we are developing a custom carrier board for the p3668 SOM based on the p3509-0000 Carrier-Board.
The Design is following the Guidelines found at the Nvidia Download Portal and only has a few modifications regarding:
USB1signal pairs → dedicated USB 2.0 Type A Adapter (no hub)
USB2signal pairs + UPHY5 → dedicated USB 3 Type A Adapter (no hub)
- the HDMI Display Out is replaced by a second DP-Out and should be used as the primary boot up display
While the Carrier Board Schematics are structured very clear and follow consistent Pin-Namings, the Platform Adaptation and Bring-Up Guide seems to follow some totally different notation, which are very confusing:
- According to the “Go through the Schematics” Chapter,
USBSS_*Pins are connected to the USB3.1 Hub on the
p3509Board. In Reference Schematics, these are referenced as
UPHY5, whereas in the Guide, the conclusion is that:
- The USB2 signal pair is wired to `UTMI pad 1` (USB2 port 1). - The USB 3 signal pairs are wired to `UPHY lane 1` (USB 3.1 port 2).
While the first statement seems rather logical (despite it is not clear what here the “pad” should be), the second statement speaks of
lane 1 instead of
UPHY4, and is enumerated as a “second” port (the
NX only has one per default).
- Earlier in the Document, the sentence
UPHYlanes are shared between
PCIE, SATA, UFS, and
XUSB, UPHYlanes must be assigned according to the custom carrier board’s requirements.
leads to the expectation, that a
UPHY assignment could have been made, which is different to the notation in the
Schematics Ref.. But in contrast to the Guides for other Jetson Boards (e.g.
TX2), the section explicitly lacks a
description of this configuration.
The name of each parameter description subnode in pads and ports must be in the form
<type>is “usb2” or “usb3”, and
<port_number>is the associated port number.
…but in the example, there is a node
pads/usb3/lanes, but a node
USB3-0 under ports which is referencing to usb2 signals on
Port 1 (probably this doesn’t mean the same in this and the aforementioned contexts)!
- The DT-Option is described by
nvidia,usb2-companion: USB2 port (0, 1, or 2) to which the port is mapped.
So which “port” is mapped here to which “port”? Why first signal names with leading incrementing numbers, and now only numbers?
- Additionally, according to the
Note: As the Realtek hub is always connected to the root hub port on a
P3509, you need not control hub power, just enable it with
VDD_3V3_SYS. Therefore, you must set dummy regulators for
…it should exists some dummy regulator reference in the example, but no
nvidia,oc-pin Options can be found in none of the nodes. Because we dont have a hub which is responsive on its behalf for these settings, we connect directly to an external
Type A Connector, so we would probably need these settings properly configured (no dummy).
Nevertheless, no matter if we put a dummy or nothing, the USB3 Connector has no connection. This is indepently from the naming, port numbering or
VBUS/OC settings we tried: all leads to the same result!
cfg-example-tegra194-p3509-disp.dtsi (3.1 KB)
For the configuration of the display outs, there is nearly no documentation targeting a dynamic assignment. A quick copy, paste/replace of the existing
HDMI-Node in the DeviceTree /w the slightly adapted DP-Port Node leads to a dark Display at boot-up. Surprisingly, there exists a whole chapter of extracting some
X11 values and transformations to
.dts code, but none of the used Notations (“Windows”,“Head”,…) is explained, nor the reconfiguration to use custom display outs /w
HDMI-Only is mentioned with a single word.
tegra194-p3509-disp.dtsi (3.0 KB)
Am I reading the wrong Docs, or could you please give me an insight which could perhaps lead me to a better understanding of these adaptation tasks?