TPS23753 Startup

A description of bootstrap startup and the conditions for continuous operation are discussed on the datasheet pages 16 - 18.  Vc must be maintained above (UVLO_1 - UVLO_H) = 9.3 - 3.7 = 5.6V for continuous operation after a startup cycle has started (see electrical characterisitcs table). 

Some generic advice for (all) bootstrap startup is to:  assure sufficient Cvc, minimize Cout, minimize load current and prevent load switchers from starting at low voltage, assure some current overhead over the loading, minimize all loading on Vc, use M1 with low Qg.  Reference disignators refer to Figure 1.

Any specific advice requires complete documentation (schematic (including any load capacitance and downstream converter), BOM if the schematic is lacking part numbers) and annotated waveforms.

Thank you for answer.

The schematic and waveform is attached.

8510.PoE_Circuit_TPS23753.pdf

4666.PoE_Startup.pdf

There appear a number of issues that may be contributors.

1)  PA1135 is "rated" for switching at 200kHz, while the R609 implies 125kHz.    Coilcraft POE70P-12L has a bias winding rated for 12V rather than the 8V of the Pulse xfmr.  Note it is rated for 250KHz operation.  A transformer change might help, but not might be the source of your problem.

2)  For the initial cycle, you have somehow gotten th Vc almost to the clamp voltage before start.  I am not sure how you did this.

3)   From the startup waveforms, it looks like your startup loading is too high.  Does a load converter start (make sure it is off until +12V is near is regulated value)?  Is Q601 Qg rating too high? Is C625 populated?  Is there additional output capacitance in the rest of the design (not shown)?

4) Once you straighten out the excessive output loading, additional things to look at are:  Adjust C614/R621 to just filter out the leading voltage spike as current limit is approached.  C611 could be made a little larger (e.g. 22uF).  Do you really need R616 at all - it looks a little large and we often make provision for a small capacitor from CS to RTN.   R620 seems small.  The compensation seems a bit unusual as well.

You might start with running the converter at the right frequency, eliminating excess output capacitance, making sure any load converters are not starting, checking Q601 Qg, fixing R616 and R620.  IF you get it almost starting increase C611.  If these do not work, maybe the Coilcraft transformer.  

They found that the situation be improved by replacing the trans.

But they can not replace the transformer because the MP board had made.

This phenomenon affects the rising time is slow, so they review of compensation parts.

They use this transformer and modify the circuit to avoid affecting the peripheral.

Thanks and Regards,

PAN