The new warp-speed Thunderbolt interface is pushing the limits on serial data rates — and USB’s response (so far) is pushing the limits of power delivery. While Thunderbolt is four times faster than the 5Gbps USB 3.0, the new USB Power Delivery (USB-PD) specification is set deliver up to 10 times the power!
I recently attended the USB-PD Developers Conference in Washington DC and have been talking with many customers globally. Here’s my take: USB-PD could be a major game changer. But there’s a lot more work to be done before product certifications begin in early 2013.
New USB-PD silicon needs to become generally available. New chipsets will include the logic that will be able to:
- Detect when USB-PD-compliant cables are plugged in.
- Communicate with other USB-PD devices over VBUS to request, approve or deny power capability and availability.
- Expose the power delivery status to the operating system.
Also, new connectors and cables need to be generally available. New connectors and cables will be required to:
- Support higher power levels – up to 20 volts @ 5 amps. (Higher current requires bigger wires in the cable and more robust contacts in the connector.)
- Differentiate between legacy lower power USB cables and new USB-PD cables. (You don’t want to put 100Wonto a legacy cable and cause safety issues. The same thing applies to connectors.)
- New Standard-A plugs have an extended shell, while Standard-A receptacles add insertion detect and PD-detect capability. Micro-B/AB plugs include passive components inside the connector or over-mold that are used to signal the USB-PD devices of their power capability.
Lastly, new circuit protection components are needed to protect the higher-power interface for safety reasons. This will require that:
- The allowable power must stay under 100W (per the USB-PD specification).
- Protection devices must meet necessary specifications such as UL60950. It is unlikely that USB-PD power management ICs (PMICs) will include constant current features due to cost and complexity. External overcurrent safety protection device(s) are the economical and robust choice.
I’ll keep you updated as USB-PD gets off the ground. In the meantime, let me know:
Do you think USB-PD will spell the end to proprietary power connectors including barrel jacks in laptops and ultra-thin computers?
Pingback: The Connecting Edge - Martin Rowe - Tea & USB
A few comments, there is no need to involve the OS in this to get it to work, PD communication is strictly port to port.
It will work even if only a few devices support it, it will work better as more support is added.
The general rule for USB as such and therefore also for Power Delivery is the the power source is responsible for the voltage and it’s own protection and the power sink is responsible for managing the current according to agreement.
There is obviously a need for some circuits to be built and tested for this to work. This is the first USB spec. to actually specify RF circuits and parameters. The communication has specifically been designed to function also through existing cables. Some negotiations can be useful even with legacy cabling, voltage must remain at or below 5V and currents can be negotiated up to 1.5A – the same as for BC1.2.
Sten
Hi Sten- I agree there is no need to involve the OS to get USB-PD to work, although if for example a 60W consumer is plugged into a 36W provider, it would be nice to know through an OS notification that the power supply is not adequate.
If one of the power providers or consumers are legacy USB then power will default back to 5V@2A / 10W (the default start up profile). Because USB-PD does not support legacy data-line manipulation (a “short” from D+ to D-, or D-/D+ pulled to specific voltages), quick-charging of consumer will not be enabled. This can lead to negative user experiences due to very long charge times, especially for tablet PC’s.
I agree with your point that the power source is responsible for the voltage and its own protection, and the sink (consumer) is responsible for managing the current draw. This is highlighted in USB-PD because sinks (consumers) will be tested to ensure they comply with the power consumption they advertise to the provider (host). With that said, especially in higher power providers (hosts), care should be given to choose an over-current device which can protect to the safety specifications (i.e. UL60950) while maintaining acceptable cost, size, time to trip, and series resistance budgets.
It will be interesting to see how quickly things play out over the next few months.
Hi Patrick
Still a few comments.
The intention was that a consumer that could not get enough power to operate properly would flash a LED or in some other way indicate this. No silent failures.
In case of a legacy consumer or provider, there is no negotiation and therefore no agreement. They fall back, not to any profile but to BC1.2 if they both handle this or back to USB.
PD does support BC1.2 in the sense that you may use that as well if you wish. The two protocols are mutually independent and do not share lines. The press release states that PD will coexist with the other older standards.
The range of up to 100W was decided because it just slips into a light weight classification of UL. You can specify more than that in the protocol but you will be responsible to adhere to the more strict requirements if you do. Obviously you as the designer must take care to use proper protection circuits.