VirtualLink standard could pipe VR pixels through one USB-C cable

Anybody with a VR headset has known the joy of plugging in a mess of USB, HDMI, and possibly power cables. A new industry group wants to change all that with a new USB Type-C alternate mode called VirtualLink. If the new standard catches on, folks looking to jack in will need only to run a single USB Type-C cable from their future head-mounted device back to the PC.

VirtualLink's raw specs include four lanes of DisplayPort HBR3 connectivity for video signals, a “USB 3.1” data link for transporting signals between headsets and host systems, and up to 27 W of power delivery to juice up compatible headsets. The consortium says these features will help VirtualLink support headset features like integrated cameras for inside-out tracking and augmented reality, as well as the raw bandwidth requirements of higher-resolution and higher-refresh-rate headsets yet to come.

So far, VirtualLink is supported by Nvidia, Valve, Oculus, AMD, and Microsoft. The members of the consortium tout VirtualLink's open nature and the fact that it's “purpose-built for VR,” including considerations for latency that more general display and audio standards might not account for. The VirtualLink 1.0 specification is still a work in progress, but interested parties can sign up for an overview of the spec at the consortium's website.

Comments closed
    • Namtrooper
    • 1 year ago

    Still not the same as untethered. But better.

    • sweatshopking
    • 1 year ago

    Do we still want cables on our heads?

      • Redocbew
      • 1 year ago

      I haven’t really been following VR, but I believe the facts are that cables suck, but they suck less than wireless.

        • GrimDanfango
        • 1 year ago

        Indeed. The problem is, a super-low-latency, super-high-bandwidth wireless connection is no simple thing to create.
        There are already 3rd-party options, but running at the ragged-edge even on the current lacking resolutions of 1st-gen headsets.

        Clearly they don’t think wireless will keep up with the requirements any time soon – they wouldn’t be bothering trying to create a unified standard wired connection only as a short-term stopgap.

      • Generic
      • 1 year ago

      Are you saying that Intel should introduce Light Peak for IR blasters?

      ‘Cause that’s what I’m saying. 🙂

    • willmore
    • 1 year ago

    How long of a cable is this going to support? Most of the alt-modes of USB-C are only good for 0.5m with a passive cable and 2m with an active one. Does this have some secret sauce to allow longer cables? I don’t see much value to 2m worth of teather length.

    If you’re looking into room scale VR, then you’re going to want 5m at the least and likely 10m once you consider all the plumbing it’ll take to arange things so you don’t trip over them.

      • VincentHanna
      • 1 year ago

      Well, assuming for a moment that we aren’t moving to a standard where the GPU driving the experience is either attached at the hip or the face….

      You would have a DisplayPort/Power/USB/3.5mm headphone (for the lolz) into a interposer, and the output would be the one 2M USB cable.

      • psuedonymous
      • 1 year ago

      [quote<]How long of a cable is this going to support?[/quote<]The example cable in the VirtualLink spec docs is 5m, but all cables are assumed to be using conditioning ICs by default so as long as you conform to the signal requirements (minimum dB drop, maximum xtalk, etc) you can make one as long as possible. One gotcha is that in order to comply with the existing Type C spec for repurposing of the USB 2 pins, the cable cannot be reorientable at one end. That means it's captive at the HMD end.

        • willmore
        • 1 year ago

        Thank you, that’s very clear. Sounds like it’s a spec for a USB-C socket then, not really the cable–if the cable has to be part of the HMD.

        Second point, isn’t the cable the most likely part of one of these to be damaged and break? Making it part of the HMD seems problematic. Maybe a different socket/plug on the HMD side so they’re not *one* part?

    • JosiahBradley
    • 1 year ago

    Yes Yes Yes!

    • UberGerbil
    • 1 year ago

    27W sounds like a lot, but then I came to realize I have zero sense as to how much power a typical set of VR goggles actually draws. Anybody know?

    (Yes, I googled but other than some frustrating Reddit threads I didn’t see anything, and then i got sucked into this Vive Pro tear-down:)
    [url<]https://www.ifixit.com/Teardown/HTC+Vive+Pro+Teardown/106064[/url<]

      • Duct Tape Dude
      • 1 year ago

      The Rift is powered by a single USB 3.0 cable, which is specced at under 5 Watts.

      • Growler
      • 1 year ago

      The average goggles use 1.21 jigawatts.

      No, wait. That’s time machines. Time machines use 1.21 jigawatts.

        • Namtrooper
        • 1 year ago

        I think you meant Jizzawatt.

      • Chrispy_
      • 1 year ago

      They’re just smartphone screens with almost none of the CPU/GPU overhead of a smartphone. At a guess, 2-4 Watts, since most smartphones charge slowly on a 5W charger when the screen is on but they’re idle, and lose charge very slowly under load.

        • psuedonymous
        • 1 year ago

        The screens can draw more power than an equivalent smartphone panel (need to achieve same luminous flux but with far shorter illumination time due to low-persistence driving vs. sample-and-hold, plus driver electronics drawing more due to increased refresh rate) but there are other power draws too. The tracking system may draw an appreciable amount (Constellation uses bright LEDs, SteamVR Tracking needs relatively beefy DSPs to gather accurate timing data), and the IMU requires active heating to reach a stable operating temperature above ambient for accurate operation. If you calibrate at room temperature, then if room temperature ever varies then IMU output goes out of calibration. Calibrate at e.g. 60°C (above any normal ambient temperature) then your IMU will need active heating but by modulating that active heating you can maintain that temperature stably. A stable temperature to calibrate at means that calibration remains valid during use.

          • willmore
          • 1 year ago

          Even if the backlight LEDs are strobed, that doesn’t raise the *average* power to them, so just put some small capacitors in there. Secondly, the IR LEDS use very little power and are *super* bright. Decades of development of handheld remote controls for TVs and other appliances has left us with extremely efficient IR LEDS. They normally have forward currents of .1A and forward voltage drops of 1.2V. That’s just 0.12W each. Need a dozen? That’s 1.44W. Even if you really need them at full power, that’s a tiny part of your power budget–chances are you don’t need them running full out like that.

          Your point on the IMU is a good one, but how small are MEMS IMUs these days? Chip scale package? Surround it with a ring of resistors, PWM that from whatever SoC’s controlling things and use the temperature sensor that’s already part of the IMU to keep it in check–simple PID loop. Again, you’re talking a fraction of a Watt.

          I think the 27W leaves them plenty of headroom for what they need to do. Sure, they could have used the USB-PD spec and gotten 100W, but 27W should be enough for what they’re doing.

          Do you know the specific voltage/amperage? I assume it’s 9V/3A from the 27V figure. 18V/1.5A would be smarter for longer cables, though…

            • psuedonymous
            • 1 year ago

            [quote<]Even if the backlight LEDs are strobed, that doesn't raise the *average* power to them, so just put some small capacitors in there.[/quote<]For the most part, no stroked baclight because there is no backlight panels. OLED is the norm. Capacitors are already present, but you can't just wait over a frame to fill them then pulse the pixel, because the whole exercise to to minimise latency. Instead, fill + discharge needs to happen in the shortest possible time, which means high peak power. [quote<]Decades of development of handheld remote controls for TVs and other appliances has left us with extremely efficient IR LEDS. They normally have forward currents of .1A and forward voltage drops of 1.2V. That's just 0.12W each. [/quote<]For the tracking LEDs, a remote controller LED does not cut it. You need consistent high-angle illumination, with high output power (to provide higher range without needing to raise camera sensitivity, which introduces noise and affects discrimination). And you have a good number of them distributed across the HMD and strap to ensure good coverage (IIRC ~40 for CV1). IMU heating is in singe-digit watts during peak startup. [quote<]I think the 27W leaves them plenty of headroom for what they need to do. Sure, they could have used the USB-PD spec and gotten 100W, but 27W should be enough for what they're doing.[/quote<]27W offers headroom over current HMDs, but future components (e.g. high grade eye tracking) will eat into that quickly. The standard proposes that HMDs that need the higher power level should -"where reasonable" - gracefully downgrade to a lower power draw if connected to a port with the lower power capability. [quote<]Do you know the specific voltage/amperage? I assume it's 9V/3A from the 27V figure. 18V/1.5A would be smarter for longer cables, though...[/quote<] 15W: 5V @ 3A 27W: 12V @ 2.25A Negotiation is via standard USB-PD. 12V was probably chosen as it's the standard internal voltage GPUs run off of.

      • Puiucs
      • 1 year ago

      many headsets have peripherals that can connect directly into them. the extra power can be used to power them or similar devices.

    • NTMBK
    • 1 year ago

    Hoorah, another incompatible USB-C Alt mode!

      • chuckula
      • 1 year ago

      I am somewhat offended that AMD and Nvidia are both supporting the same standard.

      How are we supposed to re-post the same FreeSync complaint for automatic upthumbs when both of them use the same VirtualLink!?!?!

      • UberGerbil
      • 1 year ago

      Obligatory reference to XKCD #927

      • psuedonymous
      • 1 year ago

      [quote<]5 plugs on the side of a device, one power input, one DP, one Thunderbolt, and two USB[/quote<] This is fine. Trying to push the wrong device into the wrong port and causing physical damage is just a fact of life (so many ethernet ports killed by USB plugs). [quote<]5 plugs on the side of a device. One can be used as DP (labelled), one can be used as Thunderbolt (labelled), all can be used as USB, any can be used to charge.[/quote<] This is terrible! I am too confused to plug things in (despite the billboard class reporting what is plugged where and notifying you which input to use)!

Pin It on Pinterest

Share This