Just curious if anyone knows the reason why a USB-A expansion card draws more power in the USB 4 ports on the AMD Framework laptops. I’ve seen it mentioned a few places but never with a reason.
My guess work:
In order to be fully functional usb-a ports, they need to supply a default voltage of 5V. USB-C ports usually do not supply much of a voltage (usually below 1V) until either some line gets pulled down (or was it up?) with a resistor to trigger the 5V behavior or communicate with the power supply to deliver another voltage.
I suspect these thunderbolt/USB4 ports become inefficient when asked to continually supply 5v or even have no means to conserve power with no device connected. Likely a limitation of the controller or something like that?
1 Like
From Framework support pre-shipping updates:
We learned recently that the Kandou retimers don’t enter a suspend state with an empty USB-A Expansion Card or with a USB 2.0 device plugged in, resulting in high power draw. This also occurs with HDMI (3rd Gen) and DisplayPort (2nd Gen) Expansion Cards, but we are currently working through a firmware solution for this.
1 Like
I’ve seen a lot of this on my Framework as well, my current solution is to have the USB-A cards on hand and not keep them in the laptop. Here is a handy graph if you ever need to know about the ports:
1 Like
USB4 supports a protocol known as PCIe, which serves as nearly direct high bandwidth link to the CPU and is useful for eGPUs and other high bandwidth peripherals.
The problem is that PCIe is designed for use inside the computer, where the signal doesn’t have to travel very far between components and doesn’t have to be squeezed through a thin cable that is also carrying power and other data. USB4 peripherals on the other hand can be connected to the computer through multiple meter long cables, which can result in a meaningful increase in time delay on the signals as well as additional signal degradation, both of which can cause instability.
So USB4 ports have PCIe implemented through a retimer. A retimer is a repeater that is able to adjust for the time delay and repeat the signal to overcome any degradation (hence the name retimer).
Intel has been pushing Thunderbolt (what USB4 is based on) for years and there are plenty of available retimers for Intel CPU that work without any major issues (which is why this issue doesn’t affect Intel Framework Laptops). However on the AMD side of things the retimer options are a lot more limited.
The retimers that Framework is using on the AMD Framework 13 have a behavior where for some reason when the USB-A card is connected it prevents the retimer from properly entering the low power state thst the retimer is supposed to enter at idle.
6 Likes
AIUI the Fw13 Intel versions all had problems with increased current draw when the USB A port expander was fitted. Someone had a thread on here where they were modifying them with a microswitch so that when a device was plugged into the port it operated a microswitch that notified the USB-C port that it now had a device fitted. The native USB A port expander had a wire that caused the USB-C port to think there was an active device plugged in all the time.
1 Like
Indeed there were similar issues (ie. Presence of a card causing retimers to not enter low power states when they’re supposed to) with the Intel models, however those were able to be resolved through software (at least for the USB-A and DisplayPort Expansion Cards, HDMI needed a hardware revision).
In this case however it sounds like there is no software fix for this: