• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 31 subscribers
  • Views 1901 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TUSB4041I: TUSB4041 stacked and SMBus communication issue

Chris Greg
Prodigy 40 points
Part Number: TUSB4041I
Other Parts Discussed in Thread: TPS62085

Hi there

I'm struggling very similar issue described here.


However, in my case 3 ICs are stacked giving 10 downstream ports at the end (first IC to MCU, one downstream port to upstream port of the second and so on).

SMBus is activated on all chips (with the unique addresses on each, 1k pull-ups). On PCB I can control power supply (enable/disable), where VDD11 if derive from VDD33 (via TPS62085).

The problem is I cannot communicate over SMBus. I've x-ray board to check if an issue is related to the one mentioned and the beginning.

Moreover if do a very short shortcircuits on VDD11 (like 100ms, TPS62085 is fine with this) chip gets unlocked!

In none SMBus mode IC works properly (tested separately, as polarisation on intermediate upstream ports needs to be change - that's why SMBus is in use).

Any idea? 

As this is a prototype I get stucked:(

X-Ray of the chip (see soldering quality of PAD pad)

  • 0
    TI__Expert 7395 points
    Hello Chris,

    Soldering on the pad is definitively a problem. The pad must have an homogeneous solder over its entire area.
    Another thing I am noticing is that you have no vias on the thermal pad as recommended on the datasheet. Check page 43.

    I would also recommend to check your power-up sequence. Check section 7.6 of the datasheet. If your 1.1V power supply is derived from 3.3V, then you must assert GRST (low) for at least 3 ms after both power supplies are active.

    Finally, while on SMBus mode, the USB upstream connection will be deactivated until clearing the CFG_ACTIVE bit at register offset 0xF8.

    Regards,
    Jorge
  • 0 in reply to Jorge Llamas (SWAT)
    Prodigy 40 points
    Hi Jorge

    Thanks for the reply. Problem solved (more below).

    We will keep more attention on soldering quality, especially modifying stencil. The issue is mechanical (check p. 40 for Detail A) as there is a gap between IC and PCB and not sufficient paste injection.

    Power sequence is ok, checked on the oscilloscope. Also, we are using active POR (MCP130-315).

    As this is the prototype. Next evaluation will be enhanced. We will put more vias but we have power lines on middle layers. However, the current design is stable as standalone HUB with high traffic on all 4 downstream ports. PCB is 4 layer with ground everywhere. Thermal performance is ok. Now it is time for more stress tests!

    The problem was... SMBus issue caused by I2C driver in SMT32 host controller. I2C simply stalled! The root cause was timing between power-on (by enabling VDD&VDD11 regulator and pullups powered not from MCU but from the hub. By software, this issue is fully solved.

    Sorry for bothering.

    Chris
  • 0 in reply to Jorge Llamas (SWAT)
    Prodigy 40 points
    Hi Jorge :)

    My original issue with SMBus is solved (it was a timing issue with STM32).

    However, I'd like to ask more about soldering issue (I might have symptoms of it). Recall my case: Host connected to IC1, IC2 to IC1-DN1, then IC3 to IC2-DN1 (cascade, level 3).

    I assume that lack of solder is a thermal problem. No doubt here. May it cause interoperability issue?

    In my case, IC1 talks with the host but not with IC2 (and farther with IC3, enumeration fail on IC1-DN1). But IC2 talks with host and IC3 is working too (if I skip IC1). As host works PC with USBlyzer.

    Note that DN1 is connected with upstream port via port polarity swap (to simplify PCB).

    What is your opinion?
  • 0 in reply to Chris Greg
    TI__Expert 7395 points
    Hello Chris,

    Well, the problem with bad soldering on the pad is beyond a purely thermal issue. The pad is the only ground terminal on the device; so the TUSB4041 will malfunction if proper grounding isn't implemented.

    Can you confirm you are setting "p0_usb2pol" at the "USB 2.0 Port Polarity Control Register" either trhough an external EEPROM or SMBus master?

    Regards,
    Jorge
  • 0 in reply to Jorge Llamas (SWAT)
    Prodigy 40 points
    Hi Jorge,

    I have spotted that thermal pad is the only ground terminal. However, I still cannot believe soldering can have such an impact :)

    About controlling the chip over SMBus - I can confirm, all bits are properly set (by readback of the registers state).

    Note that IC2 is talking to IC3 where the connection is identical. I have even cut a trace to disable any electrical connection and solder wires to debug this over PC. Moreover, all combinations were tested (with and without changing polarity in chip - to remove the possibility that something is wrong there).

    Tomorrow this chip will be replaced by new one. Will see what happens.

    Any other idea rather than IC malfunction?

    Cheers,
    Chris
  • 0 in reply to Chris Greg
    TI__Expert 7395 points
    Chris,

    Can you please provide me the sequence of data being written by your SMBus master to the different TUSB4041 devices in order to configure them?

    Thanks,
    Jorge.