Hi sorry about the issue. Could you please provide the schematic and layout of the design for review (in pdf format)?

The SW waveform appears reasonable.

This post is the part II of another post where all infos have been given. The previous post has been locked and I could not add any reply.

See: LM5575: Failure after 3 years of usage - about 7000 hours

Hi Thierry, 

I located your previous post. e2e.ti.com/.../4744252

Do you have the BOM and the additional waveforms that we previously requested?

I will be working with my colleague to review the materials with you and hopeful locate some things that can be potentially improved to better follow datasheet best practices.

This is the boom:

BOM_410078_G500_CPU_DRIVER.xlsx

Regarding the DC input and DC output there are flat signals no disturbances there.

Silk Screen Top Component:

Component Layout - Top Side.pdf

Schematics:CPU-DRIVER_5-7.pdf

Hi Thierry,

Thank you for providing this information!  

Could you also provide the waveforms for VIN/VOUT/SW on a working unit?

Regards,
Carol

Hi Carol,

The requested information has already been given in previous posts.

Regards

Hi Thierry,

Thank you for your patience. We are still reviewing this issue, please expect additional comments within a day or so.

Regards,
Carol

Hi Thierry,

Thank you again for your patience. We have reviewed your schematic, BOM components, and component layout, and nothing stands out that could be a potential source for device failure.

Regarding your layout, could you confirm that (1) your PCB has a ground plane under the device that the input caps, output caps, and PGND pin are connected to and (2) there are vias under the device to said ground plane to help dissipate heat? Do you have any additional cooling components in your setup? 

Have you been able to reproduce this issue? If not, could you try replicating it at your application's max Vin, load, and temperature?

Regards,
Carol 

Hi Thierry,

We have reviewed your layout further by comparing your topside image with the layout photo you provided in your previous post and a few things stood out.

1) The input capacitor (C17) is quite far from the IC (U5). While your SW node image on a working unit looks reasonable, it is possible that this long path can cause high SW node ringing that violates the abs max input voltage of 76V and cause the device to fail. This can become more probable as time goes on and your device undergoes more stress and/or your input capacitor derates. We would suggest placing the input capacitor as close to the IC as possible. 

Additionally, are transients regularly applied to your circuit? This can also cause larger spikes on SW. 

2) It appears that you have pins 1-4 and pins 13-16 shorted together when they shouldn't be. Could you confirm if we interpreted the orientation of your IC correctly and if this is the case? Could you also provide us with either your board files or a more detailed image of your PCB layout?

Regards,
Carol

Thank you very much Carol.

Even though we have large number of boards having this issue, redesigning the pcb is not an option as it might work or not. It took several years to notice that we had a real issue.

We are going to add another board to substitue the faulty DC/DC converter.

We are going to use a LMR38020 instead of LM5575.

Could you review if possible our schematics and layout, see below?

Schematics:

Top Layer

Best regards

Thierry

Hi Thierry,

Can you tell us:

1. The RT resistor you are using
2. The inductor 
3. The output capacitor

Those are hard to make out in the schematic

Thank you.

Hello,

1. The RT resistor you are using:

R-64K9-0603-1%-75V-100ppm-100mW

1. The inductor:

WURTH  7447709330 (33 µH, 4.2 A, shielded, 5.5 A)

1. The output capacitor:

C-22UF-25V-1210-CERA-10%-X5R

Regards

Thierry

Hello,

I suppose that your post is regarding our issue with LM5575.

As I wrote earlier we stopped improving the current design and move to another designwith another component that you are currently reviewing. We can still send a faulty CPU if it can help but we would prefer that you review our new design as it is what we are going to produce.

Regards

Thierry

We already did the board we do not need an evm.

We just need a schematic and board review as requested in a previous post.

You requested some more info and I gave them to you.

Regards

Thierry