TPS82150: Buck converter functions correctly briefly then shuts off

Kyle,

Are C100 and C108 ceramic or electrolytic?

They are ceramic. Originally, I thought they'd be electrolytic but we switched to using ceramic.

By the way, I just hooked up V25D to 2.5V and saw that the part is pulling down the PG pin to logic zero. I also measured the voltage on the SS/TR pin and it is basically at zero (.043 mV). I measured the resistance between the SS/TR pin and ground while the power supplies were off and saw 62 kOhm. 62k seems a bit low, but it should still be enough for the current source to drive it above 1.2V.

-Kyle

Kyle,

Try 3.3 nF for the C98 SS capacitor.  LMK if this fixes your issue.

I found a 10nF capacitor lying around and installed it on the SS/TR pin. The chip doesn't come up at all now.

Kyle,

How many out of how many total are showing this behavior?

3 out of 3 have this problem. Units #1 and #3 seemed to work fine until the load circuit was switched on. Unit #2 seemed to work OK the first time I powered it up, but not this morning. Unit #2 has never had the load circuit switched on.

I have a fourth unit, but I'm afraid to power it on until I understand this issue better.

Kyle,

I'll need to see start up waveforms to assist:

Vin with SS, VFB, and SW.  Start with 1 msec/div then zoom in.

Here are waveforms from two different cards. Signal names are labeled. In all tests, Enable was tied to VIN. The jumper to the load circuit is not connected in these tests.

Kyle,

SS/TR is staying low.  Therefore the output will never come up.  Just as a sanity check con you look at EN as well?

As I mentioned, EN is tied to VIN during that test.

-Kyle

OK I decided to power up VIN to steady state before switching on the EN signal. I did this by leaving EN unpowered while I switch on the +12V, then I switched on the second supply for EN. While doing this, I made an intereresting observation. The EN signal seems to float up and settle at 8V (when VIN is at 12). I took a screenshot of the transient behavior.

First, please double check whether the EN pin is really high with multi-meter after Vin ready.

And also you can change FB resistor to make different output to try, such as 1.8V

Finally, please share your layout for review..

Yuchang

Here is a picture of the layout. It is an 8-layer board in which there are 4 ground plane layers. 

The FB pin is at 73mV. I'm not able to probe the enable pin itself because it's underneath the package. 

I will try changing the FB resistor.

Hi Kyle,

    The fact that EN is floating up without a voltage connected to it is suspicious. As per the datasheet, the EN pin has an internal pull down and should be low when not connected.

Also, the reference designators on the layout are different from the schematic. Can you please check?

regards,

Gerold

Hi Kyle,

    Did you get a chance to look into this?

Regards,

Gerold

Sorry, they re-labeled all of the reference designators after the final layout approval was given. I was looking at the version prior to that. Anyway, here it is with the correct reference designators.

Thanks Kyle for the updated layout. Were you able to check what is causing the EN pin to float?

I don't know what is causing the En pin to float. My thought is that if there was something on my board that is causing it to float up, then the device should turn on anyway?

Hi Kyle,

In addition to Gerold's question, I had a couple more. Is the V12 connected to any other circuits? Is the C100 10uF the only capacitor at this node, or there are others? When you apply your load and start seeing the output fall, is the V12 node (as measured at the input cap C100, not just reading off of power supply) still stable 12V?

Regarding your layout, the ground and other power pin traces would benefit from being wider and ground tied together on top layer, as shown in the recommended layout on p. 17 of the datasheet, to minimize the parasitic inductances and resistances in this important paths.

Kyle, any feedback on my questions?