AM5748: USB3.0 Loopback mode does not work

Hello Yuichi,

Can you please confirm that you are using the ist instance of USB? (USB1)

Adding this thread in case the information helps..

• https://e2e.ti.com/support/processors-group/processors/f/processors-forum/567349/am5728-usb-compliance-test-patterns/2230318?tisearch=e2e-sitesearch&keymatch=thread%3A567349#2230318
• Here is a copy of user-guide that used to be available publicly: Linux Core DWC3 User's Guide - Texas Instruments Wiki.pdf

-Josue

Hi Josue,
Thank you for the information.

Josue Zamitiz-Ayala said:

Can you please confirm that you are using the ist instance of USB? (USB1)

We are using USB1 and can communicate at SuperSpeed, but the compliance test (Receiver Jitter Tolerance Test) has failed.
As for transmitter testing, our device passes without any configuration.

I read the links.
Regarding USB3.0, it seems that no settings are required for operation in test mode.
Is it correct that operation in Loopback mode does not require any settings as in Compliance mode?

Hello Yuichi,

I will be out of the office until 11/28 due to holiday.

I appreciate your patience.

-Josue

Hello Yuichi,

The USB expert is OoO until 12/11.
I apologize for the delay and really appreciate your patience.

Unfortunately the information publicly available is the only help we have here at the moment.

Regards,

Josue

Hello Josue,

Thank you for reporting the situation.
We are looking forward to your response as we have not been able to improve the situation.

Regards,
Yuichi

Yuichi,

My pleasure.

-Josue

Hello Yuichi,

I don't believe that anything needs to be setup from the software side to enable loopback mode. This is something that is done by the test device. I will let our HW USB engineer take it from here.

-Josue

Yiuchi,

So the TX tests are passing ( LFPS and CP0, CP1 included ). Did you have to put the device in loopback mode to get the TX to pass compliance?

How were the CP0 and CP1 patterns sent out from the DUT to the scope?

Also, for the JTOL, I would think you need a BERT which will send the USB patterns into the RX of the DUT which is sent out through the TX of the DUT  connected back to the BERT for allowing the comparison and measurement of jitter tolerance.

Hello Shreyas,

So the TX tests are passing ( LFPS and CP0, CP1 included ). Did you have to put the device in loopback mode to get the TX to pass compliance?

No software processing was performed during the TX test.
As for during TX testing, I don't think loopback mode is needed.

How were the CP0 and CP1 patterns sent out from the DUT to the scope?

I don't know the actual pattern, but I received the following report from a testing agency.

TD.1.9 Receiver Jitter Tolerance Test at 5 GT/s (Type-C): Cannot Test
Timeout occurs in sequence (Check Detector Sync) after switching to Loop Back mode.

We have confirmed that the DUT transitions to Loopback mode.
However, although the pattern coming back from the DUT complies with the 8B/10B encoding rules, it is completely different from the pattern expected by the test tool.

As I know for the RXJTOL testing, the device must be in the loopback mode such that it receives the pattern coming from the BERT into its RX pins, loops it back and send it out through the TX pins back to the BERT which then compares it to the sent pattern to determine its BER.

If the tool is not seeing the right pattern back, then likely that the pattern sent out from the BERT / tool needs changes. It also might have reference clock sync issue, may need to ensure that the same ref clk from the device is used for sync with the tool for pattern generation and for receiver.

Yuichi,

The AM5748 device supports and is compliant to USB3.0 but you are trying to validate USB3.2 compliance test.

the USB3.0 would switch to compliance mode once there is no LFPS ping received and the termination is seen. but maybe that the scope is still sending the LFPS pings, not sure if the device is entering the complince mode.

Hello Shreyas,

The test agency tried other device instead of ours. In that case, the exam could start.

> The AM5748 device supports and is compliant to USB3.0 but you are trying to validate USB3.2 compliance test.
>
> the USB3.0 would switch to compliance mode once there is no LFPS ping received and the termination is seen. but maybe that the scope is still sending the LFPS pings, not sure if the device is entering the complince mode.

Does that mean that AM5748 will not pass the USB3.2 compliance test?
I thought USB 3.0 was equivalent to USB 3.2 Gen 1x1.

Yuichi,

For the jitter tolerance test, I put the device in the PCS loopback mode ( for TDA4xx devices).

The BERT would send the pattern and device would return back the same pattern back to the BERT which it compares and determines the BER.

could the sent pattern (from the bERT) be captured on the scope and compared against the device TX output?

if the device is in loopback mode as expected, then the RX signals should loopback to TX.