LM5137F-Q1: Not starting up

Stephen,

The issue may be related to the DCR current sense circuit, as it may be causing the pin check diagnostic to fail at startup. Try reducing the Rsense to 1.5k (and increase Csense accordingly to maintain the same time constant).

Regards,

Tim

Thank you for this feedback. I also feel the inductor DCR sense circuit is at fault, and I've been trying various values. Unfortunately there are no examples of this particular circuit to learn from, and the documentation is fairly weak for this specific use case. Hopefully someone from the TI apps team can jump in...

Working within the available component values to test with I tried:

• Rcs = 200Ω, Ccs = 4.7uF
• Rcs = 909Ω, Ccs - 1uF
• Each of these are as close as I could get to the Lo/Rdcr of 4.7uH/5.11mΩ of the selected inductor

I also tried removed the voltage divider R205 and R206, which are intended to set the overcurrent threshold where I want it. I also tried replacing R213 and R237 with 0Ω, but no difference.

Not sure why TI this thread is set as "TI thinks it is resolved" when it is not.

Can you install shunt current sensing to see if that eliminates the issue?

I tried a 4mΩ shunt and I just see the exact same behavior.

And you had the 249Ω in series with ISNS1/2+? Must be something unrelated to current sensing then. Send on a completed quickstart calculator

Yes, 249Ω in series on Isns1/2. I've attached the calculator sheet for my design.

LM5137F-Q1_quickstart_calculator_CMBrevA.xlsm

The only significant differences in my schematic to the TI examples for 12V are:

• Inductor DCR sensing
• Fixed 12V mode (4.99kΩ to VOUT1) vs adjustable output with Vsns and FB1 voltage dividers

I've reconfigured my board for current shunt sensing and there is no change in behavior.

I have not tried the adjustable output mode as it is a bit more rework, but that is what the TI EVM uses, even for the 12V single output version of the EVM. Is there a possible problem with the fixed output mode?

Monitoring the nFAULT summed outputs, I am seeing faults generated at startup. Now just trying to determine the source of the fault...

• PG1/2 pins are pulled high (added via rework so they aren't open)
• CNFG, EN, INSNS1/2+, NFAULT1/2 are all at expected levels or not open
• VCC is 5V
• VDDA is 5V

Not sure what else to check.

Correct. I have a 48.7kΩ pullup to VDDA on FB1. VSNS1/2 is connected to VOUT1/2 (respectively) via a 4.99kΩ series resistor.

I have not. My inductor doesn't fit so some more involved rework is involved...

This is a 2-phase design. If so, you only need FB1 and VSNS1. FB2 and VSNS tied to GND (this is what you have in the schematic as well).

Send on an updated quickstart file with the shunt resistor configuration and correct/derated Cout value (44uF doesn't look right). Also, looking at the schematic, why is there a 4.99kΩ from ISNS1/2+ to AGND?

Correction on my part... FB2 and VSNS2 are connected to AGND.

This was intended as a divider to get the desired overcurrent sense threshold. I have depopulated 4.99kΩ from ISNS1/2+ to AGND, and there's no change in behavior.

Hi Stephen and Tim:

I noticed the current schematic above has the UVLO configured like this: 

But the datasheet has it like this:

Would this impact the EN pin diagnostics it does at bootup?

Nick

That divider resistor to reduce the current sense signal would go from ISNS to VOUT, as it's a differential voltage (not to GND)

Are you saying the body diode of the NFET is possibly causing the chip to detect the EN pins to be shorted to VIN? The intention here was to have an active high enable signal from the host controller. I guess didn't anticipate the protection would see it that way. Maybe a diode in series would help?

Now seeing an actual output, though it is not staying on. I've now switched to the shunt based circuit, using the same parts on the 12V version (variant 002) of the EVM. I've also updated the compensation circuit values to be the same as the EVM, which are different than the calculation spreadsheet recommends. You can see the output below.

Got it working... I needed to add the 10pF.

Also, the FET won't work where it is. I'll have to figure out the enable signal.