Advertisement

Article

SuperSpeed USB (USB 3.0): More than just a speed increase

SuperSpeed USB (USB 3.0) has been getting a lot of attention now as products become available in the market. The most obvious benefit is the more than 10 times increase in speed over USB 2.0 high-speed; 480 Mbps to 5 Gbps – but there are several others. This article looks at what is new and better with SuperSpeed USB protocols and power management versus USB 2.0.

More than just a leap in speed
Hopefully. by now you have heard of SuperSpeed USB and are familiar enough to know it is the next evolution of the Universal Serial Bus (USB). If not, I recommend watching a short video (Reference 1 ) that talks about the basics of the specification and how it compares to High-Speed USB, or USB 2.0. The most obvious difference in SuperSpeed USB is the over 10 times speed increase from 480 Mbps for USB 2.0 to 5 Gbps.

The USB 3.0 Promoter’s Group focused on delivering customer value in three other primary areas: 1) foremost, after the speed increase, was improving the power efficiency of the bus; 2) next was maintaining backwards compatibility; and 3) was improving the data transfer efficiency itself.

Use less power to move the same amount of data
The first key area of development was to improve power efficiency. This is needed for extending the battery life for portable devices, whether hosts or peripherals. There are multiple aspects of the new specification that were developed to address reducing the overall power footprint of new USB devices including:

  • Elimination of device polling
  • Elimination of broadcasting packets
  • Intermediate low-power states
  • Sata transfer speed increases by 10

The first change to help reduce power consumption was to eliminate device polling. In USB 2.0, the host controller continuously polls each of the devices in the tree to check if they have data that they need to send to the host system. Device polling means that all devices must be fully “alive” and capable of transmitting data at all times, and that each is “always” burning power by transmitting NAKs to the inquiries of the host system when they do not have data to transmit. Finally, the host is continuously consuming power asking the devices if they have any data, which in most cases they do not.

The next change was changing the packet transfer from broadcast to directed. When a USB 2.0 host has data to send to a device, it broadcasts the data on each of its ports. Each hub in the tree must also re-broadcast the packets on each of its downstream ports. Lastly, each of the devices on the bus must process the data (consuming power) to determine whether they are the intended target of the transfer.

In SuperSpeed USB, the protocol was changed to direct the packets to only the intended target. This requires a little more intelligence in the host. It must know specifically where in the tree each device is, including what hub port (or ports, if multiple hubs are between it and the host) from which it is downstream. This reduces the overall power consumption in that only the specific downstream host and hub ports to which the device is connected must transmit data, and only the target must process the data.

The third power reduction-focused change was to define two intermediate idle states. In USB 2.0, there are two states: ACTIVE and SUSPEND. In SuperSpeed USB, there are also FAST EXIT IDLE (U1) and SLOW EXIT IDLE (U2), in addition to ACTIVE (U0) and SUSPEND (U3). This allows devices to lower their power consumption when they are not transmitting or receiving data.

In the FAST EXIT IDLE mode, the link goes idle but the clocking on the device stays on. While in SLOW EXIT IDLE, both the link and clocking are turned off, which requires a longer time to re–train the link before data can be transmitted. The ACTIVE and SUSPEND modes remain the same in both USB 2.0 and USB 3.0.

The 10-times speed increase also enables lower overall power consumption. No, I am not saying that a 5-Gbps transceiver requires less power to transmit data than a 480-Mbps transceiver. I agree with all who have claimed that a 5-Gbps transceiver requires between two and five times the peak current than that of USB 2.0 transceiver.

However, what I am talking about is lowering the overall power footprint, not just the peak current that is consumed at a small slice in time when the transmitter is active. When you factor in the approximately 10 times decrease in the transmitter’s actual active time, the total power required to transmit a fixed amount of data (for instance, moving a file from the PC to a flash drive) is between 20 percent (two times peak and 1 /10 the time) and 50 percent (5 times peak and 1 /10 the time) the total power required for transmitting that same amount of data over USB 2.0.

When you combine the bus usage efficiency (no broadcast packets and eliminate polling), the improved IDLE power states, and the lower average transmit power, SuperSpeed USB consumes approximately one-third (or less!) the power of USB 2.0!

What does backwards compatible really mean?
The next key attribute that was a focus during development was maintaining backwards compatibility with what Brian O’Rourke of In-Stat has called “the most successful PC interface of all time” (Reference 2 ). It was determined during the development effort that the existing cable and connector solutions were not going to be adequate to reliably transfer data at 5 Gbps. The developers determined that the signaling would need to be done over separate conductors, versus those used for USB 2.0. They choose to use a full duplex differential signaling method based on the PCI Express electrical specification.

At the same time, it was decided to not make any changes to the existing USB 2.0 signaling. This required adding at least two new differential pairs, in addition to the existing USB 2.0 differential pair and VBUS and GND. When you include the ground shield for the two new SuperSpeed differential pairs, this brings the total conductors to nine in the cable and nine contacts in the connector.

So what then does backwards compatibility really mean? If we approach this from the end-user perspective, it means that ALL my existing products that are compliant to the specification will seamlessly connect to and work with all new products supporting the new specification! This means that the existing cables (i.e., plugs) must be able to be inserted into the appropriate new receptacle. The reverse is also true, that the new cables must be able to be inserted into the old receptacles – again, where appropriate.

Obviously, any cable and plug that will support the new SuperSpeed USB specification and data transfer will have both new conductors in the cable portion and new contacts in the plug. Also, any new receptacles must have new contacts as well to accept the new connections required.

In the USB world there are two basic connector types. The A-receptacle is what we are all very familiar with from its presence on every PC shipping today. The A-receptacle accepts A-plugs; these again are very familiar from the captive cable that mice and keyboards use as well as from the flash drive connector. The B-receptacle is what we see on peripherals and has three sizes: standard, mini, and micro.

The key to being backwards compatible is for the new SuperSpeed USB A-receptacle to accept both new (USB 3.0) and old (USB 2.0) A-plugs, and for the older USB 2.0 A-receptacle to accept the new USB 3.0 A-plug. Obviously, if either the plug or receptacle is only USB 2.0 compliant, then the data transfer will be limited to USB 2.0 speeds. The solution was to add five new conductors on the insert side of the existing plugs and receptacles. This allows for the same mechanical interface as USB 2.0, and enables full backwards compatibility (see Figure 1 ).


Figure 1: SuperSpeed USB A–side receptacle and plug
(Click on image to enlarge)

On the peripheral side, the challenge is more difficult from one perspective – there are more connector sizes to be concerned about, but at the same time easier.

Why easier? The new USB 3.0 B-plugs do not have to be able to be inserted into old USB 2.0 B-receptacles. You only need the new B-plug (and the cable), if the device is USB 3.0 capable. If the device is only USB 2.0 capable, then the existing USB 2.0 cable with the old plugs on both ends is sufficient since the other end of the cable is capable of being inserted into new machines – see previous paragraph. The new B-receptacle (regardless of size) must accept both the old and new B-plugs for backwards compatible. Therefore, any change to the form factor can not preclude insertion of an existing USB 2.0 cable.

Now let us look at the three USB 2.0-compliant B-side connector options. The most straightforward is the standard B type. This is the larger format receptacle that you will typically see on a printer, scanners, and other larger form-factor peripheral devices. The decision was to add a “bump” to the top of the B-connector to facilitate the placement of the new contacts (and conductors) for the SuperSpeed signals. Figure 2 shows how this was accomplished in the standard B size plug and receptacle. The receptacle accepts either the old USB 2.0 plug or the new USB 3.0 plug.


Figure 2: SuperSpeed USB B-side plug and receptacle
(Click on image to enlarge)

The Promoter’s Group made the decision to not update the Mini-B connector type. There are multiple reasons for this, but the primary reason is due to legislative decisions being made around the world. Multiple agencies around the world have determined that the USB plug should be the standard interface for charging the battery in handsets. There was concern that every new phone required a new charger with a new custom connector.

This was causing a disposal problem that mandating a standard interface would help alleviate. The Micro-B receptacle was chosen as this standard interface and all handsets moving forward will be migrating to it, as are many other small portable consumer products that have started using USB for charging their battery. More information can be found by joining the USB Implementer’s Forum and joining the Device Working Group focused on battery charging (Reference 3 )

The existing Micro B receptacle (Figure 3 ) has insufficient room to add the five new conductors – requiring a major change. Therefore, the working group came up with the side-by-side solution which can accept either the USB 2.0 Micro B-plug or the new SuperSpeed Micro B-plug, also shown.


Figure 3: SuperSpeed USB micro B-side receptacle and plug
(Click on image to enlarge)

Along with mechanical backwards compatibility, the goal was to maintain the extensive device driver infrastructure. The same data transfer types, interrupt, bulk and isochronous, were maintained. Finally, this standard preserves the existing USB ease-of-use expectations. All existing USB 2.0 devices will continue to work as we have always come to expect them to on our new SuperSpeed USB enabled machines.

Reducing the wasted bits
The fourth key value is improving overall bus usage efficiency. I have already touched on the first aspect of this: the elimination of polling. Additionally, the full duplex architecture of SuperSpeed USB allows for concurrent bi-directional data flow as opposed to the half duplex USB 2.0 architecture.

The host polling of devices is clearly a waste of bus usage, but the question is what does elimination of polling really mean. In honor of my mom who was a school teacher for many years, I use a classroom analogy to describe what I mean. If the classroom was a USB 2.0 system, the teacher would go around the classroom and ask each student one-by-one if they had a question. When a student had a question, the teacher would answer that question. She would then continue on with the rotation around the room until each student had been queried at which point she would begin again.

If we think about the classroom as a SuperSpeed USB system, the student would simply raise their hand when they had a question and the teacher would acknowledge the student and answer the questions as needed. This is the asynchronous notification method of IN data transfers. The peripheral device sends an ERDY (Endpoint Ready) to the host when it has data to send to the host. The host then sends an ACK (ACKnowledge) to the peripheral when it is ready to process the transfer.

In USB 2.0, the half duplex bus with only a single differential pair for data transfer causes two issues with bus efficiency. The first is that the bus has to “turn-around” every time the direction of data flow changes. This means that the transmitter has to turn off on one end of the connection while at the other end the receiver is turned off. Once this completes, the reverse happens where the receiver turns on in the first device and the transmitter turns on in the second device.

This inherently drives significant down time on the bus, cutting into the efficiency. A second consequence of this is that any given transfer must complete before the next one can begin. This means that the receiving device has to acknowledge receipt of the data and the transmitting device must see the ACK before the next data payload can be sent across the bus. Figure 4 shows at a high level what an OUT (from host to a peripheral) transaction would look like in USB 2.0.


Figure 4: USB 2.0 OUT transaction
(Click on image to enlarge)

In a SuperSpeed environment, where there are two differential pairs on every device, one for transmitting and one for receiving, the bus turn around dead time is eliminated. It also allows for the additional payloads to be sent from the transmitting device before the acknowledgement has come back from the receiving device.

This change drives the need for a little more intelligence on both devices just in case of errors. A device may need to go back in time and re-transmit a previously “completed” payload, if the “acknowledge” comes back as an error in the received data. The protocol uses credits to determine how many data payloads a device can have “active” at any one time before it can no longer transmit/receive new payloads before the successful ACK had been processed.

Essentially, each time a data payload is sent from the host to a peripheral (or from the peripheral to the host), a credit is deducted from the peripheral’s account. When the ACK is successfully processed by the original transmitting device, a credit is reissued to that peripheral’s account. See Figure 5 for how SuperSpeed USB OUT transactions occur.


Figure 5: SuperSpeed USB OUT transaction
(Click on image to enlarge)

Don’t forget that increase in speed
In any data transmission environment, there is always a bottleneck – or the location in the signal chain that limits the performance of the entire transmission. While USB 2.0 high-speed (and even USB full-speed, 12 Mbps and low-speed, 1.5 Mbps) is more than adequate for some applications, for many PC-centric applications USB 2.0 has become that bottleneck in the last few years. This is actually the whole point of developing a new specification aimed at data transfer: eliminate the current bottleneck!

Whether it is an external rotating media (i.e., hard disk drive, CD drive, DVD drive), a solid-state drive based product, or the ubiquitous flash drive – the storage media has become capable of transferring data at a rate that exceeded what USB 2.0 could do.

Are many of those media capable of reading and writing data near 5 Gbps? There are not many today – but that is the point, move the bottleneck elsewhere. Will SuperSpeed USB become the bottleneck again? In time, probably. But with the exception of some of the very best hard disk drives, SuperSpeed USB should provide headroom to keep the pinch point elsewhere for at least the next five years.

In summary, the USB 3.0 Promoter’s Group had four key values in mind as they developed the specification. Decreasing the power required to transfer data, maintaining backwards compatibility, increasing the bandwidth utilization efficiency, and of course the over 10 times increase in the raw bit rate. These values will enable increase, both in the Sync-and-Go experience of the consumer as well as extend the battery life for these content-rich consumer products, and provide the needed headroom to do the same for flash-based products for the next five years.

References
[1] What is SuperSpeed USB Video on TI’s E2E Community, http://e2e.ti.com/videos/m/analog/121743.aspx
[2] USB: The Universal Connection, IN020007MI, Multimedia Interfaces, March 2002, http://www.instat.com/abstract.asp?id=161&SKU=IN020007MI
[3] USB-IF Device Working Group, http://www.usb.org/about/dwg_charter/

To read more from Dan on SuperSpeed USB, please see his Consumer & Computing Interface blog (http://e2e.ti.com/blogs_/b/interface/default.aspx on TI’s E2 Community at http://e2e.ti.com/.
To learn more about SuperSpeed USB interface solutions, visit: http://www.ti.com/superspeedusb-ca.
To see our video that demonstrates the speed of SuperSpeed USB3 versus High-Speed USB 2.0, visit http://www.ti.com/usb3video-ca.

About the Author

Dan Harmon is Product Marketing Manager for Consumer & Computing Interfaces at Texas Instruments as well as serving as TI’s USB-IF Representative and TI’s USB 3.0 Promoter’s Group Chair. He earned a BSEE from the University of Dayton and a MSEE from the University of Texas in Arlington. You can reach Dan at dharmon@ti.com

0 comments on “SuperSpeed USB (USB 3.0): More than just a speed increase

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.