This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPS63070: Power save mode vs input voltage

Part Number: TPS63070

Dear Sirs,

We have some problem with power save mode of TPS63070 in our device. Our device has been in production for several months, and a few days ago we notice a problem, because our customer wants to change a standard power supply (so far it was 12V)  to 15V. We assumed that the input voltage range is from 9V to 16V (16V is the maximum input voltage of TPS63070), output voltage of TPS63070 is set to 3.3V. Current consumption isn't constant. When radio module transmits data it is about 130mA, otherwise a few mA, so the average current consumption is about 15-20mA (depends on input voltage). When exetrnal supply fails, the device is supply from internal backup battery (LiFePO4 3.2V). The TPS63070 has enough wide input voltage range. It works fine only to 13.5-14V of power supply. Above 13.5V-14V power save mode doesn't work. Once TPS63070 switches to PWM mode, it remains in PWM mode forever. So the averange current consumption increased from 15-20mA (at 9-13.5V of input voltage) to more than 50mA (at input voltage >13.5V) and is not accepted.

I appreciate any suggestion why it works so strange.

  • Hi Wojciech:

    Thanks for asking. 

    Firstly, please share the schematic for us, including the input ORing. 

    Then may I know how do you judge if the device enters PFM or not? From the current consume or waveform? And what if remove all load?

  • Hi Minqiu

    Because of a specific purpose of our devices I shouldn't share a complete schematics. Any changes in existed devices are a little bit complicated, because devices are encapsulated. Moreover any relevant changes reqiure re-certificate. There is a simplified schematic of a power circuit below. Battery backup circuit is negligible. F1, D2-D4, D6, D7 are necessary to fulfill safety regulations. D1 and T2 switches the power supply between external 9-16V and battery.

    I made a few tests with a real load (other parts of our device), one resistor (simulating current consumption of real load) as well as without load. I observed a current consumption. Then I cutted off PS/SYNC track and pull to low (for forced PWM) and I made the same tests. The current consumption with forced PWM mode and with PWM/PFM mode (PWM/PFM mode at input voltage higher than about 13.5V) were very similar.

    Observed current consumption (rounded values):
    1) PWM/PFM mode, Vcc < 13.5V,
    - no load, Icc = 1mA
    - after attached and detached a resistive load, Icc = 1mA
    2) PWM/PFM mode, Vcc > 13.5V
    - no load, Icc=1mA (I turned it on many times and sometimes it started with Icc=25-30mA)
    - after attached and detached a resistive load, Icc = 35mA
    3) forced PWM
    - no load, Icc=25-35mA (depends on Vcc)

    Best regards

    Wojciech

  • Hi Wojciech:

    Thanks for your supplement. Please allow me do a quick reply first.

    What's the resistance you use to simulate? I'll try to reproduce the issue in our lab on EVM if the equipment is available today.

  • Hi Minqiu,

    Rload = 10 Ohm, I think that it's a good approximation of the maximum current consumption. The average current is a few times smaller.

    When you switch the load very slowly, you may not observe the problem. Sometimes Icc falls to 1mA, and sometimes to 35mA. I'm not sure why. In our case the load changes from 3 to 100 times per second. It depends on the frequency of data package sending by a radio module.

    I'll appreciate if you try simulate this in two ways:
    1) turn the power on with no load and then attach and detach the load
    2) turn the power on with load and then detach the load

    In both cases I've got similar results

    Best regards

    Wojciech.

  • Hello Wojciech,

    On how many boards do you observe this behavior?

  • Hi Brigitte


    Our customer checked 3 devices and reported a problem to me. Probably most of the 170 provided devices are already installed and working (with lower Vcc=12V). I checked 4 devices in our office (2 pcs from production part and 2 prototypes), with two different marking numbers. In all cases I observed the same behavior.

    Best regards
    Wojciech

  • Hello Wojciech,

    With the same behavior you mean that input voltages above 13.5V cause the device to not enter PFM mode, correct?

    To be honest, I did not observe this behavior up to now and wanted to be absolutely sure and checked 3 EVMs in the lab with 16V on the input and 3.3V on the output and all of them behaved as expected and entered PFM mode at startup.

    Could you please check if there is possibly a connection issue for the PS/SYNC pin?

    Please share the part number of your inductor. Please check if the problem disappears when you are using a different inductor with higher current rating.

  • Hello Brigitte

    Yes, input voltage above 13.5V causes the device do not enter PFM mode. Sometimes the device enter PFM mode at startup, but after change of a load current it remains in PWM mode. I also observed an output waveform (not only current consumption) and it confirms that the device do not enter PFM mode.

    I wanted to eliminate the influence of the rest of my circuit and I cutted off the whole circuit behind TPS63070 (I left only a resistor to simulate any load), as well as the whole input circuit  and applied the input voltage behind a ferrite bead L2. I also tried to change the value of input and output capacitors, but the problem remains.

    Connection of the PS/SYNC pin seems to be fine. The voltage on this pin is equal to the input voltage.

    The part number of my inductor is LQM2HPN1R5MG0L. I will try another inductor soon.

    Best regards
    Wojciech

  • Hello Wojciech,

    If you have the ability to measure the current through the inductor with a current probe or a shunt resistor, this would be very helpful to dig a little more into the issue you face.

  • Hello Brigitte

    There are waveforms of inductor current below, measured with a shunt resistor of 0R1.

    Fig. 1. Vin = 12V, attached load R = 10R

    Fig. 2. Vin = 12V, detached load

    Fig. 3. Vin = 15V, attached load R = 10R

    Fig. 4. Vin = 15V, detached load

    All above waveforms are for an inductor of L = 1.5uH, Isat = 1.5A (LQM2HPN1R5MG0L).

    I repeated this test with a different inductor (L = 2.2uH, Isat = 2.5A, it was the only inductor suitable for this application in our stock). The inductor ripple current is smaller, but the problem remains. First I thought that the problem is a saturation of the inductor. But as you can see, the maximum current is smaller than Isat. I may buy a few different inductors with a higher current but I'm not sure that it is a right way.

    Best regards
    Wojciech

  • Hi Wojciech,

    The boundary between the PFM and fPWM mode is typically 650 mA of the inductor current, which is in this case (buck mode) the output current. At R=10 Ω, the output current is around 330 mA and the device should still operate in PFM mode. It puzzles me why it ever switches to fPWM mode at all. In any case, even with the output current of 1 A, i didn't observe the device staying in fPWM mode after the load was removed, independent of the input voltage or startup conditions.

    Could you please do the following:

    • Connecting the input supply directly to VIN pin (bypassing the diodes, switch and filter).
    • Driving PS/SYNC and EN externally with 3.3 V, without the series resistor.
    • Making a scope plot of VIN and VOUT when the load (resistor) is turned on and off.
    • Sending the PCB layout. Just a cropped print-screen of the area around the TPS63070 is enough.

    Best regards,
    Milos

  • Hi Milos

    As I wrote earlier, I already made a similar test bypassing whole input circuit and cutting off the whole load (the rest of my PCB). I left only a resistor as a load. But PS/SYNC and EN were connected to Vin through a resistor.
    Now following your sugestion I connected these pins directly to external 3.3V. The results I got were a little bit different. After removing the load the device entered to PFM mode. Why 3.3V on PS/SYNC and EN is better? But one issue remains. I didn't mention about it earlier, because I thought, that it is a result of higher Vin voltage, like my other problems. There is some problem during startup with Vin>13.5V. Sometimes (not always) the device started with Vout about 2.2V-2.5V, not 3.3V as it should be, and remained at this level. Of course in this case the current consumption was higher than expected (about 25mA without load).
    There are the results I observed following your sugestions.
    Yellow - Vin, blue - Vout

    Fig.1. Vin = 12V, no load

    Fig.2. Vin = 15V, no load

    Fig.3. Vin = 12V, R = 10Ω

    Fig.4. Vin = 15V, R = 10Ω

    Fig.5. Vin = 12V, left half - no load, right half - R = 10Ω

    Fig.6. Vin = 15V, left half - no load, right half - R = 10Ω

    There is my PCB layout below.

  • Hi Wojciech,

    I am sorry for the delay, and thank you for the additional info.

    I think in this case the problem is the sub-optimal PCB layout, in the first place the GND layer routing. These are my remarks:

    • You should add more GND vias close to the GND ends of input (C8, C9) and output (C10-12) capacitors, at least 2 GND vias per capacitor. Especially for C8 and C9, since in your case issue appears in buck mode, and in buck mode the layout for input capacitors is crucial.
    • You should add more GND vias as close as possible to the PGN pin.
    • The GND side of R13 should be routed to GND pin directly instead to the PGND pin.
    • GND pin should be routed to GND center point (in this case bottom GND plane) instead directly to PGND pin.
    • GND plane below the device should be uninterrupted. Currently you have a GND line from VOUT to R11 resistor "breaking" the ground path between the GND via close to the GND pin, and GND vias connected to PGND and input output capacitors. Since you don't use PG output, you can omit R11 and leave PG pin floating, or connecting it to ground, and make the GND plane below the device completely filled.

    This might look like a lot of remarks, but the PCB layout is extremely important for switching converters, and we insist on following the guidelines. Otherwise, I think there is no easy workaround for your current design, except trying the TPS63070EVM wired externally to test it. With weak GND routing, you likely have GND issues caused by switching waveforms. Please check the PCB layout guidelines in the datasheet, and how we did it on our TPS63070EVM.

    Best regards,
    Milos

  • Hi Milos

    I am sorry that I didn't reply earlier, but I am very busy with other projects. As soon as I find some free time and get permission, I will try to improve the PCB by following your remarks. Due to space limitations and components on the opposite side of the PCB, some changes may be difficult to make (e.g. vias for C8 and C9). I know that my PCB is not optimal, but sometimes it is difficult to do something as recommended by 100%. By the way, I've done a lot of projects with switching converters so far, but I didn't have such problems as with this one. The engineer have to lifelong learning.

    Best regards
    Wojciech

  • Hi Wojciech,

    Thank you for the reply.

    I will then close this thread, and if you have any update (or further issues) with the TPS63070 you can open a new one.

    Best regards,
    Milos