UCC28C58: UCC28C58 minimum PWM get stuck at 30%

Mathias Mahn

Part Number: UCC28C58
Other Parts Discussed in Thread: TL431

Tool/software:

Hello all together,

I want to build a flyback converter reference design (In: 200-1000VDC, Out: 24VDC 1.5A). I don't have much experience in flyback converters. So I created Designs with TI Webbench. I selected a Design using UCC28C58DR. I recalculated according the datasheet and build up the design with some little changes.

First Test showed that the design is working but was unstable. So I had to change the compensation to get stable PWM cycles. Efficiency Testing showed expected peek perfomance but the curves are somehow confusing with some ups and downs.

I was looking at the PWM signal at the source pin of the switcher to check operating modes of the controller. At 200V Input Voltage the design is working in CCM Mode with round about 60% PWM at full load. When lowering the load, it transitions into DCM Mode and then at light loads it stucks around 30% PWM. Further lowering the load leads to blanking of full PWM cycles. Output remains stable at 24V for the whole load range.

I'm confused about why PWM stucks at 30% and is not lowering further with decreasing load? I'm also confused about the feedback path in the design. According to the webbench I connected it to the Fb Pin of the controller. I thought it should also be possible connecting it to the comp pin with FB tied to GND. Because the error amplifier is made by the TL431 in the feedback path?SiC_Flyback_001.pdf WBDesign200to1000V_Diode.pdf

9 months ago

0 Hong Huang 9 months ago

TI__Guru 74115 points

Hi,

Did you mean the duty cycle changed from 30% to 0% when load changed at a point? It is likely the load change is not linear. Please provide details at what load is 30% and at what load is 0% while the output voltage still ok to be 24V. Please provide the waveforms of Q2 gate to GNDH, drain to GNDH, U1 CS pin, U1 FB pin, and U1 COMP pin, and VOUT to GND to help understand your description.

0 Mathias Mahn 9 months ago in reply to Hong Huang

Prodigy 20 points

Hi, I mean that when lowering the load at some point when lowering the load the duty cycle doesn't decrease anymore. Insteadt of that the converter is skipping pwm cycles. I found that this is due to the slope compensation circuit. When the Q2 is off, the slope compensation causes a increasing voltgage at the CS PIN when this equals a certain voltage next PWM cycle is skipped.

0 Mathias Mahn 9 months ago in reply to Mathias Mahn

Prodigy 20 points

CS Pin(blue), Voltage at RCS(yellow) at 0.3A Load without slope compensation R19 is removed

0 Mathias Mahn 9 months ago in reply to Mathias Mahn

Prodigy 20 points

First picture showing voltage at CS(blue) and the voltage at the RCS shunt(yellow) at 0.3A load it shows that it skipps cycles, at 0.4A it is continious. Without slope compensation (R19 removed) everything is fine.

So what could solve this problem, I need the slope compensation for duty cycles > 50%?

Best Regards

0 Hong Huang 9 months ago in reply to Mathias Mahn

TI__Guru 74115 points

Hi,

It looks the slope compensation is a bit too strong. You can change R19 value bigger to make the slope compensation a bit weaker to help solve this issue.

0 Mathias Mahn 9 months ago in reply to Hong Huang

Prodigy 20 points

I was thinking about that also. But how is it calculated correctly? I used the suggested value of the webbench design. When following the calculation in the datasheet at page 34 it would be even lower at about 2K2?

0 Hong Huang 9 months ago in reply to Mathias Mahn

TI__Guru 74115 points

Hi,

Not sure. You can provide your calculation in detail so I can check if anything can help this.

Usually the calculation is only for starting point then trial-and error tuning up.

0 Mathias Mahn 9 months ago in reply to Hong Huang

Prodigy 20 points

I increased R19 to about 6.8K. Higher values leading to instability when entering CCM mode at high loads. With 6.8K the issue is not solved completly. As expected it is moved towards lower loads starting below 0.2A.Wouldn't it be possible to use the gate drive signal of the mosfet switcher to build up some circuitry disabling the slope compensation when the mosfet is in off state?

0 Hong Huang 9 months ago in reply to Mathias Mahn

TI__Guru 74115 points

Hi,

What if further increase R19, say 24k or higher. As you mentioned open R19 can help then you need to try much larger value to find if the issue.can be improved.

0 Mathias Mahn 9 months ago in reply to Hong Huang

Prodigy 20 points

As I already wrote I tried higher values. Even 10k are leading to instability at high loads. So this is not an option.

0 Hong Huang 9 months ago in reply to Mathias Mahn

TI__Guru 74115 points

Hi,

If at higher load you get unstable, it looks increase R19 starts improve. You need to try further increase R19 to get weaker slope, say 24k., as open R19 helps resolve the issue per your observation.

Another thing you can try is page 5 of this below to add R22 and C21 to help reduce the slope compensation at light load

www.ti.com/.../sluucn1c.pdf

0 Mathias Mahn 9 months ago in reply to Hong Huang

Prodigy 20 points

Open R19 helps to get rid of the issue at light loads but leads to instability at high loads. So further increasing R19 will not help. I will try the configuration at page 5.

0 Hong Huang 9 months ago in reply to Mathias Mahn

TI__Guru 74115 points

Hi,

So increase R19 to be much bigger not open you may get a balance to cover both light load and heavy load.

You can still increase R19 to try but also ok if page 5 circuit can help resolve the issue.

Power management

Power management forum

UCC28C58: UCC28C58 minimum PWM get stuck at 30%