TPS23756: Please help to find the problem for my POE

Thank you Darwin,

I will check other parts about the short.

The circuit has no load now. After the checking I may put a 5W load at the output and see.

Bests,

Kevin

Good day Darwin,

I tested the circuit based on your suggestion(the part after the primary transformer). The way what I did:

Apply external DC power supply to the circuit. DC power supply setting: +5V, current limitation 200mA.

1) remove L2_POE inductor, apply +5V DC on the output terminal of the whole circuit (5V net in the schematic), the current consumption is lease than 6mA.

2) Put L2_POE inductor back, the current raise to the limitation 200mA, voltage drop to ~3.8V.

3) remove Q5_POE(Partnumber: SI4894BDY-T1-E3 ), Q2_POE(Partnumber: ‎SI7852DP-T1-E3) the current consumption is back to 6mA. 

Voltages: Q5_POE: Vgs = 0.04V; Q2_POE: Vgs = 4.6V.

4) put a new Q5_POE, Q2_POE on, the current goes up to 200mA, voltage drop to ~3.8V. If I increase the power supply's current limitation to 1A, the voltage still clip to around 3.8V~4.0V with 1A.

Voltages(at 200mA): Q5_POE: Vgs = 0.08V; Q2_POE: Vgs = 2.67V.

5) replace D10_POE, D11_POE(Partnumber: BAS16LT1G ), Q3_POE(Partnumber: MMBTA06LT1), and D15_POE, D16_POE(Partnumber: BZX84C12LT1G), still 200mA/3.4V.

6) R2_POE, R4_POE are tested as they should be. C14_POE, C16_POE are tested as unbroken.

Comparing: The EVM can stand at +5V with the same power supply connecting to the EVM's output terminal after 3 seconds.

EVM voltages: (Q4, same as Q5_POE)Vgs = 1.92V/  (Q2, same as Q2_POE)Vgs=2.27V. 

Looks my Q2_POE/Q5_POE are at an undesired situation.

Anymore suggestions?

Regards,

Kevin

Hi Darwin,

Do you know why SIR606 (Q2 on EVM schematic, the print on the chip is R606, so I guess its SIR606) is used instead of the SI7852DP(the print is 7852) on the TPS23765 EVM?

I took this chip(SIR606) out from the EVM, and placed the chip I bought(SI7852DP) on to the EVM, the EVM didn't work (which same as my PCB);

I put the EVM's original Q2 back to the EVM, the EVM worked again.

Could you please give me the BOM the EVM using?

Bests,

Kevin

Sorry I need to call back previous analysis -- previous 3 posts.

The problem doesn't locate behind the secondary transformer. Indeed it locates before the primary transformer.

I removed the pads connection from the secondary transformer pins, and floated them;

I soldered the +5V DC to the pads to mimic the output of the transformer, and there was no short; the current was 6mA, voltage was 5V.

Then I connected the POE Ethernet cable to the PSE, still the output of the PSE was not constant. The voltage of 50V pulse was there which indicated there was no short on my POE board. Also the multi-meter testing proved that.

I will check if any other problems. It may be the broken chip, or the wrong configuration. 

If you can provide any help based on my first and this posts of this thread, I will be thankful.

Regards,

Kevin

Hi Darwin,

The chip TPS23756 was broken. The pin 5 GATE to ARTN/RTN was about 500 ohms. Remove the chip, test the MOSFET, the Rgs of the MOSFET was correct (as open circuit).

Put a new TPS23756 on it. Then the PSE could constantly provide the 50V. However, this time the TPS23756 broken again. The resistor at the the GATE is 140 ohms to RTN.

Output from Vc(0.12V), Vb(0V), CTL( -0.12V, negative) are wrong; Vdd -Vss = 50V, but Vdd-RTN=40V.  V_RTN - Vss = 10V.

Before I turn on the PSE, the resistor between GATE to RTN is normal as the EVM (about 0.95M).--- Before I turn on the PSE, I tested all pins, the resistors value to RTN/Vss of my board were all same as the EVM.

I feels like whenever I turn on the PSE, the chip TPS23756 will be burned on my board. Anymore suggestions?

Thanks,

Kevin

From the document, the DEN pin should be high (pulled up by 24.9k to Vdd), but my board has 0V  between DEN - Vss. Remove the PSE and measure the resistors, DEN to Vss is couple M ohms; the pull up resistor is 24.9K.

Good morning Darwin,

Thanks for your analysis and suggestions.

I just read the document and found that if the TPS23756 thought there were short circuit, the chip would go to the protection mode, then everything would be abnormal. --- Which just as you said.

For my current board, since the offline resistors from the pin VC/GATE etc. to RTN(270 Ohms/140Ohms) are not correct now(measured by multimeter), I am assuming the replaced chip has gone. Just waiting new order of the chip coming. Whenever I get the new chip, I will test the waveforms as you mentioned.

Could you please let me know how to test them safely? I mean which components I should remove from the circuit, so if I connect the PSE to the PD, the TPS23756 will not be damaged.

The photocoupler(U2) is the TCMT1107 as the document suggested. 

You mentioned "did you try removing the pull down resistor that you added to the gate?". Maybe my explanation was not clear. There is no added resistor to the gate. I just use the multimeter to measure the resistor between the pin GATE to pin RTN on my board, and compare the result with the EVM. (to see if the chip is still good or not).

If you means other pull down resistor, please tell me which its.

Regards,

Kevin

Dear Darwin,

I appreciate your attention and replies. The problem has been solved with the new chips. Not sure what happened about the old chips. Maybe the heat gun's temperature is too high or some mysteries.

The chips I replaced includes the main chip TPS23756, the three diodes BAS16/BAS20(D5, D6, D9 in the EVM schematic, or the D12_POE, D17_POE, D20_POE on my board) at the primary side, and the MOSFET controlled by the GATE (Q5 in the EVM schematic, or the Q4_POE on my board).

Since this board is just a prototype soldered by hand in our lab, so I believe there would no such problem in our mass product with SMT stream line.

Again, thank you for your helps, and have a great weekend :-)

Sincerely,

Kevin