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.

TPS23770: Can't get the circuit to sustain a load

Part Number: TPS23770
Other Parts Discussed in Thread: TPS23750

Hello,

I have a design using the TPS23770 that was copied almost identically from the SLVU136a circuit.  When plugged into PoE, I have a 5V output but if I apply a resistive load to get me just under 500mA, the input 5V shuts off for a few seconds after the load is removed.  I have also tried to use the "AUX" power with a 12V source through the diode but I can't even get the 5V rail to power up there.

Thanks,

Mike

  • Greetings Mike. Our staff is on Easter holiday today through Sunday. One of our applications engineers will get back to you on Monday.

    Thank you for your interest in PoE from Texas Instruments.
  • Hi Mike,

    5V rail shuts off for few seconds after the load is removed because with no load, the current through the PSE goes below the MPS (maintain power signature) threshold at 10mA. Thus the PSE will remove power. It is the PD's job to have the load to allow the PSE to maintain power. You can add a dummy load to your PD system.

    The EVM was not designed for 12V adapters. It was designed for 48V adapters. At 12V, looking at the EVM schematic, it is below the UVLO of the PD at wouldn't startup.

    Regards,

    Darwin

  • Darwin, 

    Please see response below I've highlighted.

    Darwin Fernandez said:

    Hi Mike,

    5V rail shuts off for few seconds after the load is removed because with no load, the current through the PSE goes below the MPS (maintain power signature) threshold at 10mA. Thus the PSE will remove power. It is the PD's job to have the load to allow the PSE to maintain power. You can add a dummy load to your PD system.  

    I'm not having a problem here.  My issue is is when I apply a resistive load of ~500mA, the 5V output shuts off, not a minimum load problem.  According to the SLVU  schematic, this circuit should be able to do 2A but I can't even do 500mA.

    The EVM was not designed for 12V adapters. It was designed for 48V adapters. At 12V, looking at the EVM schematic, it is below the UVLO of the PD at wouldn't startup.  Understood, I miss-read the EVM schematic.

    Regards,

    Darwin

  • Hi Mike,

    Yes, the EVM can go up to 2A. It sounds like the sense resistor might be too high, can you confirm it is 0.18 ohms like in the EVM? If it is, can you send your schematic that I can review? Thanks!
  • Darwin

    I will double check the sense resistor.  Attached is the schematic.

    Thanks, Mike  

  • Hi Mike,

    Nothing catches my eye on the schematic. The connections and component selection looks okay. It might be something on the board. You could try comparing the impedances on the EVM with your board to narrow down where the issue might be.
  • Thanks Darwin, I will pick up an EVM to compare.  In the meantime, I took a few scope shots and took some screenshots of the layout.  Do you see anything that stands out as glaring to help narrow my focus?

    First image is Top Layer, You can see Layer 2 beneath it with some planes.  Light Green is the Output Ground (system ground for the 5V rail), Dark Green is the Input Ground (after the diode bridge), Dark Blue is the RTN plane/connections (the Isolated net)

    Second image is Layer 2 showing the planes as mentioned above.

    Third image is Layer 3 with Layer 2 in the background.  You can see signal nets crossing under here for other portions of the design unrelated to the PoE.  On my demo, none of that circuitry is stuffed yet so those lines aren't doing anything.

    Fourth image, this is the Bottom Layer of the board.  The entire area under the POE circuit is poured with the System Ground (mentioned previously).  That one Blue Pour that runs vertical is the 5V net from the PoE circuit that would go and feed a 3.3V regulator.

    Additionally, I validated the sense resistor was 180mOhm.  I thought maybe it was possible that I didn't have the best kelvin connections for the RSP and RSN lines (RSP went didn't go directly to the resistor, it went closer to the source of the FET, and the RSN line was dropped to the inner pour as done in the EVM).  I lifted the pins on the part so they did not make connection with the circuit board, and then created kelvin connections with two 30AWG wires directly to the Sense Resistor.  This didn't change my results.

    That being said, I captured the voltage measured across the sense resistor before applying my 10ohm load and at the time of application.  Shots are below:

    This scope shot is idle across the sense resistor @ 1uS divisions

    This scope shot is idle across sense resistor @ 100nS divisions

    This scope shot is at the connection of the 10ohm load (500mA) @ 1uS per division.  The voltage is jumping up to around 400-600mV, which converts to 2.2-3.3A.

    Thank you in advance for any assistance

    Mike

  • Hi Mike,

    One test I just thought of: when you put a 48V into your adapter connector, can you go up to 2A?
  • I have a 50V 10A supply, Hooked that up and had 5V on the output. Then I added my 500mA load and the voltage wasn’t regulating. It was getting to roughly 8.5V on a DMM. When this happens I’m hearing a buzzing from one of the inductors (or possibly an electrolytic cap). How did my scope shots and layout look? Thanks
  • Hi Mike,

    Looking at the waveform, at 500mA the CS voltage is reaching the TPS23750's current limit threshold (~0.5V). It's either the sense resistor is larger or maybe there is an incorrect inductance? Maybe it's 3.3uH and not 33uH?

  • I'll double check the values again.  Did you have any feedback on the pictures of the layout?

  • Darwin

    The sense resistor and inductor are correct.  When I look at the board using a thermal camera, I can see that the 10uH inductor is getting hot and is likely the source of the buzzing.  I keep asking about feedback on the layout to see if you see anything glaring there, but no response.  Any feedback on layout?

    Thanks, Mike

  • Hi Mike,

    The only thing that caught my eye on the layout is the feedback divider is tapping off of the diode (switching and noisy). I would recommend have R36 tap from the output cap.
  • Mike,

    It has been > 10 calendar days since our last correspondence from you.

    Please let us know if additional help is required or if we may close this thread now.
  • Thomas/Darwin My apologies for the delay, I meant to get back to you sooner. The problem ended up being the two 1uF caps between the inductor and the diode. These caps were too small accidentally so the board was loaded with two 10,000pF caps. Once I swapped these out it worked as expected. Thank you for all of your help in troubleshooting! Thanks, Mike