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.

tps54160: Power output not correct or not stable. Always shuts down when hooked up. Used a design off of the web bench.

Part Number: TPS54160

Ok so I made a schematic I will provide of this TPS54160.  The output voltage is correct at first but when I hook it up to what I want to power it shuts down.

Input is about 48v (coming from PoE+ which is rectified and clean DC)

Output should be 12v (normally is until it is used)

This design was copied from the web bench tool and is meant to power a camera.  The camera shouldn't be attempting to draw to much power.  Any help or suggestions would be greatly appreciated.  

  • On the schematic, both TPS_PH is connected to both C11 and R7, but it is not labeled elsewhere on the schematic. I'd assume it is the PH pin of the TPS54160. If that's the case, it is correct to connect it to C11, which it the boot capacitor. But it shouldn't be connected to R7, R7 should connect to the 12V output voltage, which is the other side of the inductor.

    Please recheck the connection, make sure C11 is connected to the PH pin of the device and R7 is connected to the output side.

    Regards,

    Yang
  • Yes it is not clear considering I didn't mark it. However TPS_PH is connected to the PH pin so C11 and R7 is connected to the output side. Also I have noticed that R7 should connect to the other side of the inductor and have worked around this on the PCB. however it has not fixed the issue.
  • Please capture some waveforms to help debugging: VIN, PH, and 12V output on one screen before and after loading. Please slowly increase the output load current, and monitor at which load the Vout will drop. When the Vout drops, does the voltage at the VIN pin stays stable?
  • I will get the scope captures but first I want to note that when I move R7 to the other side of the inductor the 12V as apposed to the PH side the voltage becomes more like 5v. Does this give you any clue what is going on? and not sure the scope captures will help until that problem is resolved.
  • It shouldn't be. Something is not right. Can you also share your layout?
  • Might be tougher to get right this second but the guidelines were fallowed. It should hooked up as the schematic shows. (except what Iv told you different) is there any reason you can think of that would happen.
  • Here is the layout.  of course R7 got disconnected from its bottom bad and a jumper ran from it to 12v.

  • The layout can be improved, especially the loop formed by VIN pin of the IC to Vin bypass cap C13, to power diode D2 and return to the ground pad of C13. These components should be as close to each other as possible to minimize the area of this loop, to reduce swithcing noises on the board. The main concern with the layout is that all the sensitive components are grounded at the diode ground to Cin ground return path, which contains all the switching noises. The most sensitive component is R6, which is part of feedback divider. It should be grounded to pin 9, and away from the diode ground return path. Noise coupled to R11 and C9 can cause problem as well. The Vout connection to R7 jumped by an external wire can couple noise to the feedback path as well.

    I did some calculation on the EN divider values. Using the equation (2) and (3) on page 16, the resulting Vstart = 35.4V and Vstop =35.2V. If you are using Vin around 35V, this could be reason why the Vout turns off with loading. If that's the case, then change R8 to 1MOhm and R9 to 38kOhm will result in Vstart = 33.2V and Vstop = 30.3V. But if Vin is at 48V, this shouldn't be an issue.
  • Yes the Vin is 48V DC.  That being said you think the layout is bad enough it just wont work?

  • It is really hard to know just by looking at the board. Would you capture some waveforms so that we know what actually happens in the circuit?
    The nodes to pay attention to includes
    VIN pin of the IC: is there a voltage drop, or ringing, or spike when the output is loaded?
    PH node: the PH node should be a pulsing waveform with high level at Vin, low level near ground during start up. After startup, if there's no load, it should have a positive pulse at a pretty low frequency. If part works properly with load, this node should be a continuous squarewave pulses, with duty cycle depending on Vout/Vin
    Vout: Vout should be a near DC waveform with small ripples. I don't understand why Vout will be 5V when you tie R7 to 12V rail. I'd trace the voltage from 12V to top of R7, to the other end of R7, to debug. The other end of R7 is Vsense pin, which should be 0.8V with or without load. Please probe Vsense node carefully, since it is pretty noise sensitive.
    EN pin: since Vin is above the threshold, this pin voltage should be above 1.2V threshold as well. Please monitor it on scope to make sure.

    How is your loaded, with electronic load or resistive load, or an active device? Can you gradually increase the load from zero ampere to see what changes with loading?
  • Sorry it took me so long to get back on this priorities changed at work some.  Please note I did make the change were (in the schematic) TPS_PH aka the R7 is on the 12V line and not PH.  

    Another thing that might give you a clue as to whats going on is C10.  on the web bench and my schematic it called for 1.2pF  and often it would not give 12v on the output. below is a capture of the PH pin with the quoted value at C10.  

    We have found when C10 is changed to around 15 to 20 pF the output voltage is more stable at 12.6v and works under a small load.

    Keep in mind i cant just change the load to much it is an active component load.  when we hook the board up to the camera it has an encoder, camera, and other things that kick on.  When the board is not hooked up (as in all these captures) it still has 2 LED's, a msp micro, RS232 converter IC, and a LDO.

    Now with a 15pF across C10 the PH looks more like this.  

    Its almost as though it gets to much noise then levels out to the wave form you mentioned for a bit then messes up again.  

    The EN and VSense pins where correct to be on I have captures of that if you want. 

    The VIN was a clean looking 48v dc.  Never sagged even when the IC shuts off or drops.

    Maybe these captures will help and maybe knowing that it seems to work better with C10 being such a higher capacitance might give you the clues you need.  any help is always greatly appreciated.   

  • Please see my latest post on this thread. The issue is not resolved.
  • Was figure one above taken with 12V output voltage or Vout had dropped to a lower value already? How much load was there?

    The 2nd scope shot actually showed the operation was not stable. It shouldn't change from discontinuous condution mode to continuous condution mode without a load change, and frequency shouldn't change as well.

    I would recommend using the original C10 value, try to add more ceramic capacitors to the input and output ports. Note that ceramic capacitor derates a lot when the voltage on it is close to the rated voltage. Try to use ceramic cap rated >=25V for the output and >80V for the input. For experiment, you can just double up the output capacitor on top of the existing one to see it changes the behavior or not.
  • These scope captures were with output roughly 12v. The load on it was very small. I cant tell you exactly how much but there was the minimal components that are just on the board and of course the camera and encoder and such were off the board.

    I have tried adding various different capacitors to the input and output with rated voltages at least double. It doesn't seem to have any effect. I tried changing C10 and sometimes it gives a better response but never a solid square wave on the PH pin. I have also tried changing R11 to see if the switching frequency needed to be raised or lowered.
  • Robert,

    I would recommend debug the board with resistive load first. I would be great if you can find those adjustable ones. Start from lower current to higher. I think it simplifies the issue to the converter itself.
    Would it be faster to spin the board with better layout than debugging this one? You can copy the layout on the datasheet or EVM next time.

    Best regards,

    Yang