The ports need retimers to ensure proper signal integrity. And unfortunately, Framework said that there aren’t any AMD compatible retimers currently available without this issue.
It does.
If a USB-A adapter is present, it’s an issue. As the illustration shows. I don’t believe it matters if anything is connected to the USB-A adapter.
It’s just about USB-A adapters.
I’m sure it applies to any passive USB-A adapter. Otherwise, Framework would have just created one that avoids it.
Not completely certain, but as long as it’s a USB-C hub, I do believe that will avoid it.
Thanks!
One thing I forgot to ask: What about USB3 (or even USB2) devices connected via USB-C?
For example, there are many USB soundcards (or “headphone adapters”) with USB-C connector (because their primary usage is smartphones), but they still only use USB2. Similarly, my smartphone has a USB-C port and thus could be connected with a USB-C to USB-C cable, but also only supports USB2.
And if it’s just about USB-A adapters, why does it make a difference if I connect a USB3 devices with a USB-C cable or a USB-A cable with a passive adapter?
I don’t really know much about USB retimers, but I thought that they are needed to translate USB4 to older USB standards, not USB-C (which, as mentioned before, can also be used with USB2) to USB-A?
I mean, in the end, what’s the difference between a passive USB-C-to-A adapter and a cable with USB-C port that connects to the same device?
Does anyone have a link with technical details?
Also, an official statement from Framework on this would be nice - ideally a proper explanation on the specs page.
If there is a device connected the whole point is kinda moot since the port needs to stay on anyway. Not being able to sleep when nothing is connected isn’t really a problem if something is connected and you aren’t supposed to sleep anyway.
I don’t think that’s really the issue. Otherwise, it would just be an unavoidable problem. It would exist on Intel board as well. And also Framework wouldn’t be looking for an AMD compatible retimer without the issue.
Retimers are about signal integrity at high frequencies. They receive signals, understand them, and then re-emit them – at high frequencies, traces running the (halfish?) length of a motherboard are long enough for signal degredation (somebody correct me if I’ve got the wrong idea). They weren’t needed with the comparatively lower speeds of USB 2, but they are needed for Thunderbolt / USB-4… not sure about USB 3 stuff, not that it’s particularly relevant here (only learned retimers existed while reading about Framework laptop power drain during sleep).
There’s been this longstanding issue where some of the expansion cards consume disproportionate amounts of power, especially while the laptop’s asleep, even when they’re empty. USB-A expansion cards keep the retimers awake I guess, or something like that, and they use enough power to be noticeable. That’s what this is about. I don’t know, but doubt that there’s a big difference among expansion cards if you’re actively using the port or have something else plugged into it, aside from the power requirements that you’d expect.
And in a subsequent reply, he seems to confirm that it’s part of BIOS updates for all intel generations.
I don’t know if it applies to the USB-A issue, but for some issues where a few people still have problems when it’s supposed to be fixed, I think I recall mention that it could be due to component tolerance on the particular mainboard. Not that a part is “defective” / out of tolerance, but further from the target than most.
So this is only about putting the ports to sleep when nothing is connected, and not about higher power consumption when something is connected? No one mentioned that before, interesting.
If so, related question: Would it be more power-efficient to connect devices (especially ones that don’t need USB4 which, let’s be honest, currently is the absolute majority of devices) to expansion ports 2 and 4, or do they need as much power as 1 and 3 when active?
And I don’t know if Intel laptops are affected, but I think only the specs pages of AMD laptops mention “Higher power consumption” for some ports.
It has been mentioned before in older threads.
The source of the problem is that in the USB-A adapter the USB-C port pin that causes the CPU to turn on power to the port is permanently grounded in the USB-A adapter, as there is no other way to get the insertion of a USB device into the adapter to make the USB-C port on the laptop turn on power.
If you go digging back in previous threads there is quite a discussion about the problem, and someone was looking into modifying the USB-A adapter by fitting a microswitch that is physically operated by plugging in a USB device to ground the line from the USB-C power on pin instead of having it permanently grounded. IIRC FW were quite interested in what he was trying to do, but I don’t know if they were actually looking at doing a production version of the modification.
The redrivers on the intel boards are apparently able to sleep with an usb-a module inserted (as they should but the amd ones are unfortunately bugged in that reguard)
Thank you very much for all the replies and clarifications!
Framework should make a FAQ entry or something about this and link it on the spec pages as reference for that higher power consumption note, so the next person who wonders about it finds the information more easily
I am on BIOS 3.17, 11th gen Intel, can confirm it’s not fixed, at least on my hardware. There probably was more than one power usage issue, though…
Just came across [RESPONDED] AMD Batch 1 Guild - #123 by Zachary_Burnett which has a copy of an email sent out to AMD pre-orderers describing the issue as increased power usage with those ports either empty with USB-A, or populated with USB 2 devices, as opposed to USB 3. So this is probably not exactly the same as my laptop’s sleep issue. Sorry for the confusion. (It’d be nice if there was a “state of power usage by motherboard and expansion card” knowledgebase article someplace…)
Interesting - so apparently the issue is a bit more complex than
(because an empty USB-A adapter has high power draw, and so does a USB2 device connected to it, but an USB-A adapter with a connected USB3 device doesn’t)
So I guess my question about connecting USB2 or USB3 devices to those ports with a USB-C cable is still relevant and unanswered
So why does a connected USB2 device (or no connected device) create high power draw, but a connected USB3 device doesn’t?
Why is it different if a USB2 device is connected directly with a USB-C cable?
Also, the port does not need to stay on when the laptop is in suspend mode, I think?
If anything, usb2 or 3 or displayport alt mode or usb4 with pcie tunneling or whatever else you can pump over the port is active there is no extra power draw as it is active. There is extra power draw when nothing is plugged in because the usb-a expansion card prevents the redriver from sleeping right (cause the redrivers are bugged) when nothing is plugged in.
The problem is extra power use when not in use, when in use it doesn’t matter as is is in use.
It doesn’t need to but the usb4 re-drivers definitely seem to have problems actually doing that.
having an emptyUSB-A adapter plugged in does draw extra power, because the redriver (retimer?) doesn’t sleep right
having a USB-A adapter with any kind of USB-device connected also draws power, but only as much as one would expect from a USB-device being active?
emptyUSB-C adaptersdo not draw extra power
USB-C adapters with any kind of USB-device connected also draw power, but only as much as one would expect from a USB-device being active
When the laptop is in suspend mode:
having an emptyUSB-A adapter plugged in still does draws extra power, because the USB4 redrivers (retimers?) screw up even when they should be sleeping
(ok, to be fair, there is this whole “wake computer by pressing key on USB keyboard” nonsense that might require USB-ports to be at least half-awake)
I haven’t looked into sleep much as I don’t use it but let’s see what I can answer there
USB 1, 1.1, 1.2 and 2 use the same pins and physical interface and are usually handled together so I’d expct thm to behav the same.
That is probably because usb2 (+ 1 and co) and usb3 are entirely separate interfaces, though usb3 is required to have a usb2 interface taped to it. If you look into an usb3 port you can see the front 4 pins, the outer pins are 5v and gnd and the middle pins are the usb2 data lines, then in the back there are the 2 usb3 pairs and a signal gnd, the only thing the 2 interfaces share is power. You can actually use both at the same time even if the spec at least frown upon that, laptops sometimes do that internally when they run out of ports for smartcard readers and fingerprint and stuff.
I find that bit quite funny, “How do we make our new interface backwards compatible? Just tape the old one onto it” XD
