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.

TPS23730: Non-isolated PoE Pd

Part Number: TPS23730
Other Parts Discussed in Thread: TPS23755, TPS23753A, TPS2372, TPS2372-3EVM-757, LMR36520, TPS2373, TPS23753, PMP40355, TPS2372-4EVM-006, TIDA-050045

Hi All,

is it possible to use TPS23730 in Non-isolated PoE Pd design? Do you maybe have some reference for Non-isolated PoE Pd?

Best Regards,

David.

  • Hey David, 

    Yes it is possible to use the TPS23730 in a non-isolated design, but it is not typically used here. At this time we don't have any designs, but it can used in a buck. The second gate drive will have to be disabled, and use the feedback directly. You can look at some of our buck designs with the TPS23755 or TPS23753A. 

    More typically customers use a stand alone PD (TPS2372), and then pair it with a buck controller (LMxxx). Check out the TPS2372-3EVM-757 for an example. 

     

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • Hi Michael,

    thank you for the info. Can i ask some more questions:

    1) Does the TPS23730 has any additional features regarding to TPS2372 ?

    We can't use TPS23755 or TPS23753A because we need PoE PD  802.3bt Type 3 Class4 40W

    2) What do you think can achieve better efficiency TPS23730 vs TPS2372 + LMR36520

    or would you suggest rather something like this:

    3) Regarding Isolation, is it ok to use non isolated PoE PD regarding to the Ethernet alliance PoE Certifications (https://ethernetalliance.org/) and the 802.3bt standard?

    Because i saw that the TI TPS2372-3EVM-757 EVM is non isolated PoE PD device and it is certified under ethernet alliance number Logo Certification # 202176913

    Best Regards,

    David.

  • Hello, 

    1. Yes the TPS23730 has a DCDC controller integrated into it, so there are a number of features associated with it. Comparing the TPS23730 and TPS2372 is not a fair comparison. 

    Yes I know you cannot use these parts, I mentioned them as examples of a PD + PWM device used in a buck. 

    2. It is hard to speculate because efficiency is dependent on every component choice. A synchronous buck controller requires a PWM square wave to control the operation --- the LM3650 does this as its only function, and the TPS23730 does this as well as the PoE negotiation. 

    3. Yes, DCDC isolation is not part of the IEEE standard. 

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • Hi Michael,

    thank you for the feedback.

    I'm trying to make a demo board and test the TPS23730 in non isolated buck design.

    Previously you said that Ti has non isolated  buck designs with TPS23755 or TPS23753A. but i can't find them in the reference library, could you point out all the non isolated buck designs?

    Also you said that the GAT2 needs to be disabled, does this mean that it is not possible to set the  TPS23730  as synchronous buck with GATE ( high-side mosfet) and GAT2 ( low-side mosfet)  and vice versa( and adjusted DT) ?

    Best Regards,

    David.

  • Hi Michael,

    we had internal discussion , maybe it is easier to go with PoE frontend ( TPS2372 or TPS2373) + buck DCDC (LMR36520).

    Which one would you recommend TPS2372 or TPS2373 for non isolated design?

    TPS23730 has " Soft-start control with advanced startup and hiccup mode overload protection", does the TPS2372/3 also has this feature?

    Are there any other special feature that the TPS23730 has over TPS2372/3 that is not associated with integrated DCDC?

    Best Regards,

    David.

  • Hey David, 

    The TPS23753 is used in a buck in the PMP40355. I can private message you the TPS23755 buck. 

    I guess you could use the secondary gate drive to make a synchronous buck in the TPS23730. Typically the design choice behind bucks in PoE is to minimize cost. So when I encounter bucks, they are not synchronous, and definitely not using parts with multiple gate drives :) Usually if efficiency is of concern, most customers just jump to a flyback. But yes in theory you could make a synchronous bucks -- we just don't have a reference design to follow for you. 

  • Hey David, 

    For TPS2372 vs TPS2373, I would recommend the TPS2372 for a buck. The TPS2372 does not have advanced startup -- this feature is optimized for flybacks and ACF's, and would need to be disabled if used with a buck. So the TPS2373 is extra cost and extra parts for a feature you're not using. So we recommend TPS2372. 

    The TPS23730 has a programmable UVLO for shutdown, and not all DCDC controllers have that. The TPS23730 also has SSFD (dithering), which is part of the DCDC but not all controllers have it. 

    The TPS23730 can also present the TPH/TPL signal in series or parallel, and the TPS2373/TPS2372 can only send it in parallel. 

    The advanced startup and hiccup features are for the DCDC converter portion. The TPS2373 /TPS2372 does have overcurrent feature, but it is for total power through the input. The TPS23730 is monitoring both this current and the primary DCDC current, so it just depends on the buck controller you choose. 

    But as I go through our reference designs, most of them are standalone PD's and then buck controllers for non-isolated designs.  I think the nice thing here is you can basically pick the TPS2372-4EVM-006 and copy it directly, and that choice is independent of the buck. You can work with our buck controller team to find the best buck solution, and just stick it on the output of the TPS2372-4EVM-006. So hopefully it is more plug & play. I think that is more a typical trend in PoE, because as I said, bucks are typically low cost designs, so the attention to DCDC design details is usually not the focus vs cost down + size reduction. 

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • Hi Michael,

    one more thing regarding isolation :) ( i'm not 100% clear with that)

    You stated  previously " Yes, DCDC isolation is not part of the IEEE standard. " while you collage provided this from the standard

    So our device has only one connector ( Ethernet Connector) , and there are no accessible external conductors, but it is in metal housing (Metal housing from connector is connected to metal housing of the device), is it possible to use non isolated PoE PD DC/DC in this situation?

    Because the standard mentions frame ground. is this referred to metal housing ? 

    Does this meant that the TPS2372-3EVM-757 can't be used in metal housing where RJ45 metal part is connected to metal housing, but it only can be used in plastic housing?

    Best Regards,

    David.

  • Hey David, 

    Our guidance is that if the end user has access to the ground plane or if there is any exposed metal, the design should be isolated. For example, a USB port. 

    You can have a non-isolated DCDC and still isolate your USB port, so the standard is not forcing you to use a flyback. That's what I was trying to say, is that the standard does not force the DCDC topology to be isolated. However, if you need isolation because of other factors, might as well make the DCDC isolated since that will probably deliver the overall best solution. Most designs are isolated. 

    If I were designing the product, with the given grounding scheme, I would isolate it. To comply with the standard, the metal casing would need to be isolated from the PoE input, and if the DCDC isn't isolating it then you have to come up with another way, and personally I can't think of anything cost effective. The only real benefit of the buck here is board space / height. but an isolated DCDC will give you better thermals and better efficiency, so to me the trade off is worth it. 

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • Hi Michael,

    we don't have any other connector then EHT connector and metal housing, please see the block diagram below.

    -> Metal housing is touching the metal shield of the connector

    -> The primary termination of the PoE magnetic is connected to the metal housing through 2.2nF 2kV capacitor

    -> System GND is connected to metal housing with 1nF 2kV capacitor through the metal screw

    The User don't have direct access to system GND plane but to metal housing, The system GND is floating and has a high voltage cap 1n 2kV that connects System GND to the metal housing ( so i'm not sure is this is proper isolation for the System GND to metal housing e.g. frame ground that is mentioned in the standard? )

    1) You say that the standard not needing to be isolated:

    "You can have a non-isolated DCDC and still isolate your USB port, so the standard is not forcing you to use a flyback. That's what I was trying to say, is that the standard does not force the DCDC topology to be isolated"

    and then later: 


    "To comply with the standard, the metal casing would need to be isolated from the PoE input, and if the DCDC isn't isolating ... "

    Which standard 802.3bt ? or an other?

    2) we have to problems

    2a) Efficiency -> the Non isolated approach yields better efficiency because

    -> if i only have 5V in the system maybe we get 1-2% better with only flyback

    -> but i also need 120V and thus when using isolated design i have to do 48V to 24V and then to from 24V to 5V, and 24V to 120V

    -> and if i have non isolated than i can do 48V to 5V and 48V to 120V so i don't have to cascade DC/DC converter.

    2b) Space, the transformer for 45W or 51E PD is big

    2c) thermals, is directly tied to the efficiency.

    3) Can you please comment on the above requirement and block diagram if this could comply with the PoE and ETH standards?

    Best Regards,

    David.

  • David, 

    1.The standard you posted is the IEEE802.3.bt standard. 

    You stated  previously " Yes, DCDC isolation is not part of the IEEE standard. " while you collage provided this from the standard

    So

    2a.) For 5V/50W output, our TIDA-050045 gets 94% efficiency on the converter. I don't know of a 50W buck close to this. Bucks are typically in the mid to high 80% range. 

    I am not sure why you couldn't do 48V-5V with an isolated solution, and 48V to 120V with a non-isolated solution?

    b.) for smaller space transformers you could look at planar transformers. They are more expensive but smaller. 

    3.) yes seems fine. If you are wanting to ensure your design is IEEE802.3.bt compliant you can get it Ethernet Alliance Certified: 

    https://training.ti.com/zh-tw/introduction-poe-certification-program 

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • Hi Michael,

    2a) ok tis design looks good, but i don't need 50W on 5V but only 5W. What is the point if in the same device i do one portion isolated ( 48V to 5V) and the other part non-isolated (48V to 120V) ? Because the 48V is form PoE and if do BOOST from 48 to 120V then this is not isolated from PoE or i'm wrong? (note the 120 V boost is adjustable it can go from 96V to 144V, so it can't be flayback) 

    Either i do non isolated as i draw on the block diagram , or i do everything isolated 48V to 24V Isolated flyback and then 24V to 5V (buck) and 24V to 120V (boost)

    2b) yes we are also looking into that

    3) Just to confirm , you think that my block diagram (non isolated design in metal housing) is ok regarding to 802.3bt and could get Ethernet alliance certified?

    Best Regards,

    David.

  • Hey David, 

    1. if you only need 5W, then maybe a flyback would be better. I assumed you needed this power level based on the quote below. 

    We can't use TPS23755 or TPS23753A because we need PoE PD  802.3bt Type 3 Class4 40W

    the 120V could be isolated / floating from the 5V plane. Ultimately you know the grounding scheme better than I do, so it is Up to you on how you want to proceed with cost and the isolation requirement. 

    3. Appears acceptable but I cannot speak on their behalf. 

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments 

  • Hi Michael,

    "3. Appears acceptable but I cannot speak on their behalf. "

    do you maybe have some contact that you could share through PM, so we can contact regarding the certification and if this is possible with our block diagram?

    Best Regards,

    David.

  • Hey David, 

    I think the best way to get into contact is through their website: https://ethernetalliance.org/poecert/ 

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

     

    Regards, 

     

    Michael P.

    Applications Engineer

    Texas Instruments