LM5176: Circuit drawing excessive current under no / light loads causing poor efficiency

Hi Michael,

thank you for using the E2E forum.

The LM5176 does not support DCM like the LM5175.
With the used CCM mode therefore the losses are higher in light or now load operation.

This you can see in the loss calculation graph of the quickstart calculator - see the inductor loss ~ 0.5W at no load.

You can get lower with the core losses at no load by increasing the switching frequency but then the losses in the MOSFET will increase.

You might change to the LM5175 which supports DCM operation.

Both devices are Pin compatible but require some different external components, e.g  RT resistor or for LM5175: It is recommended to add 7.8V or 8.2V Zener diodes from SWx to BOOTx to stay below the abs max rating of 8.5V (layout dependent).

Best regards,

 Stefan

Hi Stefan,

Thanks for the reply and the suggestion for the LM5175. Can you help me understand the discrepancy between the estimated 0.5W loss at no load vs. the observed 3.5W on the actual circuit? Do you see any issues with the schematic as is?

I have the eval kit for this part and I observe that it draws very little power (<0.5W) with no load. Could you help me understand the tradeoffs that design has made to make that low quiescent current draw?

Thanks again,

-Mike

Hi Michael,

Stefan is out of office this time.
He will answer you tomorrow.

Best regards,

Johannes

Hi Michael,

thank you for checking back - i missed that the losses seems to be around 3.5W with no load - this looks really to large.

Can you check the output voltage at now load. Is this table or does it have some oscillation?

Do you have an option to check the inductor current?

Can you check the COMP pin voltage: is this stable or oscillating?

Note: I checked the schematic but could not see an issue here.

Best regards,

  Stefan

Hi Stefan,

The output voltage seems to be stable across the input of 12V-55V, ripple is around ~3% if I'm calculating it correctly.

COMP becomes unstable as the voltage approaches 55V.

At no load the current consumption is a constant 195 mA from 12V-55V. It doesn't vary with input voltage it seems.

Can you propose a method to measure the inductor current? The layout for the design follows the eval board closely, so not sure if there's really any easy way to measure this.

I think some of the ripple/noise on the COMP pin is from my test setup's long ground lead. I can try and get a better capture if needed.

Can you explain why the current consumption is so high even when COMP is stable at 12V in?

Thanks,

-Mike

Hi Michael,

Due to bank holiday, please expect a reply by Friday.

Best Regards,

Feng

Hi Michael,

based on the scope plots the output looks stable and the losses should not come from instabilities.

Other options for the higher losses would be the switching losses in the MOSFET where you have on some more as the number of MOSFETs used are higher.

Also the core losses of the inductor can have an influence as the inductance is smaller which gives higher peak currents in the inductor giving higher core losses. Losses through DCR of the inductor should be less as the used inductor has a smaller DCR value.

Best regards,

 Stefan

Hey Stefan,

I'm pretty certain the MOSFETs are not dissipating any excessive heat based on thermal camera images. It's clear to me it's localized almost entirely in the inductor itself.

Here's an image at no load, 55V in, after 3 minutes. The power dissipation during this time was ~10W.

The inductors were sourced as samples from Coilcraft, so I trust they are genuine. We see this across three of the samples, so I believe it is not an assembly issue.

Could you propose an inductor value that would reduce this current, but still satisfy the design parameter requirements?

Thanks,

-Mike

HI Michael,

if I understand the location of the thermal image and component mapping right the most heat is generated by the Inductor. As the heat also indicate losses this should be the critical component for the losses in this operation point. As already mentioned before the core losses of the inductor can be the reason for that. To reduce this an inductor with higher inductance can be tried out. As you can see the inductor on the EVM was also with higher inductance.

But keep in mind for inductor it is often the case that optimizing it for one point is making it worse on the other. Finding the right compromise is the challenge.

Best regards,

 Stefan

Hi Stefan,

I chose the 2.2 uH since it was used on the PMP21278 reference design from TI. I switched out to use a 6.8 uH inductor but it seems some other components need to be changed since the part wont start up and the output voltage isn't being regulated. I am having to move onto other projects for now so I don't think there will be more troubleshooting on this design. I at least wanted to give you an update on the project and will reopen this is any further troubleshooting is done.

Thanks,

-Mike