USB3.0 reliability design

The most successful PC interface today is non-USB, with over 6 billion installations and nearly 100% penetration on PCs and interface devices. Although high-speed USB's 480Mbps data transfer speed can meet the existing needs of many consumers, the increasing demand (such as high-definition video and faster download of digital audio and video files) has promoted the development of SuperSpeed ​​USB (3.0). With the new generation of general interface specifications, the data transmission speed is usually doubled, while the USB 3.0 bandwidth is increased by a factor of ten. In addition, the USB 3.0 specification no longer uses a simple master-slave, packet-broadcast data transmission architecture, but uses a more complex two-way packet switching architecture.

The main challenge that USB3.0 system designer faces is to solve the problem that 5Gbps signal transmission speed brings, the designer must solve the problem including the sensitivity of the system to signal attenuation and increase of shaking. In addition, USB 3.0's downward compatibility with the USB 2.0 interface complicates matters because USB 2.0 was originally designed for lower transfer rates.

Perhaps the biggest problem that product designers encounter is to reach consumers with low-cost expectations for previous-generation USB products. USB3.0 uses double data rate technology to increase the transmission rate up to 5Gbps, thus requiring high-speed signal integrity solutions. High-speed signals can be transmitted over cable lengths up to 3 meters, multiple interfaces, and on long-haul lines on the PCB. Product designers must handle signal attenuation and jitter issues with care.

Signal margin budget

Compared with low-speed signals, the loss of quality of high-speed signals is much more. The accumulation of signal losses caused by PCB lines, connectors, and cables will quickly impair signal quality. According to the signal margin requirement, the allowable channel loss of USB 3.0 (from the transmitter eye diagram to the receiver eye diagram) is 6 to 9 dB at 2.5 GHz. In addition, SuperSpeed ​​generally has -3.5dB de-emphasis, so the overall signal loss budget is 9.5~12.5dB.

In order to meet the compatibility requirements, the USB 3.0 signal must be able to pass 3 meters of cable and keep the signal eye opening with sufficient width. But this is only part of the path, because in the compatibility test is measured after the receiver has been equalized signal. For example, in a typical notebook computer architecture, the distance from the USB controller chip to the connector is approximately 10 inches, so the signal may actually go through about half a meter of path and several connectors.

In a typical laptop-to-peripheral application, the signal will go through a 12-inch PCB line (10-inch line in a laptop, 2-inch in an interface device), which will result in 3.552dB line loss (12 inches x 0 .296dB/in, using 9/10/9 FR4 traces, 2.5GHz signal speed, does not include any PCB vias). In addition to the -1dB loss produced by each connector, the total loss produced by the two connectors is 2dB, so the signal loss budget left for the cable is only 5.948-8.948dB. The signal attenuation of the shielded differential signal pair cable at 2.50 GHz is approximately 1.9 dB/m at 34 AWG 4.4 dB/m to 26 AWG. Table 1 lists the maximum lengths of SDP cables that can meet the USB 3.0 specification calculated from these numbers. The third column to the last column in Table 1 shows that even without the use of signal conditioning products, even the high quality 28AWG cable cannot reach the 3 meter length required by the USB 3.0 specification.

Product manufacturers can choose to use high-quality 26AWG SDP cable to pass the certification test, but this kind of cable is very expensive. If the product is shipped, the cost of the cable will greatly increase the cost of the product. In addition, even if such cables are shipped with the product, there is no guarantee that users will not mix other low-quality cables. Although it can be emphasized in the product that the proper cable must be used, users of the USB product are accustomed to connecting the product without worrying about the quality of the cable. The key issue here is that when the product performance is low due to the use of low-quality cable connection products, the user intuitively believes that it is a product problem rather than a cable problem. This may result in higher product returns and significantly reduce the early profits of adopting USB 3.0 specification products. Another important factor affecting the adoption rate of new technologies is the difficulty of use. If users feel that USB 3.0 technology is difficult to use, it will seriously affect the growth of the market.

Using signal conditioning techniques to restore signal quality

When the signal quality is degraded due to jitter (jitter) and attenuation, the signal conditioning device can be used to recover the signal quality. Signal conditioning devices are also referred to as redrivers because the signals are redriven before being transmitted. The redriver can be placed between the transmitter and the receiver to recover, adjust, and send the received signal.

If placed before the receiver, the redriver can effectively open the eye of the signal eye to restore the signal quality to the acceptable range. The redriver can adjust the signal and use the signal equalization function to reduce signal attenuation and jitter, thus providing a clean signal transmission path over longer distances and through multiple connectors. The signal can be sent back to its original strength again, as if the signal path were broken into several paragraphs and the quality of the signal was restored in each paragraph. Redrivers also allow product designers to adjust the best signal-equalization effects for specific application types, thus ensuring that the device passes the most stringent USB 3.0 certification requirements.

The signal conditioning function can increase the signal margin, allowing product design engineers more room to expand the signal transmission distance, or more flexible design of the path the signal travels on the PCB, especially if it may require the use of fewer layers of PCBs to achieve Better signals are isolated from ground, or there is a greater chance that one design can pass the verification. Improved signal quality not only gives design engineers more flexibility and choice in the signal chain, but also increases product stability due to lower bit error rate, reduces transmission errors, and increases actual and effective information throughput. Makes the system work more efficiently.

In consumer applications where the device needs to be frequently connected and removed, the adaptive signal equalization function has the best effect because of the use of different lengths of cable plus signal wiring inside the pluggable peripheral device to make the overall signal path Can be changed at any time. For example, the signal-equalization parameters optimized for 3-meter cables will impair signal integrity when applied to USB-free mobile storage. The redriver ensures continuous adjustment of the actual signal path by means of continuous retraining.

Conversely, if the signal path is stationary, the redriver does not need to use the re-adaptation function. In this kind of environment, for example, servers that never need to remove any cables, and then adapting functions may cause negative effects. In a stable application like this, configurable signal equalization is the best choice.

High signal quality drives cost reduction

The retuning driver increases the signal balance of the receiver by 6dB, which can greatly increase the transmission distance of the signal on the SDP cable, and thus can help solve the trouble of the cable. If the transmitter is also equipped with a redriver driver, a total of 12dB signal equalization can be added, ensuring that the signal is transmitted over 34AWG cables longer than 3 meters and passed the certification test (Table 1).

Increasing signal equalization allows signals to be transmitted over longer and cheaper cables, allowing consumers to use inexpensive, flexible, aesthetically pleasing, and finer cables without having to buy thick and expensive cables to reduce signal loss. High quality cable. Therefore, if you consider the cost of the cable, using a redriver not only ensures that the product works as intended, it also reduces the overall cost of the product. Although the use of a redriver will add an on-board component, the redriver only has a very small PCB area (4mm2), which is very important in portable devices.

Electrostatic discharge (ESD) is another consideration when using a redriver. To reduce the cost of the system, the USB interface is likely to be integrated into the main chipset. Not only does the USB interface control device have no protection against ESD, and the entire chipset may be damaged, making the entire system unusable. Because the redriver chip is located between the connector and the transmitter/receiver, the chipset with the built-in USB interface can be isolated. If an ESD event occurs, the redriver can protect the system's control chip, and in the worst case, only a single USB port is lost, allowing the system's other functions to work properly.

Redriver must have the ability to recognize the communication protocol in order to efficiently perform the signal balancing function. Since the redial drive does not have a device identification ID, it cannot stop the transfer of information like a USB terminal device. If the transfer driver can recognize the communication protocol, it can support functions such as receiver detection and electrical idle, not only reply and transmit signals. Ideally, the redriver will support the continuous-time linear equalization technique defined by the USB 3.0 specification, which is the same technology that the transmitter expects the remote terminal device to use to equalize the signal. Finally, the redriver must work without being noticed, otherwise the root device after the redriver will not be identified, even if this condition can work in the lab, but in practical applications No way.

Bidirectional signal conditioning

Implementing a consumer application is the same as the user expecting USB 3.0 to be at USB 2.0. One of the important design considerations of USB 2.0 is that it can be produced at a very cheap price, so many manufacturers mistakenly believe that USB 3.0 devices can be produced in the same way. In addition, the pressure to reduce costs will allow manufacturers to consider the use of low-quality cables, or allow the receiver to solve signal conditioning problems and other shortcuts. Because of the market's expectation of low-cost hubs and peripherals, manufacturers will produce these products and use high-quality cables to achieve the results required for passing certification tests. However, the average user wants to use any USB cable and the system can work stably.

In this case, responsible manufacturers have to face the problem of the secondary products in the market: These products have price advantages, but when connected to devices that only have receiver signal conditioning Can't work stably. In other words, these devices can receive signals stably, but when sending signals to these secondary products, there is a significant reduction in performance or even errors. For example, an external hard disk constantly emits incorrect information from the receiver due to poor signal quality. The user will think that this is a problem with the hard disk, not a low-quality cable or hub that does not appear to be problematic when used with other products. Equipment problems, which will lead to increased return, profit loss.

A receiving adapter is installed on each external connector to ensure that the signal can be correctly restored both when entering and leaving. In addition, if the transmitted signal is also adjusted, even if other devices do not perform signal conditioning at the receiver, the product design engineer can also ensure that the device passes any length of cable and communicates with other devices.

Redriver products on the market are divided into single channel and dual channel. The dual-channel redriver has signal conditioning capabilities in both the receive and transmit signal paths, thus ensuring connection to low-quality devices that do not receive signal conditioning. The single-channel repeater driver has the flexibility to separately adjust the receive and transmit signal paths.

Another feature when using a redriver product is the flexibility of the supply voltage. For example, a server typically uses 1.2V and a PC uses 3.3V. A re-driver product that can select any of the previous voltages can satisfy both markets, so it can use a larger overall shipment to drive down prices. In addition, its built-in LDO feature saves an external LDO component.

Questions about clocks and handshaking

USB controller chips, devices, terminals, hubs, and processors with built-in USB interfaces all require an accurate and reliable clock source. The clock signal plays an important role in maintaining good SuperSpeed ​​USB signal quality because the available jitter budget is reduced as the clock speed increases.

It is a common misconception to think of clock technology as an old technology. In fact, clock technology (especially at high speeds) is an extremely sophisticated technology. Since high-speed clocks are expensive, the more cost-effective method is to multiply the clock signal at a lower speed by several times. However, when the low-speed clock signal is multiplied several times to a high speed, the generated jitter is also amplified by the same factor, thus occupying a limited signal margin. The product designer's challenge is to balance the input speed (equivalent price) and resulting jitter, so the clock buffer can only tolerate very low jitter, so that when the clock signal is doubled, it will not produce excessive jitter.

Likewise, USB switches must provide smooth transitions. For example, the switch can reduce the signal interface used in practical applications such as docking stations and reduce costs. By maintaining low resistance and low impedance, the switch maintains low input signal loss and does not allow feedback of the signal to the transmitter, thus maintaining signal integrity.

Use USB to manage intelligence to save power

Many consumer USB 3.0 terminal devices use batteries, so USB 3.0 has several new power-saving modes, including idle, sleep, and suspend modes. Product designers can take advantage of these new features by turning off the redriver signal circuitry to extend product operating time.

The USB interface is dual channel, and the transmit and receive channels can be individually switched. The power management of the transponder driver of the transmitter is relatively simple and can directly enter the sleep mode when no signal is transmitted. USB has the ability to be hot-swapped, and the management of the receiver is complex, because it is possible to receive signals at any time, and the redriver must have a mechanism to support complex front-end signal detection capabilities in order to maintain device transparency.

When connected to other devices, the redriver can use the electrical idle threshold more frequently to check for signal transmissions. When the signal is at the threshold voltage, there is no signal input and deep sleep mode can be entered. This threshold voltage is typically 100mV, but if you want to increase the sensitivity of low-voltage signals transmitted over long cable runs, you can adjust it to a lower 60mV. For example, the digital TV's USB interface may be connected to other very distant electronic devices and therefore require higher sensitivity to detect signals. The USB 3.0 interface in the server application, because of the longer PCB layout, the lower threshold voltage is also a better choice.

USB3.0

Product designers familiar with PCI Express Gen 2 may be more comfortable with USB 3.0 designs than are familiar with USB 2.0 because USB 3.0 and PCI Express Gen 2 use similar link initialization, packet structure, and bug fixes. In addition, the two communication protocols use identical transmit and receive modules and units, including scrambling codes, 8b/10b encoding, serializers/deserializers, and the like.

The similarities between USB3.0 and PCI Express Gen 2 will greatly reduce the application of PCI Express Gen 2, especially because of USB 3.0's signal throughput and the need to retranslate the communication protocol to USB. For example, USB 3.0 has been designed in the docking station because USB 3.0 does not need to be converted again (the HBA or PCI-USB controller is required) as PCI Express Gen 2 signals. Using the USB 3.0 docking station will have the ability to fan out the signal from the hub, thus simplifying the overall architecture and reducing costs.

USB is the most common consumer electronics communication protocol today, and has a strong advantage over PCI Express. Because USB 3.0 connectors and controllers must support USB 2.0 devices downward, vendors transitioning from the PCI Express Gen 2 platform to USB 3.0 can immediately enter this vast market. Supporting the USB2.0 protocol device downwards did increase the difficulty of designing USB 3.0 products. However, due to USB 2.0 and USB 3.0 sharing connectors, most of the challenges are in chip design.

PCI Express Gen 2 still has a place in the clear off-chip interface, including the processor's internal bus interface and other servers, storage, embedded applications. Of course, the premise of this existing advantage is that PCI Express Gen 2 is still the main built-in interface.

It is expected that USB3.0 will compete with other new and old interface specifications, such as HDMI, DisplayPort, PCI Express, DVI, and other monitor-connected interfaces. USB3.0 has a high channel width and low price. When integrated into a chipset, there is no reason why it should not be used to connect a monitor. USB3.0 will not completely replace the HDMI and DisplayPort communication protocols, as they each have their own applications. For example, HDMI has an impact in the surveillance camera market because of its reverse channel capabilities. However, it can be expected that USB 3.0 will erode the market share of other communication protocols that are using at least one USB interface.

USB3.0 will change the status quo in the electronics market in many ways, but in order to achieve high performance and stability, it is necessary to use signal conditioning products to compensate PCB, connectors, and most importantly cable signal loss. Consumer applications cannot accept the high-price, high-speed components used to maintain signal integrity, but they also cannot tolerate the decline in application performance and reliability due to signal integrity issues. If the use of a redriver at the transmitter and receiver ends in restoring signal quality, engineers can maintain a good signal margin to use longer cables or to provide greater flexibility in PCB routing. They will also ensure that the product design not only passes the rigorous USB 3.0 certification test, but also uses low-quality cables and has excellent interoperability with other USB 3.0 devices.