JTAG-SMT2-NC with USB 3.1 extension cables

[Eck]

Hello,

I'm using a JTAG-SMT2-NC module on a custom board with a Xilinx ZU9EG as the target.

The USB connector on the board is USB C. The board is mounted in an enclosure with a USB C to C bulkhead connector to gain access to the JTAG-SMT2-NC when the enclosure is closed. A USB 3.1 male to female extension cable (~300mm long) connects the board to the bulkhead connector.

If I connect direct from the board to my PC everything works as you would expect, Vivado connects to the target with no problems. I use a 2m USB C to USB A for the cable.

However, If I connect through the bulkhead connector (through the extension cable) the JTAG-SMT2-NC is not recognised as a valid USB device (error message attached) and Vivado cannot connect to the target. If I connect straight from the board to a USB C host port on my PC i get the same result; JTAG-SMT2-NC only works when using a direct connection to a USB A host port.

So, for now we have to keep the lid off the enclosure to gain access to the JTAG port but this isn't ideal, there are hazardous voltages inside.

Is it possible to operate JTAGE-SMT2-NC through a USB 3.1 extension cable or direct to a USB C host port?

 

Thank you,

 

Alex.

[JColvin]

Hi @Eck,

I apologize for the delay. I reached out to another engineer and was reminded about how USB type C connectors have two sets of D+ and D- pins at the connector in order to allow the cable to be flipped either way when attached to a host connector, though there is only a single pair of D+ and D- pins inside the cable itself.

To make this two sets of D+ and D- on the connectors and only one set in the cable to work, both the device and host side must either short both D+ pins and both D- pins through traces (though this creates stubs, which creates reflections, and can cause signal integrity issues) or the D+ and D- pins must be routed to a mux that's controlled through a CC controller to select which set of pins is connected to the onboard USB controller. Correspondingly, super speed pairs are much more sensitive to signal integrity issues and (more or less) must go through a mux that is controlled by a CC controller.

This is to say that the engineer I reached out to suspects that the host side USB C connector shorted the pairs together using a mux and that the same was done on the device side, creating enough signal integrity issues for the SMT2-NC to not be detected. They also mentioned that extension cables in general (even when USB C is not involved) can easily cause their own signal integrity issues; if the extension cable includes a USB C connector that has stubs due to the designer shorting the pairs instead of using a mux will be much worse.

Thanks,
JColvin

[Eck]

Hi JColvin,

Thanks for your help and taking the time to look into this for me.

On our board D+ pins and D- pins are linked on the PCB as you describe. The stubs created are small but could be causing a problem, however a second USB C connector uses an identical layout with a USB/UART device and it works fine with a 300mm extension inside the unit and a 2m cable outside.

My next step is to make a custom M/F USB C cable so that I can see what happens as the device is connected and identified. I expect JTAG-SMT2 to identify as a full-speed device at 12Mbps. Can you conform that is the case, please?

 

Thanks,

Alex.

[JColvin]

Hi @Eck,

The JTAG SMT2 (NC) should identify as a high speed device at 480 Mbps, though of course it's backwards compatible with full-speed.

Thanks,
JColvin

[Eck]

Hi @JColvin,

OK, thanks for that information. So, just to be clear, does the device identify at 12Mbps if identification at 480Mbps fails, or does identification take place at 480Mbps?

 

Thanks,

 

Alex.

[JColvin]

Hi @Eck,

Yes, that is my understanding; if enumeration fails at high-speed, then it should fall back to identify at low-speed.

Thanks,
JColvin