TPS23752: TPS23752 Efficiency at 5V 200mA

Carsten Bode

Prodigy 10 points

Part Number: TPS23752
Other Parts Discussed in Thread: TPS23758EVM-080

Dear Sirs,w

we tested the Evalboard for the TPS23752 with a load of 5V 200mA.

The efficiency is lower than expected.

Is there a way to tune the design for better performance at this load?

BR Carsten Bode

over 2 years ago

0 Michael Pahl over 2 years ago

TI__Mastermind 21320 points

Hello Carsten,

It is the nature of DCDC switching power supplies to be less efficient at light loads. This is because the duty cycle is lower, and therefore MOSFETs and diodes are off longer, translating to more power loss. Therefore light load will always be lower efficiency than higher loads for any switching DCDC supply.

The best topology for light loads and 25W is the diode flyback -- but they are only really good for 12V output designs. So there is just a limit to efficiency at light loads.

All that being said, you could get more expensive MOSFETs that have lower gate charge. The parts here are on the lower cost/performance side of the spectrum. Similarly you could get more expensive diodes as well.

You could also try to play with the snubbers to reduce losses there.

you could get a larger BJT for Q4.

You could also change the input diode bridge to MOSFETs.

If this post answers your question, please indicate so by marking this thread as resolved. Thank you.

Regards,

Michael P.

Applications Engineer

Texas Instruments

0 Carsten Bode over 2 years ago in reply to Michael Pahl

Prodigy 10 points

Hello Michael,

thank you for the detailed answer.

If we change the secondary side from 5V to 12V we gain at the POE side but loose on the secondary sides at the core voltages we need for our SOC and DDR.

Do you recommend a lower gate charge MOSFET for the primary side or for the sync rect. I guess we need it on the primary side. With light loads we change from sync rect to diode anyway. For the primary MOSFET we can try one with half the gatecharge of the SI7898 and a lower RDS-ON.

We can replace the input bridge with an ideal bridge, that should increase efficiency by approx 3%.

We tried to reduce the frequency, but that didnt help.

We didnt try so far to change the snubber.

BR Carsten

0 Michael Pahl over 2 years ago in reply to Carsten Bode

TI__Mastermind 21320 points

You can look at the FET on TPS23758EVM-080 or search on a vendor website by gate charge. Gate charge will have a bigger impact than Rdson, so prioritize that. Additionally you are going to need to switch FET package.

Yes changing the frequency will not help since the transformer inductance and feedback loop is based on the switching frequency -- you would need a serious redesign to get the benefits.

Let us know if you have any other questions.