TUSB1046A-DCI: Question about design requirements

Hi,

TPS65982 is not recommended for new design, please use TPS65987D, https://www.ti.com/product/TPS65987D.

But you do need the PD controller such as the TPS65987D to support USB and DP Alt Mode over USB Type-C, please see this doc for a general overview, https://www.ti.com/lit/wp/slyy109b/slyy109b.pdf.

Thanks

David

Thank you for your reply. 

I can't seem to figure out why my application is not working. 

My design consists of a board that combines DP and USB 3.0 using TUSB1046A-DCI. Then with a USB-C cable I connect to a secondary board that uses USB1064 to demultiplex the signals. The secondary board consists of a USB 3.0 Hub and a DP to MIPI converter. Both redriving switches are operating in GPIO mode and SNOOP is disabled. I am able to get the USB SS signals across, but I am not having luck with the DP. 

I found two EVM reference designs for each module, but I realized some of the configurations are different than what the datasheets recommend. 

TUSB1046-DCI

1. Do CONFIG1 and CONFIG2 resistors need to be populated? if so, which ones? Reference design has them as DNP

2. Do AUXn and AUXp resistors need to be populated? If so, which ones? Reference design has them as DNP

3. Do AUXn and AUXp need coupling capacitors? Reference design does not seem to include them.

4. Do SBU1 and SBU2 need 2M pull down resistor. Reference design does not include them.

5. Should snoop be enabled or disabled? 

TUSB1064

1. Do SBU1 and SBU2 need 2M pull down resistor if they are already used on the other board?

2. Do AUXp and AUXn resistors need to be populated? If so, which ones? Reference design has them as DNP

3. Do AUXn and AUXp need coupling capacitor? Reference design does not seem to include them

Thanks in advance. 

Hi,

1. Do SBU1 and SBU2 need 2M pull down resistor if they are already used on the other board?

Yes, they need to be present on TUSB1046 and TUSB1064 side. The 2M on SBU1 and SBU2 are representative of the leakage of ESD and EMI/RFI components, including termination to ensure that there are no floating nodes, and are intended to show compliance with zSBUTermination in USB-C.

2. Should snoop be enabled or disabled?

Snoop is only used for TUSB1046 and TUSB1064 power management. If power consumption is not a concern, I will leave it disabled for now. 

3. Do AUXp and AUXn resistors and capacitors need to be populated? If so, which ones? Reference design has them as DNP

If the DP source is designed on the same board as TUSB1046, then both AC coupling capacitors and resistors are required. AUXP should be pulled to ground through a 100k resistor. AUXN should be pulled to 3.3V through a 100k resistor.

For the reference design, TUSB1046 is being connected to an outside source. The AC coupling capacitors and resistors are already populated inside the outside source so you will not populate them on the reference design.

If the DP sink is designed on the same board as TUSB1064, then both AC coupling capacitors and resistors are required. AUXP should be pulled to 3.3V through a 1M resistor. AUXN should be pulled to ground through a 1M resistor.

For the reference design, TUSB1064 is being connected to an outside sink. The AC coupling capacitors and resistors are already populated inside the outside sink so you will not populate them on the reference design.

4. Do CONFIG1 and CONFIG2 resistors need to be populated? if so, which ones? Reference design has them as DNP

Are you connecting to a DP source? The DP source should have 1M pulldown on CONFIG1 and CONFIG2. In this case, you will leave TUSB1046 CONFIG1 and 2 to be DNP.

A. Do you have a PD analyzer that can decode the CC traffic between the PD controller in the source side and the sink side? This will make sure PD controllers are entering into the DP Alt Mode. You can also probe FLIP pin, it should go high or low depending on the Type-C cable orientation. CTL0 pin which should be low for 4 DP lane mode or high or 2 DP lane mode, CTL1 must be high to enable DP mode.

B. Can you also probe the voltage on AUXP and AUXN? AUXP voltage needs to be around 0.3V, and AUXN voltage needs to be around 3V for SBU/AUX switch to work properly

C. HPD pin, is it going high as well?

D. Can you share your schematic for review?

Thanks

David

1. Okay current designs have 2M pull down resistors on SBU1 and SBU2 on both devices. 

2. No need for power management so I will keep it disabled. 

3. TUSB1046 uses an external DP source. The current design has AC coupling caps and pull up/down resistors populated on AUX lines, but it sounds like these are not required so I will remove them.

TUSB1064 uses a built in DP to MIPI converter so it seems like I need to add the corresponding pull up/down resistors and caps on the AUX lines.

4. I am using external DP source for TUSB1046 so I will remove the CONFIG pull down resistors 

I do not have a PD analyzer, but I can probably get one. Do you have one you recommend? 

Flip does change HIGH/LOW depending on the USB-C cable orientation. 

I am using jumper headers to set CTL0 LOW and CTL1 HIGH to use 4DP lane mode for testing. 

I will probe AUX signals and report back soon. 

On TUSB1046 I am pulling HPD high with a 1K resistor. On TUSB1064, the DP to MIPI IC drives the HPDIN high when it is initialized.

Attached you can find the schematics for each module (Without the suggested changes mentioned above)

DISPLAY_PORT_USB_3.0_TO_USB_C_23thNOV2022.pdf

AR 2.0 Screens Schematics.PDF.pdf

This is very helpful information Thank you!

Hi,

For the DISPLAY_PORT_USB_3.0_TO_USB_C_23thNOV2022 schematic, 

1. You can replace C5,6,7,8,11,12,16,17 with 0ohm resistor since AC coupling capacitors on already populated inside the DP source. Having two sets of capacitors reduce the total capacitance, may cause DC wander.

2. Please change C32 and C35 to 330nF or 407nF. The AC coupling capacitors should already populated on the SSTX on the USB source. Having two sets of capacitors reduce the total capacitance, may cause DC wander.

3. Add a 10uF cap to 3.3V supply, this will help filtering low frequency power supply noise

For the AR 2.0 Screens Schematics,

1. Add a 10uF cap to 3.3V supply, this will help filtering low frequency power supply noise

2. Please change C47 and C48 to 470nF or 330nF

3. Please change C49 and C62 to 100nF or 220nF

None of these recommended changes impacts the no display issue, but more on improving the signal integrity.

For the analyzer, I am aware three of them.

Total phase, https://www.totalphase.com/products/usb-power-delivery-analyzer/, can capture CC communication only, not DP AUX link training

Ellisys, https://www.ellisys.com/products/ctr1/, can capture CC communication and DP AUX link training

LeCroy, https://teledynelecroy.com/protocolanalyzer/usb/voyager-m4x, can capture CC communication and DP AUX link training

I would personally go with the one that can support both CC and DP AUX Link training analyze in case we need to start to look at the AUX data down the road.

Thanks

David

I ordered a PD analyzer, so I am just waiting for it to arrive. In the meantime, I made the changes you suggested. 

For #2 I updated the values for C32 and C35 and shorted C3 and C4, but the USB 3.0 configuration stopped working so I moved the caps to C3 and C4, mimicking the dev board. 

After this USB 3.0 worked again. But still no luck with DP configuration. It seems like the AC coupling caps placement is pretty crucial. 

I measured AUXp and AUXn voltages. On the MUX side (DISPLAY_PORT_USB_3.0_TO_USB_C_23thNOV2022) I get 0.254V for AUXp and 2.7V for AUXn. On the DEMUX side (AR 2.0 Screens) I get 0.254V for AUXp and 2.7V for AUXn

On the Demux side I added the 1M resistors and coupling caps to the DP sink in this configuration, as suggested by the datasheet 

However, I am still not able to get DP to work ...

Hi,

You do not want to short C3 and C4, you can use 110nF or 220nF for C3 and C4.

Did you remove the AC coupling caps and pullup/pulldown on AUXP and AUXN on the DISPLAY_PORT_USB_3.0_TO_USB_C_23thNOV2022? The common mode voltage does meet the TUSB1046 and TUSB1064 requirement as shown below.

TUSB1046

TUSB1064

Is snooping being disable on TUSB1046 side?

While waiting for the PD analyzer, does the DP to MIPI IC have a set of registers you can read to indicate DP link status?

Thanks

David

Got it I think I misunderstood the last message. So, I have C3 and C4 with 100nF and C32 and C35 with 470nF on TUSB1046. 

AC coupling caps and resistors have been removed on the AUX lines on DISPLAY_PORT_USB_3.0_TO_USB_C_23thNOV2022.

Snoop is disabled on TUSB1046 (pin 29 pulled high to 3.3v).

Yeah there are some registers that are not giving the expected value (failing) for Link training.

Hi,

If the link training is successful, and if you read DP to MIPI IC DPCD register 0x00202h (for lane 0 and 1) and 0x00203h (for lane 2 and 3), you will see a value of 0x77 for both registers.

The DP link training consists two parts, clock recovery and equalization training. 

For clock recovery, do you have a way to read the DPCD register 0x00202h and 0x00203h on the DP to MIPI IC bit 0 and bit 4 (CR_DONE) are set? If they are, then this implies the main link is connected between the source and the sink. If not, then we need to check the main link connection or lane mapping.

If CR_DONE bit is set, and LANE0_CHANNEL_EQ_DONE and LANE0_SYMBOL_LOCKED (bit 6, 5, 2, 1) are not set, then we will need to tune the re-driver DPEQ setting.

Thanks

David

The two registers that are failing are 0x00202h and 0x00204h. I get 0x00 instead of 0x77 for the first one, and 0x80 instead of 0x81 or 0x01 for the second one.

I received PD analyzer and here is what the output looks like when I connect the device. Not much seems to be happening...

Is it possible to get TUSB1064RNQ EVM and TUSB1046 USB Type-C Enabler EVM directly from you? There seem to be out of stock everywhere.

Hi,

Can you reach out to our local FAE and ask them to submit a request for EVM? Meanwhile I am asking the PD controller team to look at this issue to understand why there is no PD communication.

Thanks

David

Could it be that it does not show activity because I don't have PD controller on the boards?

Hi,

Yes, it would make sense since you do not have a PD controller, you need to have a PD controller on both the TUSB1046 and the TUSB1064 board. 

For both board, do you have FLIP tied low, CTL0 tied low, and CTL1 tied high? 

Is the AUX/SBU common mode voltage now meeting the TUSB1046 and TUSB1064 requirement as listed in my previous response? 

I assume HPD is also high on both boards?

Do you have a way to probe the TUSB1046 DP output and verify there is a signal? 

Thanks

David

I am mainly testing these configurations, making sure both boards are configured the same.  

I noticed on TUSB1064 AUXn is 2.69V in 4 Lane DP configuration and 2.68V in 2 Lane DP configuration, so slightly below the threshold.

AUXn read 2.7V on TUSB1046 in either configuration.

AUXp seems fine on both sides in either configuration (~250mV)

Yes, HPD is high on both sides.

Unfortunately, I currently don't have a way to probe DP output. 

I ordered both EVBs so hopefully they can provide more insight. 

Sorry, I meant to highlight these configurations:

Here is a diagram of the complete system. Basically yes, TUSB1064 input is connected to a Type-C connector through a mezzanine connector. There are 3 boards in total. AR 2.0 Screens mates on top of USB-C Hub. And TUSB1064 output connects to DP-to-MPI directly without connector on DISPLAY_PORT_USB_3.0_TO_USB_C

Hi,

So TUSB1064 basically splits the incoming Type-C signal into USB and DP, with USB signal going to the USB3.0 hub and DP signal going to the DP to MIPI IC?

We can wait for the EVM or if you can probe the DP output, it will be nice to set one lane to generate TP1 or clock pattern, and then use a scope to verify TUSB1064 is correctly outputting the clock pattern.

Thanks

David

Hi, 

I finally received the EVBs. I am waiting on the USB-A to USB-B Cable for TUSB1046. In the meantime, I started testing my version of that board with TUSB1064. Here is how I have the EVB configured: 

The billboard gets recognized on my computer, and USB 3.0 configuration seems to work. However, I still can't get 4 lane DP to work. I do get a message saying: "Display connection mat be limited". I am hoping to receive the cable tomorrow so I can test both EVBs. 

Hi,

When you are testing both EVB, can you also capture the PD log using the analyzer?

Thanks
David

I started testing the EVBs today. Here is how I have TUSB1046 setup for 4 lane DP. Attached you can find the data captured from PD analyzer

4LaneDP.csv

Still no luck getting DP to work.. 

Did you play with DPEQ setting?

Hi, 

I was able to get the EVBs to work! It turns out the TUSB1064 board had the wrong marking for I2C_EN. So, in order to successfully disable I2C I needed to use the jumper pins on the left side instead. 

Another thing I noticed is that on TUSB1046, when configuring CTL1 to HIGH, it was actually not reading any voltage. So, I ended up bypassing R19 to ensure the pin was pulled HIGH.

I have yet to get my own board to work... 

Do you know if SBU lines to be routed as diff pairs? 

Good to know EVM is working.

SBU lines don't need to be routed as differential pairs.

Regards

Brian

Hi,

Please route both SBU and AUX as differential pairs.

Thanks

David

So, I tweaked the boards a little and was able to implement the PD controller from EVB TUSB1064 in my Screens Board. When paired with EVB TUSB1046 I was able to get this (display 2 being the one of interest.)

Even though the MIPI screens are still not turning on and Display 2 shows greyed out it feels like a step forward.

Would not having SBU routed as diff pairs cause DP to not work? 

Also, I noticed I miscalculated the length match for RX2, making it 96 mils longer than the rest of the pairs. Would this also cause DP to not work?

For my next PCB rev I am planning on adding PD controller. You mentioned earlier TPS65982 was not recommended for new designs. Why is that?  What about TPS65986 (I can't seem to find it available anywhere)

Hi,

Even though RX2 is 96mil longer than the rest of pairs, I do not think this is the issue here since a DP sink can tolerate a large inter-pair skew.

If you click on Display 2 and go to advanced setting, are you seeing Display 2 showing the right vendor name? 

For the EVB TUSB1046 and your board, what is the DPEQ setting?

Thanks

David

Display 2 shows the right vendor name. 

I tried these DPEQ setting combinations and they all gave me the same results:

EVB (dB) | My board (dB)

0   | 0 

5   | 0 

10 | 0 

15 | 0 

0   | 5 

5   | 5 

10 | 5 

15 | 5 

0   | 10 

5   | 10 

10 | 10 

15 | 10 

0   | 15 

5   | 15

10 | 15

15 | 15

On my USB-C Hub board I noticed SBU lines are routed as diff pair but without length match 

And on my Screens board they aren't routed as diff pairs 

The AUX lines, however, are routed as diff pairs:

Could this be root of the problem?