• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 62 subscribers
  • Views 192 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.

Original question:

TUSB546-DCI: What's the normal behavior of DP AUX chanel when measure AUX eye diagram?

TUSB546-DCI: no screen with AUX_SNOOP enabled and DockingStation Source Device

Torsten Roeder
Prodigy 90 points
Part Number: TUSB546-DCI
Other Parts Discussed in Thread: HD3SS214

Hello all,
we are using the TUSB546-DCI in a 4:1 DP-MUX application (DP-only, no USB-C) and controlling it via I2C. So far we have not had any problems with the selected register settings. A customer of ours would like to operate his DockingStation on our unit but the display on the MUX output remains dark. The docking station has a laptop as source. Laptop and docking station are connected via a USB-C cable. The docking station is connected to our MUX with a standard DisplayPort cable. I see no data on output lanes 0-3 of the TUSB546 and endless traffic on the AUX channel. It looks like a never-ending link training. This happens when AUX_SNOOP_DISABLE is 0. If we set the bit to 1, we have a screen signal. What explanation could there be for this? In another process I read:
AUX snoop is used for power management. The TUSB546 snoops AUX writes to DisplayPort sink's DPCD registers 0x00101 (LANE_COUNT_SET) and 0x00600 (SET_POWER_STATE) and will disable/enable lanes based on value written to LANE_COUNT_SET. The TUSB546 disables all lanes when SET_POWER_STATE is in the D3. Otherwise active lanes will be based on value of LANE_COUNT_SET. If you disable AUX snoop, then you lose this power management feature.

What does "POWER_STATE is in the D3" mean?
So, why does the TUSB546 switch off its outputs even though the source is apparently still trying to perform link training?

  • 0
    TI__Guru*** 139835 points

    Hi,

    Can you send me the TUSB546 schematic, the I2C register dump and link training log file?

    0x00600 (SET_POWER_STATE) is a command sent by a DP source to the DP sink over the AUX bus to go into low power mode (D3), the TUSB546 snoops this and then place itself into the low power mode as well.

    Thanks

    David

  • 0 in reply to David (ASIC) Liu
    Prodigy 90 points

    Hi David,
    Enclosed are the files you requested. As you can see, the TUSB546 is only connected to the AUX channel for sniffing. AUX traffic goes through the HD3SS214 directly to the output jack. There are no problems with other sources, only with this docking station.

    Best regards,

    Torsten

    regdump_tusb546.txt 
    VS0 TUSB546 Reg0xA: 0x12
VS0 TUSB546 Reg0xB: 0x0
VS0 TUSB546 Reg0xC: 0x0
VS0 TUSB546 Reg0xD: 0x0
VS0 TUSB546 Reg0xE: 0x0
VS0 TUSB546 Reg0xF: 0x0
VS0 TUSB546 Reg0x10: 0x44
VS0 TUSB546 Reg0x11: 0x44
VS0 TUSB546 Reg0x12: 0x10
VS0 TUSB546 Reg0x13: 0x0
VS0 TUSB546 Reg0x14: 0x0
VS0 TUSB546 Reg0x15: 0x0
VS0 TUSB546 Reg0x16: 0x0
VS0 TUSB546 Reg0x17: 0x0
VS0 TUSB546 Reg0x18: 0x0
VS0 TUSB546 Reg0x19: 0x0

VS0 TUSB546 Reg0x20: 0x0
VS0 TUSB546 Reg0x21: 0x0
VS0 TUSB546 Reg0x22: 0x4
VS0 TUSB546 Reg0x23: 0x23
VS0 TUSB546 Reg0x24: 0x0
VS0 TUSB546 Reg0x25: 0x0
VS0 TUSB546 Reg0x26: 0x0
VS0 TUSB546 Reg0x27: 0x0
VS0 TUSB546 Reg0x28: 0x0
VS0 TUSB546 Reg0x29: 0x0
VS0 TUSB546 Reg0x2A: 0x0
VS0 TUSB546 Reg0x2B: 0x0
VS0 TUSB546 Reg0x2C: 0x0
VS0 TUSB546 Reg0x2D: 0x0
VS0 TUSB546 Reg0x2E: 0x0
VS0 TUSB546 Reg0x2F: 0x0
    UNIGRAF_DPA400_LinkTraining_Dock-MUX.htmlUNIGRAF_DPA400_LinkTraining_MUX-DISPLAY.htmlTUSB546_schematics.pdf

  • 0 in reply to Torsten Roeder
    TI__Guru*** 139835 points

    Hi,

    Can you also share the log file with MUX AUX_SNOOP disabled? 

    With the MUX sitting between the Docking Station and the Display, you can remove the 100k pullup on AUX and 100k pulldown on AUXP since the pullup/pulldown will be in the Docking Station. 

    The link training file log you sent shows the link training is stuck in the Clock_Recovery phase even with source already providing max SWING of Level 3.

    • Do you have a way to probe the TUSB546 input and verify there is a clock? The clock frequency is the Nyquist frequency of the data rate. So in this case, with 2.7Gbps, clock frequency is around 1.35GHz. 
    • Do you also have a way to measure the current of TUSB546? Looking at the TUSB546 current can tell us if TUSB546 is in low power mode or not.

    If there is a clock present at the TUSB546 input and TUSB546 is in low power mode, then it is possible that the AUX signal coming out of this particular Docking Station, going through two HD3SS214 passive mux, may get distorted so TUSB546 is not snooping the AUX correctly and incorrectly turn off its output.

    Thanks

    David

  • 0 in reply to David (ASIC) Liu
    Prodigy 90 points

    Hello David,
    R100 and R101, the two 100k resistors on the AUX lines are not assembled. Enclosed I send you the log file for the case "AUX_SNOOP_DISABLED = 1".
    The register content of 0x12 is 0x04 with working source and 0x10 with docking station (both with 0x13 = 0x00 and with 0x13 = 0x80 . Bits 6:5 are "00" in both cases. Why is the bit field for SET_POWER_STATE only 2 bits wide? According to the VESA SPEC it is 3 bits.
    Bit 4:0 Lane_COUNT is "00100" with working source (ok, that means 4 lanes) and "10000" with docking station. What does it mean when bit 4 is "1"?
    I also measured the AUX signal from the DockingStation and a working source - also at the output in the DockingStation case. The signal from the working source looks nicer.

    Furthermore, I measured the clock of the LinkTrainings, before and after the HD3SS214.
    Measuring the current consumption of the device would be difficult but possible if absolutely necessary. But the device gets warm when the lanes are switched on, otherwise it stays cold.

    Best regards,

    TorstenUNIGRAF_DPA400_LinkTraining_MUX-DISPLAY_AUXSNOOPDIS.html

  • 0 in reply to Torsten Roeder
    TI__Guru*** 139835 points

    Hi,

    The log file does show link training to be successful from the read of DPCD status register 0x00202 and 0x00203 with the AUX_SNOOPING disabled.

    If it is difficult to measure the TUSB546 power, can you probe the TUSB546 input and output? It does look like the distorted signal on the AUX is causing TUSB546 to not correctly snooping the AUX bus and erroneously turn off its output, which would cause link training to fail. In this particular case, the workaround is to turn off AUX snooping function.  

    Thanks

    David

  • 0 in reply to David (ASIC) Liu
    Prodigy 90 points

    Hello David,
    then switching off the AUX_SNOOP function will probably be the solution. Which features do we have to do without? Is there anything else besides the power saving feature?
    Finally, I measured the currents: With 0x13 = 0x80, the component draws about 191mA from 3.3V, with 0x13 = 0x00 about 1mA.

    Thanks,

    Torsten

  • 0 in reply to Torsten Roeder
    TI__Guru*** 139835 points

    Torsten

    The only impact is the power saving feature when AUX snooping is disabled, there is no other lost functions. And from the power measurement, it does look like the TUSB546 is erroneously turning off its DP lanes.

    Thanks

    David