LM3668SD-2833EV: Buck-Boost regulator malfunction during use

Would you please offer the details about the failure, as well as the operating conditions, your schematic, and any scope pictures that can help us understand the problem?

Best Regards,
Youhao Xi, System and Applications Engineering, APP-BMC-BCS, TI

Hi,

Here are schematics.

"Vout" is regulated DC voltage coming from pin 1/20 on MCP73871 (3.0-5.2VDC). Voltage to MCP73871 comes from USB or 3.7V Li-Ion battery. The combined load on LM3668SD-2833 is <200mA.

The customer has been using our device for several months. Then suddenly they are not able to turn the device on because 3.3V is not generated to the MCU. When the power button is pressed, a current of approx. 1.6 A is drawn from the battery (normally <250 mA), but no 3.3 V is generated in the circuit. Instead 0.60V comes out of LM3668SD-2833 when holding the power button. At the same time the IC heats up to about 140 degrees C.

It sounds like the LM3668 is damaged somehow. From your description we cannot tell what inside the IC is damaged, and how. Is this board in production or just a prototype? Can you monitor SW1 and SW2 voltage waveform and see if the device is still switching or not?

Hi,

This board is in production and is being used by our customers. When LM3668 fails, the whole product fails and the product is being returned to us.

I can't monitor SW1 and SW2 now since I had to remove the IC to verify that it was LM3668 who had the high current draw.

I just notice from the schematic that EP (or DAP as referred to in the datasheet) is not connected to the ground plane. EP is for heatsink and should be connected to ground plane to dissipate the heat. Otherwise the accumulated heat may damage the IC. This may explain why the failure occurred after some time of operation.

Hi Lars,

In order to better support you, would you send some info of your company and the product that uses the LM3668 to my email: youhao.xi@ti.com? If necessary we can send an FAE in your local area to assist you.

Best Regards,
Youhao

Yes, that is correct. My PCB is 4 layers: 1 - top layer, 2 - ground plane, 3 - power plane and 4 - bottom layer. Since the datasheet says "...connect the DAP to SGND on PCB layout to enhance thermal performance. It should not be used as a primary ground connection" I have not connected it to my ground plane.

Should I connect EP/DAP to my ground plane?

Hi Lars
Please refer Figure 4 in the LM3668 Evaluation Board User Guide at www.ti.com/.../snva236f.pdf
EP/DAP is connected to LARGE SGND plane.
Regards,
EL

The evaluation board has 4 layers: TOP, GND, SGND and BOTTOM. LM3668 pin 3 goes to GND layer while LM3668 pin 9 and EP/DAP goes to SGND layer.

My PCB also has 4 layers: TOP, GND, PWR(3.3V) and BOTTOM. LM3668 pin 3 and 9 goes to GND layer. My question is: should LM3668 EP/DAP also be connected to my GND layer?

Yes you need to connect it to GND. The words in datasheet about this is to avoid the noisy switching current from flowing through the DAP to affect the IC performance, but electrically it should be grounded.