LM3671 Enable Pin tied to Vin

Sorry, the part is an LM3671MF-3.3/NOPB. The fixed output 3.3V version, sot23-5

Mark, I had replied with your another post, which I believe it's same issue you seen.

I've searched the community and couldn't really find an answer to your original question. I too have designed a board with the enable pin directly shorted to Vin. The board I designed is on 1 square inch and there isn't any room to add an additional circuit. I've started some testing on the first prototypes and I believe I've run into some issues with the LM3671 (I'm using the LM3671MF-ADJ) where it seems that it is shorted. Unfortunately I am out of the office and only have some limited tools to test. The initial power up and check of signals looked good so I thought I was OK to just concentrate on firmware while I was gone.

I'm guessing a short because if I try to bypass the LM3671 by inserting 3.3V on it's output rail the inductor gets hot quickly but the current draw of the board is only 250mA so it can't be a direct short.

Fortunately I had additional space to add a startup control circuit. However I also added a schottky diode between the switch node and ground. Although this is a synchronous converter the external schottky can help overshoot on the switch node that I thought may also be damaging the part. I added a PMEG3015EH which seemed to reduce the overshoot.

My startup used a Microchip MCP112T-300E/TT with 100k pull-down. These devices tend to have an unknown output up to 1V or so but that still keeps the EN pin within spec during startup.

My layout followed the design guidelines and the components tightly placed with wide traces so little I could do to reduce the ringing without the additional diode externally.

I haven't seen any further failures of this since updating the board, but as I added the startup circuit and the diode at the same time it is difficult to say which could be improving things.

Mark

Hi Mark,

Thanks for replying. I never saw an actual response from TI on this (maybe I missed it). Is the overshoot you saw on the switch node only happening at the startup and is this why TI recommends keeping the part disabled until Vin is >2.5V. That one sentence in the datasheet is kind of a poor way to present this if there are known issues.

I do have a NSR0530P2T5G on the board already in a different circuit. It's in a SOD-923 package so it's small enough that I might be able to add one to the switch node. I guess I can hope that the failures were from this overshoot you saw and that is the condition that TI knows can happen when Vin <2.5V and the part enabled. I don't even have room to add a RC filter on the enable pin but now I wish I had. Maybe fortunately in my case the end product is a wearable product so in theory once the battery is connected at manufacturing there will/should be limited times when the voltage goes below 2.5V. I'm working on the first prototypes while I'm on Holiday so I've been doing lots of power ups and maybe that's what caused my problem. I don't have a scope with me so I can't check any signals until I get back in the office.

Thanks again,

George

I had another post on this:

e2e.ti.com/.../443265

If the link doesn't work search for LM3671 Output Short Circuit. I put my scope waveforms and final schematic on that post which may help.

I asked TI for advice on the allowable overshoot as I think the datasheet says the output cannot go above Vin plus a small voltage. Typically the switch node of SMPS has ringing even with a tight layout and mine was exceeding this. I read up on synchronous converters and how to reduce switch node ringing and several sources suggested an external schottky.

Regards, Mark

I saw that, thanks. That schematic doesn't show the schottky, just the delay circuit on the enable pin. I've only got a volt meter with me but I measure ~9 ohms from Vin to GND on 1 board and ~40 ohms on the other board. Seems like mine are damaged but not all the way. I'll see if I can find those notes you were reading about adding the schottky to reduce ringing. I'm using the adjustable version which has caps in the feedback network which might change the behavior, improving or making worse I'm not sure.

The schottky was added between pin 5 and 0V (where a schottky would be in a non-synchronous buck converter), with cathode towards pin 5, anode to 0V.

Take care in schottky selection. I tried a few types with varying success. The ringing is caused generally by parasitics to difficult to predict and measure.

Mark

Thanks Mark. Looks like I'll have just enough room to add one. When I get back I'll check the signals on the scope and compare before and after. I should be able to easily add one on these boards for testing and just a little modification to the design if it looks like this is my problem.

Hi Mark,

Instead of continuing to hijack your thread I started a new one at e2e.ti.com/.../532821. My failure mode seems to be Vin shorted to GND and I don't see large over/undershoot but then my current draw is also very small.

Thanks for your advice so far.

George