• State Resolved
  • Locked Locked
  • Replies 17 replies
  • Subscribers 39 subscribers
  • Views 340 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

PMP9563: Hiccup

nigel gordon
Prodigy 100 points
Part Number: PMP9563
Other Parts Discussed in Thread: TPS23754, LMV431A, TL431

Tool/software:

I have prototyped an exact replica of PMP9563 , but cant get it to work.
POE comes up ok but the TPS23754 cycles round and round. It appears to reset when VC droops to 8V after initially rising to 15V.
I tried increasing C22 to 47uF and changing R17 which just made the time to droop to 8V  longer.
I have no load connected.

Secondly, to eliminate overcurrent I shorted out the current sense resistor but was quite surprised that it caused something to fail because now POE doesnt even come up. 
Surely that shouldn't cause a device failure?

  • 0
    TI__Mastermind 37121 points

    Hi Nigel,

    Thanks for reaching out!

    It may be caused by the load if you connect the PD with a PSE. PSE may turn off the port if the PD draws too less current (maintain power signature).

    Also it would be helpful if you can share your schematic.

    Best regards,

    Diang

  • 0 in reply to Diang Xing
    Prodigy 100 points

    POE is up and doesnt turn off, its the TPS23754 which turns on and then off when VC drops to 8V The cycle repeats. Why should that be?
    Thats my main issue.

    My other issue is that in order to diagnose the above, I shorted the current sense resistor on one board, but that seems to have blown up the TPS23754 and on that board ( I have multiple boards) the PSE fails to power on because it doesn't receive the signature any more.
    Why has the TPS23754 blown up?

  • 0 in reply to nigel gordon
    Prodigy 100 points

    How do I attach the schematic?

  • 0 in reply to nigel gordon
    Prodigy 100 points

    What is the point of R22 on PMP9563? The TPS23754 evaluation board doesnt have it.

  • 0 in reply to nigel gordon
    TI__Mastermind 37121 points

    Hi Nigel,

    Sorry I mis clicked the "TI thinks Resolved". 

    Please see my comments below:

    nigel gordon said:
    How do I attach the schematic?

    You can drag the .pdf to the E2E text box

    nigel gordon said:
    Why has the TPS23754 blown up?

    TPS23754 uses current sense for peak current mode control. Short the CS to RTN will result control misfunction. Max duty cycle can blow up the board.

    nigel gordon said:
    What is the point of R22 on PMP9563?

    R22 is feedforward resistor, it should be connected to VDD to CS, using 750k Ohm. It is optional and used for better FB control / regulation

    Best regards,

    Diang

  • 0 in reply to Diang Xing
    Prodigy 100 points

     

    The symptoms of the blown board are that POE does in fact come on for a short time, so it must pass classification, but because the TPS23754 doesnt start switching, I guess the load doesnt't change so the PSE switches off again. I didnt expect a PSE to do that.
    The primary external MOSFET cant have blown , neither the transformer because the PSE cant provide that much current. Maybe the TPS23754 internal MOSFET blew. I will replace it.I see 825mA is specified.

    OK so my real problem is with a board with a good TPS23754.I attach the circuit.Please can you advise why VC drops to 8V.
    .e2e.ti.com/.../7485.Schematic-Prints.pdf

    I notice that the eval board uses PA2649NL  but the reference design uses PA4065NL transformer. Also the evaluation board has a transient suppressor between VSS and RTN. No such device on the reference design. I have seen other designs with a schottky diode in that place. I assume the internal MOSFET has a diode across it?
    The components in front of the opto are also different. Like I said, I copied the reference design.

    I spotted something. My reference is different! I use LMV  a 1.25V reference instead of a 2.5V. I will change that.

  • 0 in reply to nigel gordon
    TI__Mastermind 37121 points

    Hi Nigel,

    Thanks for your reply. I will get back to you in 2-4 business days.

    Best regards,

    Diang

  • 0 in reply to Diang Xing
    Prodigy 100 points

    Hi, so changing R561 to I.2k because we use LMV431A and R547 to 5.6k because we use PS2801C (higher CTR), I get 12V output with no load.(Major step forward)
    If I connect a load though,  Q17 blows up. Why?

  • 0 in reply to nigel gordon
    Prodigy 100 points

    Part of this is loop instability. Is there a safe compensation network? I notice the network is totally different on the eval board compared to mine copied from PMP9563.
    Question... why dont you set the minimum bias current for the TL431? (ie >1mA) (My LMV431A has to be >50uA)

  • 0 in reply to nigel gordon
    TI__Expert 7121 points

    Hi Nigel,

    Diang is OoO today and he will get back to you by this Thursday.

    BR,

    Ke Wang

  • 0 in reply to Ke Wang
    Prodigy 100 points

    OK, here are some more differences between the evaluation board and the reference design.
    The transformer is different..
    The zener diodes in the reference design are 10V, the evaluation board has 12V.
    The reference design has a 100pf across the other BJT in the secondary compared to the evaluation board. Is this why my BJT blows up because of the 100pF I copied from the reference board?
    The compensation components around the TL431 are completely different but that could be because of the opto difference I guess?

  • 0 in reply to nigel gordon
    TI__Mastermind 37121 points

    Hi Nigel,

    Thanks for your help to list the difference of TI's deign and your schematic!

    Q17 is the 100V BJT on the gate driver circuit. It normally is robust unless the FET's gate is damaged too. Could you check if Q16's DS or GS are shorted?

    The compensator values are different but their values look reasonable. 

    The failure may be related to the overshoot considering there is no RC snubber circuit on the Q16 or Q21.

    Can you check if the Vds of Q16 and Q21 exceed 100V especially during start up or shut down? If so, you may add a 100V 2.2nF + 47 Ohm 0805 resistor to see if it helps.

    Best regards,

    Diang

  • 0 in reply to Diang Xing
    Prodigy 100 points

    I tried the snubbers across DS of each MOSFET but no good.
    If I touch my scope earth to 0V (no load connected)  Q16 and Q17 blow straight away!!!!

    Should I consider changing the transformer from PA4065NL to the PA2649NL used on the eval board? Its 2:1 instead of 1.58 to 1.

    I just discovered that if C599 is shorted then everything is good.  With it causes severe instability.


    Also what is C600 for? The eval board doesnt have that one..

  • 0 in reply to nigel gordon
    TI__Mastermind 37121 points

    Hi Nigel,

    Thanks for your updates. The PA4065NL with 1.58:1 should work for 33 - 57Vinput and 12V output based on datasheet. 

    C599 and C600 are Y-cap for creating a common mode noise return loop that DC/DC switching caused. These Y-caps cannot be shorted otherwise the DC/DC could have a malfunction. You can make them open circuit if do not need to run EMI test. 

    Best regards,

    Diang

  • 0 in reply to Diang Xing
    Prodigy 100 points

    Tried both caps open circuit. No good. I have to short C599 and everything is good. I lose isolation doing this which is not good.

  • 0 in reply to nigel gordon
    TI__Mastermind 37121 points

    Hi Nigel,

    Thanks for your reply. It is very strange that when C599 is shorted and everything works.

    Maybe a +DC bus on the primary & secondary is wrongly connected? 

    Could you also share the circuit from RJ45 to TPS23754?  

    Best regards,

    Diang

  • +1 in reply to Diang Xing
    Prodigy 100 points

    I thought C301 and C324 could be the culprits, so I removed them but it made no difference.

    I do have a lot of output capacitance elsewhere on the circuit. I will try removing some of that.


    6837.LAN_Ports.pdf


    I got it, it was the POE_SYNC upsetting the TPS.