LMR36520: Efficiency is worse than datasheet and webench

Hi Shinichi,

Your customer has choosen LMR36520FADDAR, which is the FPWM version. FPWM mode is fixed pulse frequency switching, which will bring higher loss at light load.

The efficiency graph shows the LMR36520ADDAR, which is PFM version. This mode will change switching frequency under light load, thus reduce loss.

I also simulate this circuit on Webench, the result shows similar with what your customer has tested: 

WBDesign8 (1).pdf

Feel free to reply if you have further concern.

Regards

Sirui

Hi Sirui,

Thank you for your reply.

The datasheet described with FPM version, we agree.

According to your webench design of LMR36520FADDAR, the effciency at load current 200mA is around 80%.This result is same as our attached Webench result. 

The efficiency at the customer is around 70%.  

The difference is 10%.

Request(1) Could you please let us know what is the cause of the inefficiency?

Request (2) Could you please confirm the efficiency of the datasheet.

---The efficiency in our circuit---
Input power: Vin 27.5V x Iin 46.8mA = 1.287W
Output power: Vout 5V x Iout 172mA = 0.86W
Efficiency: 0.86 ÷ 1.287 = 66.9%

Input power: Vin 19.6V x Iin 61.8mA = 1.211W
Output power: Vout 5V x Iout 171mA = 0.855W
Efficiency: 0.855 ÷ 1.211 = 70.6%

---

I appreciate your great help and cooperation/

Best regards,

Shinichi

Hi Shinichi,

We are sure about the accuracy of the datasheet data since it's validated before release, but currently we don't have validation data of LMR36520FA, we suspect there are two possible causes: the test method, or webench result too high in light-load.

There is one thing you can recommend your customer to do: test the circuit under mid-load or heavy-load, thus the device could work on PWM mode. Under this condition, if the test result is close to datasheet efficiency, then we can say the test method is correct.

If your customer cares light-load high-efficiency very much, we recommend to use LMR36520 w/o FA version.

Regards

Sirui

Hi Sirui,

Thank you for your reply.

I think it's a good way to investigate the difference of efficiency.

I'll share your suggestion with the customer.

Also if possible, could you please confirm the efficiency of LMR36520EVM.

Looking at the EVM's results, we can clarify whether Webench is too efficient or there is a problem with  the customer's board.

Or should we contact the Webench Team?

I appreciate your great help and cooperation.

Best regards,

Shinichi

Hi Shinichi,

Yes we can validate the FA version efficiency, but it may take more time since currently we don't have EVM nor device on hand.

I'll keep this post open, in case any of us has update.

Hi Sirui,

Thank you for your reply.

I understand your suggestion.

(1) I'll make the customer to confirm the efficiency of middle and heavy load.

When I obtain the data, I'll share about it.

(2) Could you please confirm the efficiency of LMR36520 FA version. It is a good information of us.

Also I'll ask the same question to Webench Team.

If I get an additional information, I’ll inform you.

I appreciate your great help and cooperation.

Best regards,

Shinichi

Hi Shinichi,

I notice you have post another similar thread on E2E. I suggest we keep the question on one thread for better tracking.

We have checked the validation data of FA version and it appears to be similiar with Webench simulation result. Now we need to check the test method and try to duplicate.

• Please keep ask your customer to confirm the heavey load efficiency, and provide the test method(eg. measure point)
• I've applied EVM for bench test, and it may take few weeks to verify.

Regards

Sirui

HI Sirui,

Thank you for your reply.

I understand.

I'll proceed with the confirmation in this thread.

You've already confirmed the validation data.

Can you send me the validation data?

I’d like to share it with customers.

Could you please send us the EVM test results when you obtain.

I’m looking forward to waiting your response.

I appreciate your great help and cooperation.

Best regards,

Shinichi

Hi Shinichi,

Sorry we can't provide the validation data since it's NDA restricted.

I'll update once I have EVM test result.

Regards

Sirui

Hi Sirui,

Thank you for your reply.

I understand you cannot provide the validation data.

How much difference was there between the validation data and the Webench data?

I think only a few percent, is it correct?

I'd like to report to the customer as much detail as possible.

I appreciate your great help and cooperation.

Best regards,

Shinichi 

Hi Shinichi,

Yes just a few percent, but it's tested under different condition(periphery components), and it will lead to different efficiency result.

By the way, how many devices have your customer find ~70% efficiency? all of the device or part of them? what is the ratio?

Regards

Sirui

HI Sirui,

Thank you for your reply.

The customer confirmed the efficiency at  result.

Could you please see below and advise.

The efficiency was confirmed with two prototype boards, and the same results were obtained for both boards.

①VOUT＝4.978V、IOUT＝212.6mA（load 22Ω）

　VIN＝24.0V、IIN＝64.93mA

　Efficiency＝4.978ｘ212.6／（24.0ｘ64.93）＝67.9%

②VOUT＝4.940V、IOUT＝897mA（load 5.5Ω）

　VIN＝24.0V、IIN＝219mA

　Efficiency＝4.940ｘ897／（24.0ｘ219）＝84.3%

The efficiency are 10%-15% lower than  the webench's efficiency.

Could you please let us know what is the reason for poor efficiency.

I appreciate your great help and cooperation.

Best regards,

Shinichi

Hi Shinichi,

Thanks for the update, I'll discuss internally and let you know once I get the answer.

Regards

Sirui

Hi Shinichi,

Thanks for the confirmation. It seems this device efficiency is lower than simulation over all load condition. I have some questions on measurement, please help to verify:

• Is the input/output current measured at the same time?
• Can your customer use electrical load to draw certain current from the device?

At the same time, we are out of EVM in stock, so it may be take longer time for us to verify in our side. If possible, you can ask customer to send their power board to us for debugging.

Regards

Sirui

Hi Shinichi,

Thanks for your patience with our customer.

Since then, we can close this thread and continue offline if your customer agree to rent the board.

Please open another thread if you have further support need.

Regards

Sirui