TPS63001: Power consumption increases over time

Part Number: TPS63001

I am seeing strange behavior with a batch of TPS63001 regulators.  The PCB consists of a TI CC1310, some other low-power sensors, and the TPS63001.  The board will eventually be powered from a Li battery, but right now is being powered from a bench supply.  Expected max current 15mA during radio TX, quiescent current should be about 2mA (largely due to a power LED that is on for debugging).  

Of the first batch of eight boards, 4 of the TPS63001 regulators were foo'ed: on the first power-up, they started drawing 200mA with nothing else powered on.  Using the radio, the current would jump to 215mA, but it would never go much below 200mA.  Using a FLIR, the failed regulators were red hot (over 150C).  The other 4 units functioned as expected.  There were no shorts present on the 4 defective boards, so I replaced the 4 regulators, thinking maybe they were defective.  They functioned exactly as expected.  When we were done testing, we had 8 working boards.  They all took programs, and they all ran through their initial tests.  

I've been running sensor calibrations with two of the units, and noted that the quiescent power started increasing, slowly at first.  Starting two days ago, the baseline power was 5mA with peaks to 20mA during radio TX, then by the afternoon it climbed to 40mA (and 55mA peaks), and after running through the night, yesterday morning it was 100mA (115mA peaks), and by the afternoon was up at 150mA (165mA peaks), and the regulator was getting hot.  I didn't want to leave it unattended after that.

The board is relatively simple, there are three power rails - the supply rail from the battery (presently using a bench supply), the always on 3.3V regulated line, and an auxiliary line that is enabled by the MCU to turn on the sensors and is tied to the 3.3V regulated line.   I have test points on the board that allow me to directly drive the 3.3V regulated line with a bench supply and by-pass the regulator.  The quiescent power draw of the rest of the circuit is exactly as expected - 15mA peaks, 2mA for the LED when everything is asleep. 

The regulator design is based on a WEBBENCH output, but I had to make some part substitutions due to availability.  The filter caps for the regulator are all rated to 50V (input voltage would never go above 4V), input is 10uF, output is a series of 3 at 10uF, 47uF, and 100uF.  The 10 and 47uF are placed nearly adjacent to the regulator, while the 100uF is placed between the regulator and the CC1310. The inductor 2.2uH is rated to 20V and 500mA and is nearly adjacent to the regulator.

There are other temperature sensitive components on the board, so we used lower temperature solder paste that reflowed at 220C, so I don't think we cooked anything.  I'm not really sure where to begin...

4 Replies

  • Hello Tom,

    Please share your schematic, layout and BoM (if you do not want to share it here, please send it to lbs_request@list.ti.com).

    Please check as well the peak currents that are delivered from the power supply during normal operation. Especially with your output capacitors a lot bigger than the input capacitors, the peak currents at startup could be a lot higher than you want to have them.
    Are you using power and sense lines supplying from the power supply to the board?
     

    Best regards,
    Brigitte

    For more information on buck-boost devices have a look at www.ti.com/buckboost

  • In reply to Brigitte:

    Brigitte,

    Thank you for the reply.  If you prefer the actual OrCAD files, let me know and I can send them, for brevity, I'm attaching just the regions of interest.

    The board layout for the regulator is shown below.  Colors indicate:  Brown = Ground, Yellow - L1/L2 inductor connections, purple is Vbat, aqua is 3V3.  The green region under the TPS63001 is the thermal pad that is also connected to ground.

    The accompanying schematic:

    Bill Of Materials August 31,2018 15:28:13 Page1

    Item Description Qty Ref Des
    Footprint Mfg P/N# Manufacturer
    Digikey PN
    ______________________________________________

    3 CAP CER 10UF 10V X5R 0805 2 C3,C4
    smd_cap_0805 LMK212BJ106KG-T Taiyo Yuden
    587-1300-1-ND

    4 CAP CER 100UF 6.3V X5R 1206 1 C5
    smd_cap_1206 GRM31CR60J107ME39K Murata Electronics North America
    490-7217-1-ND

    11 FIXED IND 2.2UH 640MA 429 MOHM 1 L1
    smd_ind_0805 CBC2012T2R2M Taiyo Yuden
    587-1601-1-ND

    20 IC REG BUCK BST 3.3V 1.6A 10SON 1 U2
    TPS63001DRC_v2 TPS63001DRCR Texas Instruments
    296-19643-1-ND

  • In reply to Brigitte:

    Brigitte,

    You mentioned that the input current might be greater than expected.  I captured these using a scope and an ADS8210 current shunt with a 1.1 ohm resistor.  Current is scope voltage / 22.  

    The first trace is at 40us per div, and I read it to be about 3.1V peak (/22 = 140mA). 

    This capture is 400us per division/  The first peak is still present, but then it looks like the oscillator is starting up?  The peaks are all right around the same (3.1V = 140mA)

    Finally, this is after the the system has reached steady state.  The voltage is flat and level at about 2.1V ( / 22 = 95mA).  There is a CC1310 module on this supply, but its in deep sleep when this is taken.

  • In reply to Tom Briggs:

    Hello Tom,

    Thank you for this information. The input current of your power supply is too high when there is no load on the output.

    I expect that the biggest contributor is the inductor. Please replace it with one that has a lot lower series resistance. 429 mOhm is extremely high for a power supply. Please check with an inductor that has a series resistance of less than 0.1 ohm.

    And we recommend to use an RC filter connected to the VINA pin to reduce noise coupled into the VINA pin.

    In addition, your layout is not optimized for a power supply. Please have a look at the layout recommendations we have in the datasheet. I expect that your layout causes a lot of noise in the system.

    As it is almost impossible to make an existing layout better without ordering a new board, maybe it would be a possibility to take the TPS63000 EVM and check if your system can be supplied through the EVM. If this works out, exchange the external components on the EVM with your components and check if the issue comes back.

    Best regards,
    Brigitte

    For more information on buck-boost devices have a look at www.ti.com/buckboost