PMP22764: Output couldn't loading over 60% debugging

Muhsiu Wei36

Intellectual 340 points

Part Number: PMP22764
Other Parts Discussed in Thread: TL431, LM51551

Hi Team,

I following PMP22764 design a 28Vin/56Vo/100W flyback converter. The schematic is attached below.

28 to 55V schematic with CMC.pdf

During the EVT build, I find 3 of 480 pcs have issue which couldn't loading over 60% at output.

I check the Gate and CS signal, observe the PWM signal is unstable at 50-60% load (The duty will cahnge at each cycle) cause VCS has chance to trigger the 100mV threshold.

I trying to swap same components (Transformer, TL431, Photo-coupler, Rsense) to check if it's components failure. I also trying to modify the Rsense from 7mOhm to 5mOhm to see if it's design margin issue.

But, those changes doesn't work. The failure symptoms still exist.

I measure the impedance of the LM51551 pins between failures and good unit. They looks the same.

Please help to recommend debugging suggestion. thank you.

Muhsiu.

over 1 year ago

0 David Strasser over 1 year ago

TI__Intellectual 2430 points

Hello Muhsiu,

Here are some things to check:

1. C76 is 470pF on the original design. 1000pF may filtering the current sense signal too much.

2. Is capacitor EC5 on the output (68uF/80V) a standard aluminum with relatively high ESR? The original design used a low ESR (30 milliohm) Polymer Hybrid Aluminum. If so, the loop compensation may need adjusting.

3. Does the transformer have the same turns ratio and inductance as the original design?

Thanks,

David

0 Muhsiu Wei36 over 1 year ago in reply to David Strasser

Intellectual 340 points

Hi David,

Thanks for the reply.

1. I checked the Vcs signal during overload and SCP, converter can working normally and has stable trigger point around 2.1A-2.2A (output current).

3. The transformer design is correct. I swap the transformer with the good DUT, the abnormal still exist.

However, I has been resolved this issue. The root cause is the EC5 was wrong polarity installation by MFG. But, how come the circuit still can work with 50% load.

Thanks,

Muhsiu

+1 David Strasser over 1 year ago in reply to Muhsiu Wei36

TI__Intellectual 2430 points

Muhsiu,

EC5 was probably drawing a significant amount of current if installed backwards. Not only was this limiting the amount of load current available, it also affected the control loop by moving the output R/C pole higher in frequency.

Thanks,

David

0 Muhsiu Wei36 over 1 year ago in reply to David Strasser

Intellectual 340 points

David,

Yes, I can observe the Vcs and Vcomp signal carries low frequency harmonic around 30 KHz when the EC5 installed backwards.

Thank you,

Muhsiu

Simulation, hardware & system design tools

Simulation, hardware & system design tools forum

PMP22764: Output couldn't loading over 60% debugging