TUSB216: Entering Shutdown at Startup

Hi Ethan,

The "downstream" TUSB216 is connected to a downstream port of a USB Hub correct? Do you have any waveforms of CD and ENA_HS with DP and DM on the "downstream" TUSB216 when this issue is seen? Are you seeing any connectivity issues during this time?

Hello Malik,

Thank you for getting back to me on this. Yes, the downstream TUSB216 is connected to the downstream port on a USB Hub IC located on PCB E. Here are waveforms for when this TUSB216 comes up with Boost 3:

Ch1: VCC

Ch2: CD

Ch3: ENA_HS

Ch4: DP/DM differential voltage at the downstream connection to PCB E

Here is what is looks like when the TUSB216 enters shutdown on power up:

Another shot of the CD and DP/DM waveform when it enters shutdown:

Typically I do not see connectivity issues when the TUSB216 enters shutdown. Even in shutdown the signal quality is marginally passing for this 4m section of cable. It is possible I have seen a reset occur several seconds after coming up at boost 3. But this is not easily repeatable.

Here are far end eye diagrams for this "downstream" 4m section of cable when this TUSB216 is in shutdown:

PCBE side:

PCBD side:

Here are far end eye diagrams for this "downstream" 4m section of cable when this TUSB216 is in boost 3 mode:

PCBE side:

PCBD side:

I lowered the boost to 2 and I am still seeing this "downstream" TUSB216 sometimes enter shutdown and sometimes enter boost 2.

Thanks, Ethan

Hi Ethan,

This seems to be a intermittent issue with USB 2 enumeration. TUSB216 relies on USB 2 enumeration to enter high speed mode and enable boosting. If CD goes high then low and ENA_HS remains low this can signify that USB 2 enumeration was not recognized by TUSB216. Can you zoom in (~2ms/div) on the below portion of the "shutdown" waveform? I would like to see the waveform as measured on PCBE and DP/DM single ended if possible. Does your USB Hub have a function to change USB 2 drive strength?

Hello Malik,

The USB Hub on PCBE does not have increased drive strength ability. It is the CY7C65634 part. The Hub on PCBD is boosted at max level. It is the USB2517 part. Below are the single ended waveforms for DP and DM on the downstream PCBE connection.

Shutdown DP:

Shutdown DM:

For reference, here are the waveforms when the TUSB216 enters boost mode.

Boost DP:

Boost DM:

Let me know if the waveforms do not appear as expected. I have not used the single-ended mode with this differential probe before today. 

Thank you,

Ethan

Hello Malik,

After you wrote that it may be an enumeration issue with USB 2.0 prior to boosting being engaged, I dropped the TUSB216 boost setting back down to 0. I confirmed that even with a boost setting of 0, the downstream TUSB216 will still enter shutdown on startup. It seems that it is less often than when the boost is at 3, but that may just be a coincidence.

Thank you,

Ethan

Hi Ethan,

I see, this points to something in the enumeration process separate from the boost level. This makes sense because boost is not enabled until after HS mode is entered ( ENA_HS is high). I would like to see a zoomed in waveform to provide more insight.  

Hi Malik,

Thank you. Can you specify what waveforms you would like to see? What should I be looking for?

Thank you,

Ethan

Hi Ethan,

Sorry, you already provided the waveform I was looking for. Can you reduce the boosting on the USB2517 part to its normal settings?  

Hello Malik,

Sounds good. Thank you very much for the help.

Ethan

Hi Ethan,

Did reducing the Hub boosting help here? 

Hello Malik,

Whoops. I missed your request to lower the boost setting pm the USB2517. I can try that, probably early next week. I will get back to you with that data. Did you see anything suspicious in the USB waveforms or any indication of why the TUSB216 is not recognizing enumeration? Is there anything else I can try beyond the USB2517 boost setting change?

Also, do all TUSB21x parts work this way where they need to digitally recognize USB enumeration before configuring their appropriate boost? If there were a part that would just boost without needing this additional step, that would be beneficial in this case. 

Thank you,

Ethan

Hi Ethan,

Sorry left out the detail. It seems possible that the JK chrip is slightly too large in some cases and fails our check for USB 2 enumeration. I suspect that with increased drive strength from the USB Hub makes this performance intermittent. Removing or decreasing the hub boosting should help. Since TI USB 2 redrivers are designed to be transparent to the USB link this feature was added to prevent interference with the enumeration process and tell the difference between USB FS/LS and USB HS data. If boosting is left on at all time it could cause other issues with USB enumeration on the Host/device side.      

Hello Malik,

Thank you very much for explaining some more detail here. I retested this week and decreased the boost for the USB2517 Hub on PCA D to 0% extra drive strength. It appears the TUSB216 comes up and enters HS boost mode more often but still fails maybe 1/5 to 1/10 times. I've taken some more measurements on the J and K chirps. 

USB2517 12% boost, TUSB216 comes up:

USB2517 12% boost, TUSB216 does not come up:

USB2517 0% boost, TUSB216 comes up:

USB2517 0% boost, TUSB216 does not come up:

I can see a difference for the J and K chirp amplitudes when the boost was decreased on the USB2517 hub. The downstream "device" K chirp went from -825mV to -800mV and the upstream Hub J and K chirps went from +825mV and -700mV to +787mV and -756mV, respectively. But I am still seeing the TUSB216 not coming up. 

I reviewed the USB 2.0 standard and found these numbers for J and K chirp voltage limits.

The waveforms captured don't appear to be violating any of these limits. What limits does the TUSB216 have internally? Are these limits tighter?

When the TUSB216 CD line goes low, it will do so during the upstream Hub J and K chirps. Does this indicate it is the upstream Hub J and K chirps that are exceeding the TUSB216 limit versus the downstream Hub K chirp?

Do we have a proximity issue where the TUSB216 is too close physically to the Hub on PCA E?

Here is a partial layout snapshot of PCA E. The board is only about 1.5 in. long.

Thank you again for your help. I look forward to your response.

Ethan Rhodes

Hi Ethan.

Thanks for the waveforms, the JK chirps are as expected. Since PCB E is small and TUSB216 is very close to the downstream port of the hub there could a proximity issue here. Would you be able to try shorter cables to PCB C and see if the rate of failure decreases further? 

Hello Malik,

Thank you for the information. We are committed to the cable length we have so I don't want to shorten it as a part of the solution. Is there anything else to try? My design intent was to depend on the TUSB216 boosting ability to provide additional signal quality margin at the far end. With the current configuration, I can't depend on it coming up.

One other thought I had was to remove the USB Hub altogether from PCA E and just route the signals through 1 or 2 TUSB216 parts on PCA E. Would this potentially improve the reliability of the part to enter boost mode? Would you suggest 1 or 2 TUSB216 parts on PCA E? Would the 2 parts be too close if they both were in line?

Thank you,

Ethan

Hi Ethan,

I believe it would help and that TUSB216 would be able to redrive 8 m of cable on PCBE. I would say only 1 TUSB216 should be used with no Hub. As you mentioned 2 TUSB216 would be too close.  

Hi Ethan,

Any more support needed here?

Hello Malik,

No, I think we are good to close this thread. I will try the single TUSB216 on PCBE and go from there. Thank you very much for help and information.

Ethan