LM3671: LM3671MF-3.3NOPB fails or not doing a good job.

Hi Cyril,

it's a pleasure to help you.

The fact that sometimes the devices goes to a short makes me think that somehow the voltage on some pin is exceeding the AbsMax values and cooking the device. Even though you think you are applying 5V at the input, for example, the layout parasitics together with the switching operation can cause tremendous spikes. Your layout indeed seems to lack of a good ground connection to the ground plane. Actually I see just a strip on the bottom layer, the blue one, which introduces parasitic inductances.

I think this is the direction we need to go, but let's get more information:
-Could you share a more complete schematic and layout from the actual power source to the actual load connected to the 3.3V?
-Could you share the BOM ?
-Could you share also the scope plot of the Vin pin vs GND pin, and SW pin vs GND pin? But first, please watch this short video training for understanding a troubleshooting probing issues:

training.ti.com/power-tips-managing-your-dcdc-converters-switch-node-sensitivity

Hi,

Regarding the caps and the self, this is what is recommended in the datasheet.

I must admit my layout is simple, and I did not think this could be an issue in 2018.

Do you have a layout sample ? Si this possible to achieve something good with a two layer ?

Regarding the 5V, I have used either a DC-C system that converts from 24 to 5V, or even a very good lab power supply, with no success.

Rgds

PS: I've googled this part, and I've seen a lot users having issues with.

Hello Cyril,

Indeed I noticed you followed our suggestion. Anyway the device and the schematic one designs, always interacts with the system you connect this one to. In the case of a DCDC you have a power source and a load to connect. Since the nature of and SMPS is to switch at high freq (2 MHz in your case, which is just the fundamental harmonic of the whole square waveform spectrum), one must take into account the interactions and the non idealities of the source, load and connections.

Maxwell's equations, despite their validity is limited to atomic distances, well represent these electrodynamical interactions (which remain unchanged since 1862 to 2018). For example, from these, the Ampere and Faraday's laws can be derived, which combined together can define the inductance of any conductor when electric charges flow through it.

Even with the best lab power source, a long cable will result in a current loop, hence in parasitic inductance that, interacting with the impedances at the beginning and the end of the wire (output impedance of the lab supply and input impedance of the SMPS) reacts with the input current of the SMPS according to Kirchoff laws (non linearities can be neglected at this point), hence causing ringings and spikes. 

The cable connection is just an example, but since any object on a PCB has high frequency side-effects, you can understand that one has to take particular care of the layout. A general rule is to reduce the pulsing current loops and the pulsing voltages surfaces (which mimic the duality between electrical and magnetic field's effects).

So, resonances and spikes resulting from these interactions can cause both EMI and electrical overstress problem (Voltages and currents can exceed the Abs Max values). I think the latter is the one causing your issue, that's why I asked to shoe us some scopes taken with the right probing technique.

About a layout sample: at page 22 of the datasheet there is a layout suggestion, which is a 2 layers one. For more details you can consult the evaluation board documentation: www.ti.com/.../snva122f.pdf

But, as I said in the first part of this post, the system and its connections are important. If you are using cables especially, you need to reduce their parasitic inductance, and this can be done by twisting them to reduce the area depicted by the current loop (which the inductance is proportional to).

Regarding your declaration about many users having issues with our product. could you please share the sources of this information with us?

Hi Cyril,

I respect your position, and I am sorry you could not make this device working.
I just want to make a couple of points:
-In my first post I asked you to share the BOM. Indeed I have noticed the inductor you are using is a 2.2mF, not 2.2uF as suggested by the datasheet. This makes a huge difference, as it could bring to instability.
-Did you redesign the PCB to fit the new device?
-Could you kindly indicate us the web sources you are talking about?

Hello,

>I respect your position, and I am sorry you could not make this device working.
>I just want to make a couple of points:
>-In my first post I asked you to share the BOM. Indeed I have noticed the inductor you are using is a 2.2mF, not 2.2uF as suggested by the datasheet. This >makes a huge difference, as it could bring to instability.

This is a typo, I have actually purchased a 2.2uH 3.8A self, I have checked 10times the three parts around the LM3671

>-Did you redesign the PCB to fit the new device?

No, if it fails, this will be a disaster for us, I cannot afford doing it X time,  I need to have a working solution and I found one that is OK by me. It is more expensive but it works like a charm.

You should put in red characters that the PCB has to be drawn in a strict given manner, and provide a PCB file sample where it works with.

Also the enable pin seems to be an issue :

The device uses an internal reference voltage of 0.5 V. TI recommends keeping the device in shutdown until the input voltage is 2.7 V or higher.

This has caused a lot of issues; and connect the EN pin to the 5V seems to be very risky.

Just google "LM3671 issues", or fails , you've read statements that cause me to go away from it.

>-Could you kindly indicate us the web sources you are talking about?