Jetson Nano/NX USB3.2 (LANE2) Speed Reduction issue

Hi Sirs,

Current, my project design need support three USB3.0 (2x external IO + 1x Recovery) requirement, here is the port lane as lists :
(a) USB0 + USBSS1 => Recovery mode purpose
(b) USB1 + USBSS0 => External I/O
(c) USB2 + USBSS2 => External I/O

These USB3 design have follow up the reference Type-A connector example as snapshot, for “Orin NX Series and Nano Series_Design Guide_DG-10931-001_v1.1.pdf” (Figure. 7-2)

But, after internal development USB3 signal interface, have mention “USB2+USBSS2” testing meet Speed Reduction condition occur and offer test read/write result.

After check the layout routing, confirm both are routing in the same length & same electrical designs,
[ “USB1+USBSS0” + “USB2 +USBSS2” ]

Therefore, does you have any idea on this one for meet speed reduction occur? Many tks!!

BRs
By Neil Chen

update the Fig 7-2 design example (for above mention follow up the design guide)

Have you made the software configuration of device tree?

Hi Wayne,

Attached file is our device setting, please kindly review and welcome if you have any comments on here (setting), many tks!!
tegra234-pb-c1nv-zp-000.zip (1.1 KB)

BRs
By Neil

please convert your final dtb back to dts and attach that dts.

HI Wayne,

Attached file is our device setting and follow-up your order, we convert to dts. format done.
Please kindly to review it and welcome if you have any feedback on here, many tks!!
0612.zip (36.2 KB)

BRs
By Neil

What is the exact error you hit here?

Hi Wayne,

Current, my project have plan to support two external I/O connector which USB3.0 source is come from Nvidia Jetson module, and here is the design configure as lists :
(a) I/O-(1) : use USB1 + USBSS0
(b) I/O-(2) : use USB2 + USBSS2

The internal development, discover the I/O-(2) meet write (TX) speed reduction condition occur as snapshot “red mark” side. In this picture, you can see the I/O-(1) testing result is pass and meet design expect result.

But, I/O-(2) testing result : read(RX) speed is normal but write(TX) speed is reduction, after compare I/O-(1).

For Hardware part : confirm both I/O design routing length is almost the same and electrical schematic have follow-up the “Nvidia Jetson Design Guide_DG-10931-001_v1.1.pdf/Fig. 7-2” as mention above message.

For software part :
(1) Does our development configure of device tree is correct? (base on provide file 0612.zip (in dts. format)
(2) Does software have anything can check or adjustable via register setting?
(3) In read normal & write abnormal case, does you have any debug idea on here?

Above three question, please kindly to reply and hopeful have deep dive discuss on next, tks a lot!

BRs
By Neil

I don’t actually know what does that mean from your picture.

Could you just check the device on your USB2+USBSS2 status in lsusb command? Is it in USB2 speed or USB3 speed?

If you are talking about one port has lower speed, then you should check lsusb first.

Hi Wayne,

The two external I/O coneector can get the correct USB device via lsusb command in these two I/O-(1) & I/O-(2) signal lanes.

In my picture, we adopt the “dd” for testing USB read & write speed in linux env., base on the test picture can discover I/O-(2) : USB2+USBSS2 configruation, write (TX) speed is very slow as red mark highlight and read (RX) speed is normal.

Therefore, these questions want to asking :
(1) Does our development configure of device tree is correct? (base on provide file 0612.zip (in dts. format)
(2) Does software have anything can check or adjustable via register setting?
(3) In read normal & write abnormal case, does you have any debug idea on here?

We need to clarify the software configuration/register can adjustable? Or, development configure of device tree is setting correct?!

BRs
By Neil

Hi,

Just to clarify. We didn’t answer your questions because they do not really matter.

If lsusb can give you the correct speed. For example, both of the device there is in USB3 speed but not USB2, then it has nothing do with with device tree.

You didn’t share anything useful here but just keep repeating same things, so I don’t see anything I can check here.

The way you should debug here is put your device to NV devkit and see if same issue would happen in NV devkit too.

If that would happen too, then we may need to look into it. If it will not, then maybe check our Orin AGX tuning guide and see if any tuning thing you could do.