LM25085: Low measured efficiency

Eric Garner

Part Number: LM25085
Other Parts Discussed in Thread: LM27403, LM25145

Hi, we are using a the LM25085 for the 3p3 supply in one of our designs and I'm getting about 86% peak efficiency, I'd like to do better. this is the measured efficiency of the design:

this is how it's implemented:

The X's over the test points are so they don't get included in the exported BOM,

Previous designs using this controller have had measured efficiency in the low 90% range, any suggestions on how to get this design into the same range?

Thanks

over 6 years ago

0 Eric Garner over 6 years ago

Prodigy 180 points

There is 100 uF of bulk capacitance in the 12V input that's not shown in the schematic snippet

0 David Baba over 6 years ago

TI__Mastermind 33090 points

Hello Eric,

Did you use the exact components that WeBench recommended? if not, which components did you divert from? did you select the recomended switching frequency Webench suggested?

If you implemented what Webench suggested, is the supply stable? Please check the switch node. Is the Switching voltage periodic in nature, are you able to easily trigger scope on it?

Hope this helps?

Thanks.

David.

0 Eric Garner over 6 years ago in reply to David Baba

Prodigy 180 points

Vin 11.5 -13V
Vout 3.3v nom
Iout 7A max

0 Eric Garner over 6 years ago

Prodigy 180 points

previous design was
Vin 11.5-13V
Iout 5A
Vout 3.3V

0 Eric Garner over 6 years ago in reply to David Baba

Prodigy 180 points

Previous Design:

D8=STPS340UF

Q2=FDC602PCT-ND

0 Eric Garner over 6 years ago in reply to David Baba

Prodigy 180 points

It was implemented as Webench recommended the only part substitution was D4 and L3, both of which came out of the webench alternates and looked fine in the webench simulations. when tested on the the bench, the circuit is stable both on its own and during load transiants

0 David Baba over 6 years ago in reply to Eric Garner

TI__Mastermind 33090 points

Hello Eric,

I am assuming that the new values of L3 and D4 are reflected in the efficiency curve above? can you please confirm?

If true, can you give me the Webench BOM so I can check efficiency manually at full load.

Thanks.

David.

0 eric garner over 6 years ago in reply to David Baba

Prodigy 160 points

Attached is the BOM exported from Webench. the inductance values in the chart are reflective of 2 iterations of the circuit. The one marked 2.7uH is the circuit with the attached BOM, the 15 uH curve is substituting L3 for a SER2915-153 and changing the values of R17, C17, C29 and R20 to their webench recommended values.

BOM_1067547_9.xls

0 David Baba over 6 years ago in reply to eric garner

TI__Mastermind 33090 points

Thanks Eric,

Please can you confirm that you used the part numbers within the attached BOM on your board?

Thanks.

David.

0 David Baba over 6 years ago in reply to David Baba

TI__Mastermind 33090 points

I assumed you were and I run some manual calculations assuming a rise and fall time on the switch node of ~20ns. I get an efficiency of around 88% at full load. Scrutinizing the plots again, looks like you are only a couple of point out.

If you want to improve the efficiency closer to 90% at full load, I recommend you consider using a synchronous Buck Converter because most of your losses are in the Diode followed (by the high side FET).

If you want to improve what you have, you can look at reducing the switching frequency, but you will only improve by a point or two.

Further things to realize when making comparisons from your current design to your previous design is the current is more and vout is less, this drives efficiency down because the diode is conducting for a long period of time with more current.

Hope this helps?

David.

0 Eric Garner over 6 years ago in reply to David Baba

Prodigy 180 points

I'll mess with the switching frequency and live with the losses. If I did decide to go with a synchronous buck, do you have a recommended part? We stuck with LM25085 because we use it elsewhere and it simplifies inventory. Our single overriding criteria is long market availability if that makes a difference.

0 David Baba over 6 years ago in reply to Eric Garner

TI__Mastermind 33090 points

Hello Eric,

I am assuming you Vin max is much greater than 12V? If so, please consider the LM25145 (42V max vin). If Vin max is below 20V consider the LM27403.

I hope this helps?

David.

0 Eric Garner over 6 years ago in reply to David Baba

Prodigy 180 points

Vin max is 13V. I'll look into the LM27403. thank you

Power management

Power management forum

LM25085: Low measured efficiency