LM21305: Switching frequency - Light load

Hi Niels,

Your understanding is correct about Tonmin behavior. The typical value is specified in the datasheet is under 1.5MHz and voltage divider ration =1 condition. This is the min on time which the device can support. Tonmin time at 500kHz will depend on your application (Vin/Vout/Iout etc) and no limitation. Regarding Tonmax, you can refer to Toffmin and calculate Tonmax at 1.5MHz. 

Thanks,

Nancy

Thank you for your reply, however, this does not resolve my issue, nor does it answers any of my questions.

My design has Vin=8.5 V and Vout =0.8V and has an inductor of 2.2 uH. Switching frequency is set to 500 kHz. So, the Inductor ripple current should be approx. 670 mA. Further, according to the datasheet (section 8.3.4) the output current at the DCM/CCM critical conduction boundary is approx. 335 mA.

However, the design simulation, with a load of 500 mA (i.e. larger than the critical value) shows an irregular switching frequency. It appears that cycles are skipped since the controller does not go much below 400 ns in minimum on-time. The reduced switching frequency causes the inductor ripple current to increase. The controller is operating in DCM mode and the output ripple is much larger than expected.

For this design, reducing the input voltage to 5.0 V does not resolve the issue. Switching frequency is still much lower than 500 kHz. the inductor ripple current and the output ripple voltage is much increased.

My compensation network is very close the the recommended value and the values calculated by the Quick Calc excel sheet. In general, stability is fine even if I change the compensation network a lot. So this is not the issue.

The only real fix is a load current much higher that the critical value of 335 mA. Even reducing the input voltage to 5.0 V for this design does not cause the controller to operate in CCM mode at 500 kHz.

It appears that the controller uses some kind of pulse-skipping at 'light load' even though my minimum load is higher than the critical value. It also appears that the minimum on time of the controller is much larger than the typical value of 70 ns.

Regarding my question about the maximum value of the Ton-min parameter; Im not interested in Ton-max, but as stated the maximum value of the Ton-min parameter. The datasheet only specifies the typical value of Ton-min. Further, there is no specification of any dependence on the Fsw and the voltage divider ratio. I guess the specified ratio of 1 corresponds to the two feedback resistors being equal and thus the output voltage is approx. 1.2 V. But how does Ton-min depend on Fsw and divider ratio?

So, what can explain this behavior. Does the controller has some inherent pulse-skipping feature at 'light' load, even if the load current is above the critical value?

For my design the switching frequency is critical, since it determines my inductor ripple current and output voltage ripple. For some reason, the controller just does not behave as expected. So what is going on? What can be done to ensure a stable switching frequency down to the critical value of 335 mA?

Hi Niels,

Thanks for your detail information. I think you just need those numbers. Could you please send me your simulation schematic? What are input capacitor, output capacitor, compensation R/C value? What are ESR for input/output caps? I will run Pspice based on your schematic and debug the issue.

Thanks,

Nancy

Hi Niels,

Quick run your condition, you are right. The device enters DCM at 500mA load current which is incorrect. I will check further to confirm if there is something wrong in the model. Will update to you tomorrow.

Thanks,

Nancy

How can I send it to you? As far as I can see, this forum do not allow attachment of files.

However, I have put a zip file with the project in google drive here:

https://drive.google.com/file/d/1N78iKR2ETeAVdsol4aThZQLIp_azeawx/view?usp=sharing

Hope you can access that. Otherwise, please instruct how to send the project to you.

Hi Niels,

I didn't down load your zip file since I can duplicate the same issue from my side. Thanks.

I also verified the same condition on Simplis with same result. The behavior with 12Vin, 3.3Vout, 2.2uH, 500kz is expected. I am not sure the reason and report this issue to our model team today. Let's wait for their investigation. I will post here once I have a feedback.

Thanks for your patience!

Nancy

Hi Niels,

Our model team is still working on the issue. We test the real device on the EVM. The behavior is expected. So the model has bugs. Will get you update once we have the issue fixed.

Thanks for your patience!

Nancy