LM5085: lm5085: Broken issue at high temperature test

Here are the results from today's experiment. The PWM signal of the LM5085 is behaving strangely on some of the samples. Instead of adjusting the PWM duty cycle depending on the load, the value is controlled through burst operation, while keeping the duty cycle fixed at 50%. Do you have any insight into this phenomenon?

Hi Junghyun,

Please fill out this excel spreadsheet with your design parameters. It will help me get a better idea on the issue.

Regards,

Henry

2451.LM(2)5085_Quick_Start.xls

Hi, Henry

I have written and attached the file you provided. I used the current sense type1 structure and the type3 minimum ripple configuration. However, I was unable to modify it because I couldn't input the value of the R3 resistor. The actual value of the R3 resistor is 56 kohm. Additionally, I designed it by connecting a small RC filter in series with ground to the D1 diode cathode node. The values of the components in the RC filter are 100 pF for the capacitor and 6 ohm for the resistor.

Thank you for your help!

LM5085_Quick_Start.pdf

Furthermore, the PWM signal waveform differs between each sample under light load conditions. I believe the waveform of sample2 is normal, but there are peculiarities with the waveform of sample1. There is no direct correlation between the waveforms of sample1 and a faulty IC, but could this variation potentially provide a clue or hint?

This figure is more accurate.

Hi Junghyun,

I believe the R3 value in your calculator was not computing because the minimum VIN value was too low. If you change to 14V, the R3 value is able to be computed.

I need some further information on your schematic, please provide this. I can determine if it is an input transient issue interfering with VCC or not.

I'm assuming the failure is permanent with the VOUT falling to 0V and PGATE remaining at 24VIN even when you bring the device back to lower temperatures or a lighter load? In addition, please detail what VIN, load current configurations cause the strange PWM signal. I'm a bit confused on what you mean by "sample". Are you referring to two different ICs?

Regards,

Henry

When I mentioned a "sample," I was referring to another IC. I'm currently discussing broken or defective samples. There are boards that operate normally with the same schematic. Through failure analysis, we want to determine if there are any vulnerabilities in the circuit. The failure samples are permanent. The load current flows at least 10mA. The load setting at the time of failure test was close to the minimum. I would estimate it to be less than 50mA. The schematic for the VIN side is attached as a picture.

In addition, according to the Excel sheet you provided, the recommended minimum L1 for Imax 3A is 33uH, which is too large. However, we are using 15uH. Please confirm if this is acceptable.

And the occurrence of the PWM operating in bursts seems to be a problem caused by the abnormal feedback of the ripple reduction scheme.It is believed that a significant amount of PMOS switching occurred momentarily in the error amplifier due to the large ripple loaded into the FB. (Is this analysis valid?)  

Thanks,

Jugnhyun Park

Thank you for your clarification. I will take a look at the issue.

Regards,

Henry

Hi Henry

I still don't understand. Can VCC be a negative voltage even if VIN is fluctuating around 8V? Is the diode included to prevent voltage droop (or undershoot) that happens during VCC switching between ground to regulator's output or vice versa? I have attached the complete schematic of that section. Please analyze it.

Thanks,

Junghyun

In addition, the followings are FB node voltage at 10mA load.

Hi Junghyun,

The VCC becomes a flying Cap and peak charges if VIN goes below the VCC regulation point at 7.7V. Input transients can cause 0V on VCC, so you would need a Schottky to clamp if the VCC node flies negative.

In addition, please upload a higher quality image of your scope shot, preferably taken directly on the scope instead of from a phone.

From what I can see, your FB ripple may be too high and it might seem like you need to adjust your ripple injection network, but I am not sure.

If you need to send at full resolution please email me at h-kou@ti.com.

Regards,
Henry