• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 64 subscribers
  • Views 333 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.

TPS23754: TPS23754 board came up -once-, then no response

Borge Strand-Bergesen
Intellectual 710 points
Part Number: TPS23754

Hi, 

I have copied your TPS23754EVM-383 / slvu304c.pdf reference design for the TPS23754. Two boards have been built this far, and they behave the same way, they powered up -once- but never again.

The first time I powered up the boards I got 12V out. That initial test took perhaps 8s to establish the DC output voltage after plugging in injector input power. I left it on for just enough time to measure the DC voltage, then turned the injector off. The 12V output is NOT LOADED by anything but an LED in series with a 10k resistor.

While the board was on I didn't measure VC voltage, but did see the VC LED turning on. Then I repeated the test with the other board with the same result. After that I have not been able to power them up at all, not even after letting the boards rest over night. Has anybody had the same issue?

There are no obvious signs of component damage.

In my test setup I use the Injector "Planet POE-171A-60". I have previously used the same type of injector successfully with an unloaded TPS23754EVM-383 kit. The injector has a "Standard" / "Legacy" mode switch. Both work with the reference design.

Compared to the reference design I have made the following changes in my design:
- VDD = VDD1, there is no externa supply like the reference design's J1
- R9 = 63.4k, slightly different switching frequency saves a BOM line
- APD = grounded through 10R resistor
- PPD = floating
- C9 at rectifier DC replicated 4 times for symmetry in layout
- C14 = 39uF, up from 22uF due to availability of vibration proof caps

I will be able to compare my implementation and the reference kit side-by-side. I'd appreciate your input on which measurements or modifications to do in order to have my board start up each time. 

Best regards, 

Børge

  • 0
    Intellectual 710 points

    Quick update:

    - Adding a 36R / 0.33A load on the 12V output changes nothing. When plugging in I never see any voltage change on the 12V output. It remains at 0V.
    - Rectifier at RJ45 connector measures fine with 0.25V forward voltage on all 8 diodes. 
    - Voltage across D6 consists of 2VDC + low-frequency noise. With CAT6 cable in place it is established 8s after injector power-on. With injector power in place it is established immediately after CAT6 insertion
    - I see no signs of protocol negotiations on voltage across D6 during power-on.

    I'm starting to fear that something was broken when injector power was first -removed-. Do you concur with this assumption? If so, which component could it be, and how should I protect it in the future?

    Thanks,

    Børge

  • 0 in reply to Borge Strand-Bergesen
    TI__Mastermind 21320 points

    Hello Borge,

    Listening to your description, it sounds like the problem happens during shutdown. Active clamp forward designs can have a voltage/current spike during shutdown and is a phenomena of the topology.

    I would first check the primary FETs, the secondary FETs and the secondary BJT's. The damage may not be visible but ensure these components still operate and are not shorted. 

    Then we can check the IC. Measure VC now that it is not working. Please let me know the results!

    If this post answers your question, please indicate so by marking this thread as resolved. Thank you.

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • 0 in reply to Michael Pahl
    Intellectual 710 points

    Hi Michael, 

    Thanks for getting back to me so soon!

    When I start it up from the PoE injector I have a noisy VC with a DC component of 2V. The noise has components around 2-4Hz and a Vpp voltage of 225mV. 

    I replaced first the secondary FETs, then the primary FET Q6, then D22, then the controller chip. I measured the secondary BJTs and diodes, they look OK with diode drops of some +-0.5V. As do primary FET Q1 and companion D1, but I replaced them too for good measure.

    For each round of changes I plugged in the injector again. Each time I got the same result, VC at 2V, no output voltage. 

    I even tried a backup injector of the same model. Same result. Just to be sure I measured that the transformer doesn't have an open. Next, I'll replace the secondary BJTs, zeners, shunt regulator, the works....

    NOW, while it would be very good to get my prototype board up and running, a permanent solution would be even better. What is the best way to safeguard an active clamp forward design? Alternatively, should I change topologies entirely? There is a chance the finished product will not need isolation, but I can't yet design that out. 


    Thanks,

    Børge

  • 0 in reply to Borge Strand-Bergesen
    Intellectual 710 points

    Hi, 

    I did find two bugs in my schematic which has revived one of the boards, the one where I didn't change all the active parts. With this I'm able to boot it up from Planet's "Standalone" mode a few times, and from their "Legacy" mode most of the times. 

    I'll file this issue as solved for now, but I wouldn't be surprised if additional issues come up with this design.

    Thanks,

    Børge

  • 0 in reply to Borge Strand-Bergesen
    Intellectual 710 points

    Michael, 

    what can one do to prevent turn-off from burning up the Active clamp forward power supply?

    Børge

  • 0 in reply to Borge Strand-Bergesen
    TI__Mastermind 21320 points

    A spike is a phenomena of the ACF topology. There are lots of approaches of handling it. 

    One is have a slower shut down (increase soft start and/or VCC cap). Another is to have higher voltage FETs.

    Those would be my suggestions!

    If this post answers your question, please indicate so by marking this thread as resolved. Thank you.

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments