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.

UCC28782EVM-030: Using a series pass MOSFET to interface with USB sink

Part Number: UCC28782EVM-030
Other Parts Discussed in Thread: TPS55288, UCC28782, LMG2610

Hello: 

In the EVM mentioned above, there is a series pass MOSFET controlled by WT6636 from Weltrend.

For many reasons, our configuration is different- and my proposed approach to connecting the USB C PD sink is NOT to use that series pass MOSFET- because it is superfluous when you are using TPS55288 as the interface. This assumes different programming of WT6636 of course. 

I do fear that I might not have understood all the needs of the series pass although I do not see anything arising out of the USB C  PD Specifications.

Here are the scenarios one can postulate with TPS55288 connection controlled by WT6636 with or without series pass MOSFET

#1 Opon AC connection, UCC28782 module in our application, turns on for an output of 24V. That is how UCC28782 is set up in our application.

#2 TPS5588 remains OFF since I2C is not active. VREF is 0, VBUS is 0, which is VSafe0 on the USB C PD sink side.

If a sink device is connected, there is no conflict even if a series pass was present. In this state, WT6636 is not turning the switch ON anyways. If the sink has a pre-existing voltage, the series pass body diode will make the switch state ON/OFF IRRELEVANT.

#3 WT6636 does not control our UCC28782 module. . But, Vout of ACF is connected to bias the WT6636.

WT6636 is programmed to run through C1-C2 upon being biased up, its algorithm runs through USB C requirements of the Startup dialog. 

# 4 VBUS may be changed to Vsafe5V if the connection dialog allows it  to do so. WT6636 controls TPS5588 : so it can be programmed to put VBUS at Vsafe5. Again, the extra switch contributes nothing in this state that TPS55288 output cannot totally satisfy. No matter what the sink device is.

When WT6636 would have decided to turn on the switch to apply VBUS on the device, the same function is quite eminently being done by its control of the TPS55288.

#5 On the other hand, WT6636 has no control if the sink device is higher than 5V upon connetion per clause #2. With TPS55288 connected, this poses no problem whatsoever: WT6636 in fact can detect VBUS condition before setting TPS55288 to output any voltage, yet it can make VBUS be at any sink device pre-existing value by keeping TP55288 OFF. VOUT pin can have any value: 24V at the input keepd upper TPS55288 switch body diode isolated the two. WT6636 cannot: if it was controlling UCC28782.

#6 Upon Hard-reset- which is essentially described as a software-related reset although there are state changes: this change the series pass switch does not help at all in any manner given its body diode. Whatever software indicates the state should be, controlling it by TPS55288 is far more effective including "discharge" because a "0" at VOUT of TPS55288 will discharge the sink capacitor as effectively as the switch. 

#7 Finally, this setup with TPS55288 offers far better and more effective ramping of VBUS than control of the series pass switch which may be destroyed as a linear device in this mode.

SO WHY THIS SWITCH? What specific item in the USB C Spec implies its use?

It is a question of the cost of having TPS55288 and a less efficient module, prone to failures vs. one that is robust by design. Especially if you are going for higher power level applications of 65W and up.

We do require a different programmed WT6636 of course. Or prefer one that we can program ...one from TI as a replacement. 

Using LMG2610 instead of ~20 parts discrete GaN required, is a significant reason to go for our approach: ours are expected not to be fail-prone, repairable..consistent with current trends.

Any comments will be highly appreciated. 

-r

  • Hi Robin,

    Thanks for reaching out with so much detailed descriptions here.

    1. I don't know much more about the MOS, maybe it's used to prevent the pre-stage voltage into VBUS unless the PD allow it. You could ask more about it with UCC28782 team for reference design consideration in a new E2E thread.

    2. For TPS55288, I don't see other customers use like this since this device has no output before you write to Output Enable register.

    3. I have reviewed the schematic you posted in another thread and find something needs to correct, (1) EN/UVLO pin can't connect to VCC since there is no VCC before EN has raises above EN UVLO raising voltage; (2) FB/INT needs to connect to Vout with divider resistors if external feedback is used, and C20 is not needed.

    4. Please tell with the application and input voltage range, output voltage range, max output current.

    Regards,

    Bryce

  • Hello Bryce:

    Thanks a lot for pointing out the issues here.

    In response:

    #1: waiting for some others to comment in the Forum

    #2 We are fully aware of no others designing with this: although the rationale for doing so is put forward in great detail in my "reply" with 5 points. I am hoping others won't use  our scheme!

    In fact, would like to know about a similar TPS part that is as good but only works in buck mode, is there one?

    In fact, our approach is more logical! Here is a causal response to your concern about writing to Enable register of TPS55288. Notice that upon plugging in AC- which is the assumed source here- UCC28782 starts off unconditionally to set output which we redefine to be 20.5V not 26 as shown( in response to Weltrend concerns about OVP- which in reality is a non-issue). And this output is connected to the WT6636 thus starting off even before TPS55288 responds- it need not! WT6636 then goes about running its preprogrammed code.

    There is no contradiction with USB specs up to this point.

    Upon dialog with USB C PD set up BY C1/C2(D+,D-), it decides to first Enable TPS55288, then set VBUS as required, the default setup being Vsafe5 . Or Vsafe0 as the case may be.

    #3: Appreciate your pointing out the error here. Pl see attached corrections .

     WT6636 sets VBUS through I2C per program. do you think TPS55288 still requires FB connection as you indicate here?  Since VBUS could be 5v/9v/12v/20v, what should the divider be for such a case?

    #4 Application is USB C PD or adaptor/charger for any USB C type "sink" per USB C Specs.

      Input to this section is from a UCC28782 module as I mentioned above. With a FIXED OUT OF 20.5 V.

    Max output current from  VBUS to a sink is 5 amp.

    VCBUS thus most efficiently provides 5v, 9V, 12V  ( max AMP per specs), 20V max 5amp per Spec of 100W

    UCC28782 functions most efficiently with a constant output voltage with a fixed efficient transformer. Ours is a planar xmfr- best used in this manner.

    Any further comments will be highly appreciated.WT6636F_TPS55288_230308corr.pdf

  • Hi Robin,

    Please see my comments below,

    #1: waiting for some others to comment in the Forum

    -> You need to post a new thread to ask about this since this thread is assigned to me.

    In fact, would like to know about a similar TPS part that is as good but only works in buck mode, is there one?

    -> Since the output of UCC28782 is 20.5V, I think Buck converter is hard to convert 20Vo from 20Vin since the duty cycle limit, but TPS55288 is suitable for this.And I don't know if there is Buck converter with I2C function.

    #3: Appreciate your pointing out the error here. Pl see attached corrections .

     WT6636 sets VBUS through I2C per program. do you think TPS55288 still requires FB connection as you indicate here?  Since VBUS could be 5v/9v/12v/20v, what should the divider be for such a case?

    -> If internal feedback is used, the divider resistors are not needed and the FB/INT pin can be used as fault indication pin by connecting a resistor to VCC.

    -> About the schematic, (1) please tell with the NMOS part number; (2) Recommend to use the recommended values at MODE pin to select different modes; (3) CDC pin is recommended at PD application for cable voltage droop compensation; (4) Recommend R18=15k, C22=4.7nF, C21=100pF.

    Regards,

    Bryce

  • Bryce:

    Thanks a lot for these detail suggestions.

    Note that the whole set of arguments going back and forth between TI us and Weltrend is about not using the NMOS. If you have TPS55288, you certainly do not need the NMOS.

    There is a document to further see why from our perspective. PL send me your TI email: I will attach that proprietary document for your review.

    Then, the following ensued. 

    Implemented all except CDC.

    TPS55288 is I2C controlled by WT6636.

    ( or preferably by a TI chip)

    So we do not have RFUP . FBK pin is used for fault indication.

    So we have to program the WT6636 or its equivalent to set register 05h to account for it in discrete steps.

  • Hi Robin,

    Please send to my email bryce-xun@ti.com

    TPS55288 is I2C controlled by WT6636.

    ( or preferably by a TI chip)

    You can post a thread to ask if there is a suitable TI PD controller for this, I think there should be some product can be recommended and also reference design.

    So we do not have RFUP . FBK pin is used for fault indication.

    Yes, when FB is used for indication, FB needs to connect to VCC through a 100k resistor.

    Regards,

    Bryce

  • Hi Robin,

     

    We have not seen an update from you for 2 weeks, so I assume the questions are answered and the issue is solved.

    I close this thread now. If there is still something open, please reply and the thread will get opened again.

    If you have any other question or of the thread has been locked, please open a new one.

    Clicking the Resolved Button also helps us to maintain this forum.

     

    Best regards,

    Bryce