TPS2373: Schematic review

aviv frishman

Part Number: TPS2373
Other Parts Discussed in Thread: PMP22477, PMP20859, , TPS2372, LM5022, TL431, LM5122

Hello,

Hello, I need your assistance in reviewing the schematic design.

Specifically, I need you to focus on the TPS2373-3RGWT (U11) component, the 'Power Input Decision' block (Q4), and LM5022MM (U3).

The design should support the following options for power input:

1. DC external adapter only or POE only.

2. The DC external adapter should be prioritized over the POE input, and a seamless transition between them should be implemented.

2.1. When the DC adapter is connected, there should be no PSE detection for POE.

2.2. When both DC and POE are connected on power-up, the DC external adapter should be prioritized.

2.3. If DC and POE are connected and the DC adapter is disconnected, there should be a seamless transition to POE.

Please let me know if you need any further information.

Thank you

jl-pc0002471 power board poe pd.pdf

over 1 year ago

0 Praneet Kumar over 1 year ago

TI__Genius 16370 points

Sorry, team is out of office today 'coz of US holiday - Please allow a delayed response to your post.

0 Diang Xing over 1 year ago

TI__Mastermind 37121 points

Hi Aviv,

Thanks for reaching out.

We will help to review the TPS2373-3 related circuits and get back to you by 27th/Feb.

Best regards,

Diang

0 Diang Xing over 1 year ago in reply to Diang Xing

TI__Mastermind 37121 points

Hi Aviv,

Thanks for your patience.

If you want to enable the adapter -> PoE seamless transition, the APD pin could not be used considering it will turn off the PoE hotswap.

To have PoE <-> adapter seamless transition and adapter priority at Class-8, you may refer the below two approaches:

- PMP22477 has a 12VDC adapter connected to the output. You can use a 48-12V Buck to replace adapter. You also have to ensure the PoE converter voltage is lower than the adapter voltage. The PoE converter can be set lower if the application allows or the feedback can be modified to switch the PoE output lower when the adapter is connected. TPS2373-4 could be used in this case.

https://www.ti.com/tool/PMP22477?keyMatch=&tisearch=search-everything&usecase=refdesign

- PMP20859 has a good number of external circuits to achieve seamless transition. It is designed for a 48V adapter, and PMP20859 will also work with one PoE input and the adapter. TPS2372-4 could be used in this case.

https://www.ti.com/tool/PMP20859?keyMatch=&tisearch=search-everything&usecase=refdesign

Best regards,

Diang

0 aviv frishman over 1 year ago in reply to Diang Xing

Prodigy 50 points

Hi Diang,

Thank you for elaborating on other options to solve the design requirement, but I'm confused about the answer.

Our previous design was around TPS2372, and we have been advised to change it to TPS2373 (see the previous case link below).

It was a bit long time ago, so I've opened a new case here.

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1244625/tps2372-advanced-adapter-oring-solutions-using-the-tps2372-3

please your advise.

Aviv.

0 Diang Xing over 1 year ago in reply to aviv frishman

TI__Mastermind 37121 points

Hi Aviv,

Thanks for your reply.

The PoE -> adapter seamless transition & adapter priority can be easier achieved in TPS2373 since the APD pin is used. But the drawback is APD pin will turn off the internal PD hotswap to disconnect the PoE, so I have concerns for adapter -> PoE seamless transition.

I will test the TPS2373-4EVM and keep you posted.

Best regards,

Diang

0 Diang Xing over 1 year ago in reply to Diang Xing

TI__Mastermind 37121 points

Hi Aviv,

I tested TPS2373-4EVM, adapter <-> PoE transition from both direction are seamless, and the adapter has the higher priority.

So the TPS2373-4 should be the correct unit for your case.

Sorry for the extra time it took. I will get back to you for the review comments by next Monday.

Best regards,

Diang

0 Diang Xing over 1 year ago in reply to Diang Xing

TI__Mastermind 37121 points

Hi Aviv,

Thanks for your patience. Please find the review comments below:

1. The LM5022’s startup circuit may have conflict with TPS2373-4’s startup circuit. Please also check with LM5022’s engineer if LM can be only drive by VCC.

2. At the startup or stop, the Vout could be below 15V, and it may cause TL431 in off state. It may be risky in startup or stop. Please double check.

3. To ensure “When the DC adapter is connected, there should be no PSE detection for POE.” D18 may not be used. Without D18, the VDD-VSS will be charged by DC adapter and make sure the PSE-PD detection fail. You may consider add a 0-Ohm in parallel and make D18 DNP.

Best regards,

Diang

0 aviv frishman over 1 year ago in reply to Diang Xing

Prodigy 50 points

Hi Diang,

Thank you for the review.

From what I understand, there is no issue with TPS2373 in this design besides an additional resistor to bypass the diode. We will implement this suggestion.

Regarding LM5022, we are based on PMP23181 we got internally from TI, but how can I reach out to LM5022’s engineer for additional validation? Should I open another case?

I should point that this design is already produced and checked while the only issue was the adapter -> PoE seamless transition.

Thank you,

Aviv

PMP23181AP(001)_Sch.PDF

0 Diang Xing over 1 year ago in reply to aviv frishman

TI__Mastermind 37121 points

Hi Aviv,

Thanks for your reply.

PMP23181 uses TPS2372-4 which does not have the integrated startup circuit (VC_IN, VC_OUT pins). So the LM5022 needs to do the start up by itself. While in your design the TPS2373-4 is used. TPS2373-4 has integrated startup circuit which may conflict with LM5022's startup circuit at VIN pin. I know to use the adapter priority function (APD pin), TPS2373-4 is needed.

I will forward this thread to the LM5022 team.

Best regards,

Diang

0 Stefan Schauer over 1 year ago in reply to Diang Xing

TI__Guru 58641 points

Hi Aviv,

I do not fully understand the whole schematic, so can not comment on the whole system.

I will check with design if there is an issue the suppling the device via LM5022 only and let you know in 1 -2 days.

Best regards,

Stefan

0 aviv frishman over 1 year ago in reply to Stefan Schauer

Prodigy 50 points

Hi Stefan,

I appreciate your help. If something needs further explanation, please update me.

Aviv.

0 Stefan Schauer over 1 year ago in reply to aviv frishman

TI__Guru 58641 points

Hi Aviv,

i am waiting for some feedback from design - just sent an reminder.

Best regards,

Stefan

0 aviv frishman over 1 year ago in reply to Stefan Schauer

Prodigy 50 points

Hi Stefan,

To rush the process since we must release the design urgently, we will remove the advanced startup feature of the TPS2373 and leave the design as before with the additional feature of an APD pin we must use.

From LM5022 point of view, we will leave the design without changes according to PMP23181AP(001)_Sch.PDF

From the TPS2373 point of view, we will connect Vc_in with zener + R to VDD and Vc_out to RTN.

This is per Diang's recommendation here - https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1243114/tps2373-unused-vc_in-pin-termination/4700823#4700823

also here - https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/769359/faq-tps2373-wiring-of-vc_in-and-vc_out-when-no-transfromer-aux-bias-supply-present?tisearch=e2e-sitesearch&keymatch=tps2373%2525252520VC_IN#

Please approve if this is acceptable. Schematic :

JL-PC0002471 POWER BOARD ALDER POE PD.pdf

Thank you,

Aviv.

0 Stefan Schauer over 1 year ago in reply to aviv frishman

TI__Guru 58641 points

Hi Aviv,

the LM5122 part of the schematic looks good from my side but as I do not know the function of the other devices and also do not know the requirements and power profiles associated with POE, I can not comment on this parts.

Best regards,

Stefan

0 aviv frishman over 1 year ago in reply to Stefan Schauer

Prodigy 50 points

OK, thank you very much for your help, Stefan.

Diang is the design approved from your side?

Thank you,

Aviv.

0 Diang Xing over 1 year ago in reply to aviv frishman

TI__Mastermind 37121 points

Hi Stefan,

My biggest concern is the startup circuit of LM5022. Looks like it has its own startup circuit on the VIN pin. TPS2373-4 has a start up circuit too which can provide a 10 - 15V on the LM5022's VCC pin. And if LM5022 start up earlier than TPS2373-4's start up, it may cause PoE compliance issues such as the inrush current limit. I am wondering if it is ok to remove LM5022's start up circuit at VIN and only power it on VCC?

Best regards,

Diang

0 Stefan Schauer over 1 year ago in reply to Diang Xing

TI__Guru 58641 points

Hi Diang,

so if the TPS2373 would also control the UVLO then the LM5122 would only startup once this signals get above the threshold level.

I think this should work for that case.

Best regards,

Stefan

0 Diang Xing over 1 year ago in reply to Stefan Schauer

TI__Mastermind 37121 points

Hi Stefan,

Thanks for your reply.

Yes, the PG pin from TPS2373-4 will control LM5122's UVLO too. It pulls UVLO low during the inrush period. After PoE inrush, the TPS2373-4 will power LM5122 by a 10 - 15 V VCC.

So is it ok to have TPS5122's VIN pin open?

Best regards,

Diang

0 Stefan Schauer over 1 year ago in reply to Diang Xing

TI__Guru 58641 points

Hi Diang,

I still recommend to have VIN connected to the supply but control the start of the LM5022 (Note: mentioned above the LM5122 instead of the LM5022) with the EN pin.

Please do not keep VIN open but connect it with VCC - see also datasheet:

7.3.1 High-Voltage Start-Up Regulator

Best regards,

Stefan

0 aviv frishman over 1 year ago in reply to Stefan Schauer

Prodigy 50 points

Hi all, I really appreciate your feedback,

To summarize this discussion, the changes are -

* from LM5022 side:

- VIN connected to VCC and controlled by the PG from TPS2373 (per datasheet section 7.3.1)

- VCC connected to VC_OUT of TPS2373 for external power (per datasheet section 7.3.1)

* from TPS2373 side:

- VC_IN connected to the auxiliary winding

- VC_out connected to LM5022

* We will keep the option (as DNP) to remove these connections and tie LM5022 to AUX winding and VC_IN from TPS2373 to VDD using R + Zener.

Thank you,

Aviv.

0 Diang Xing over 1 year ago in reply to aviv frishman

TI__Mastermind 37121 points

Hi Aviv,

Thanks for summary.

Yes. It looks correct to me.

Best regards,

Diang

0 Stefan Schauer over 1 year ago in reply to Diang Xing

TI__Guru 58641 points

HI Aviv,

point 1 and 2 for the LM5122 section looks contradicting.

So it should be VIN and VCC connected to VC_OUT of TPS2373 for external power (per datasheet section 7.3.1) - right?

I also recommend to have UVLO connected to VIN and VCC - UVLO detection should not be required from the LM5022 as this is done by the TPS2373 -right?

Best regards,

Stefan

0 aviv frishman over 1 year ago in reply to Stefan Schauer

Prodigy 50 points

Hi Stefan,

This is correct; VIN and VCC are connected to VC_OUT. UVLO is also connected to VIN and VCC.

Thank you,

Aviv.

0 Stefan Schauer over 1 year ago in reply to aviv frishman

TI__Guru 58641 points

HI Aviv,

thanks for the update.

Best regards,

Stefan

+1 Diang Xing over 1 year ago in reply to Stefan Schauer

TI__Mastermind 37121 points

Hi Stefan and Aviv,

One minor thing:

I would recommend to have UVLO connected to PVDD. This will be the same circuit with R5 and R4 now.

If UVLO is also connected to VIN and VCC, when PVDD drops TPOS2373-4 will keep trying regulate VC_OUT and making VCC no drop. In this case you may have a low V_in and high I_in in this DC/DC converter. Connecting UVLO to PVDD can turns off the DC/DC when V_in is low enough.

Best regards,

Diang

0 aviv frishman over 1 year ago in reply to Diang Xing

Prodigy 50 points

Hi Diang,

We are adding this option to the design.

Thank you,

Aviv.

0 Stefan Schauer over 1 year ago in reply to aviv frishman

TI__Guru 58641 points

Hi Aviv,

just would like summarize the connection of the LM5022 and the TPS2373 again.

LM5022 VCC and VIN are connected together and supplied from the TP2373

LM5022 UVLO still has the Resistor divider to PVDD but also the UVLO gets controlled and pulled low with the PG (PowerGood) from the TPS2373.
This ensures that UVLO does not get high before VIN gets high but let the LM5022 stop operation if PVDD drops during operation.

Best regards,

Stefan

Power management

Power management forum

TPS2373: Schematic review