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.

TPS2372: TPS2372 LED driver advice

Part Number: TPS2372
Other Parts Discussed in Thread: TPS54560, , TPS23861, TPS23861EVM-612, TIDA-01463

We've had a board developed for us that uses TPS2372 as the PD interface. This then supplies power to a tps54560 switcher set to supply 12V at 2A. 3x MR16 LED bulbs are to be connected to the output of the PCB. There is also a switch circuit to allow the LED bulbs to be turned on/off. The MR16 bulbs used are 5W LED bulbs that can work on either 12V AC or DC (constant voltage). I expect that within the bulb, there is circuitry to drive the LED with constant current and also allow dimming. 

Having received the PCB's for test, i have noticed the following -

  1. If i supply 48V through the input RJ45 from a PSU, on pins 1,2 and 3,6, then I can turn 3x bulbs on/off without issue.
  2. If i connect the PCB to our PoE PSE (based on tps23861) then i can only load with 1x LED bulb. Any more than 1x LED bulb results in the light initially turning on then it turns off immediately, so a brief flash.
  3. I have repeated (2) using the TI tps23861evm-612 eval board with the same result.
  4. Repeating (2) & (3) with the TI PoE commander software monitoring the PSE, i dont see any fault flags for Icut which is set for 110-686mA. Are there anyother registers i should pay attention to?
  5. I have replaced the LED bulbs with a DC load set to drawn 2A at 12V. This works and i see the power draw expected when i turn the output on/off.

In (2), (3), (4) and (5) the PCB correctly classifies as class 4 and is confirmed as such with the PoE commander software. Even when the bulbs are switching off in (2) and (3) the classification is never re-done.

Could you suggest where to look for the source of the turn off in (2) and (3). To me, the PSE is supplying power correctly as the PCB classifies and maintains this, plus there is no error flag in PoE commander.

I believe that the tps2372 might be causing the blip, but i need to investigate further tomorrow, but i cant explain why all is well in scenario (1).

  • Hi Howard,

    In cases 2, 3, & 5, did you power on while the load is attached?

    In case 5, did you power on with PSE or 48V power supply?

    Regards,
    Tom Amlee
    PoE PD Apps Engineer
  • Hi Tom,

    The PCB consists of PD interface, 12V switcher and a load switch. The LEDs are connected to the load switch. Therefore the LEDs are always connected to the PCB but only electrically connected when the switch is enabled. The 12V switcher is under the control of the tps2372 PG pin.

    Case 5 was using the PSE to power the PCB.

    Further investigation today has shown dips in the 48V supply to the switcher and onwards when the LEDs are turned on. I have increased the bulk capacitance from 68uF to 100uF and this appears to sort matters out, but i was under the impression i wouldnt need such a high bulk capacitance on a class 4 device. During the 48V supply dips the PD didnt disconnect from the PSE nor were any errors flagged in Poe Commander.

    Howard

  • Hi Howard,

    Is the TPS2372's inrush delay enabled? If so, is the PG of the TPS2372 connected to the ON/OFF switch circuit of your lights to keep the load off during inrush phase of the PSE?

    Thanks!
  • Yes, i am using the inrush delay of the TPS2372. The PG pin of the TPS2372 is connected to the enable pin of the tps54560.

    The switch controlling the LEDs is off at start up and is manually turned on. Its highly unlikely the LEDs will be turned on just as the tps54560 is enabled and this certainly isnt the case in my testing. There would be seconds from the 12V switcher coming up before i attempt to turn the LEDs on.

    What i saw today was that the 12V turned off momentarily when i switched the LEDs on, this resets the switch to off, the 12V then returns. I could only overcome this by increasing the bulk capacitance to 100uF. The PSE reports no fault at this stage and it doesn't retry classification.

    Once the TPS2372 is out of inrush phase and the tps54560 is enabled is there any current draw event that would cause the TPS2372 to turn off?
  • Hi Howard,

    We also have our bulk cap as 100uF for our PoE Lighting design, TIDA-01463. Using a larger bulk cap is not unusual for a buck topology.

    There may be a spike of current when you connect the LEDs which can hit the operational current limit or foldback threshold of the TPS2372.

    To verify this, you can take a waveform of RTN-VSS when you connect the LEDs.


    Thanks,
    Tom Amlee
    PoE PD Apps Engineer

  • Hi Tom,
    Not been in the office the last couple of days to make any further measurements.

    Should it matter whether i use the TPS2372-3 or -4 as im only using a Type 2 PSE.

    If i classify as Type 2 Class 4 would the extra current limit of the -4 be available to me? What is the current limit available for type 2 class 4?

    Howard
  • Hi Howard,

    Both parts perform the same, however it would be better value to use TPS2372-3 for lower cost.

    If you classify as class 4, the TPS2372-4 operational current limit remains the same (1.9A). The current will be limited by the PSE first before the PD current limit is reached.

    Regards,
    Tom Amlee
    PoE PD Apps Engineer
  • Hello Howard,

    If we have found resolution to the issue, please click resolved. If work is still in progress, then please reply.

    If I do not see activity in the next week, I will close the thread, however, your reply will automatically reopen it.

    Thanks,
    Tom Amlee
    PoE PD Apps Engineer

  • Hi Tom,
    Apologies for not responding.

    The designer has increased the Cbulk to 330uF and this appears to overcome the issues with voltage drops when certain LED bulbs were turned on.

    As such please close this thread.

    Howard
  • Howard,

    No problem & congrats on finding the resolution. Thank you for clicking the green resolve button.

    Regards,
    Tom Amlee