23	1uF
Rvins2=R25	100k
Cvins=C20	33,000pF
Rfb2=R24	12.5k

 

[cupoftea]

Its just common mode noise. Maybe unhook the other three probes to see if its still there...or put a torroid round the lead.....or , if isolated input...then put a sense resistor in neautral, and measure v across that with a scope probe...or better still, a homebrew coax probe....but only if isolated input.

 

[Easy peasy]

We cannot see your layout - so hard to comment on the switching noise - lack of proper snubbers, overly aggressive turn on of the fets, power layout far too "open" - lots of ways to make a noisy ckt.

in spite of your attentions, the thing seems to be nearly working now

--- Updated Aug 8, 2023 ---

0.33uF for Cin is still too low for 2kW, add another at least

--- Updated Aug 8, 2023 ---

Also your PFC choke could be a lot bigger, double the uH's in fact.

 

[Easy peasy]

What is the rated ripple current for the 0.33uF C9 ?