Chris A. Ciufo, Editor-in-Chief, Embedded Intel Solutions
Want thinner or longer USB or HDMI cables? Turning them from passive to active is the only way to go.
Most everyone is familiar with common cables for USB and HDMI. They’re commodities, and are available at big-box computer stores, drug stores and discount shops. USB cables are even sold at gas stations and quickie-type markets.
But these low-cost—and usually low-quality—cables can be pretty hard on high-speed signals. At rates from 0.5 Gbits/s (USB 2.0) to 10 Gbits/s (USB 3.1), signals are “fragile” and suffer from internal (caused by the cable) and external environmental sources. The best one can hope for is that things don’t work: that the USB device refuses to function or the display stays blank. The worst result is that systems behave unpredictably, and data files or important video frames are corrupted or incorrect.
These problems can all be alleviated using active cables with embedded integrated circuits (ICs). Making cables “active” is cheaper than making them higher quality by the combination of copper wire, shielding and higher-quality connectors needed to fix the deficiencies in low-cost cables. Ironically, active cables are actually better.
And active cables offer other, less obvious benefits, too. Besides being thinner (smaller gauge wire) and longer, they can be more flexible and easier to route due to the smaller wire, and as a result they’re less bulky and take up less space. Thinner cables also improve in-system (or behind the computer) airflow. Active cables are win-win, all around. Let’s take a closer look at them.
By the Book
Specifications from the USB Implementers Forum (USB-IF) dictate wire types and maximum cable length for USB 2.0 as shown in Table 1. USB 3.0 and 3.1 don’t actually specify cable lengths, because technology has improved dramatically in the ensuing years after USB 2.0. Instead, USB 3.x merely requires that cables meet the electrical specifications defined in the specs.
For USB 3.0, according to Wikipedia, a 26-gauge cable has a maximum length of 3 meters (9.8 feet). As we shall see, signal integrity degrades substantially in these passive cables. For the twice-as-fast USB 3.1 at 10 Gbps, the length is reduced by two-thirds to only a meter. These are the USB 3.1 cables seen around offices and on desktops. Ever wonder why they’re not longer to reach your scanner? Sure, putting a hub in the middle acts like an extension cord, but it also adds hardware and requires an AC plug-in. Longer would indeed be better.
USB 2.0 has four (4) wires, while USB 3.0 and USB 3.1 add four additional wires for a total of eight (8). USB 3.0 and 3.1 are also known as SuperSpeed; after USB 3.1 was introduced USB 3.0 became known as USB 3.1 Generation 1 while “regular” USB 3.1 is Generation 2.
Table 1: Speeds and cable lengths for USB and HDMI standards. Note that these are for passive cables without any embedded electronic amplifiers or timing devices. (Courtesy: Wikipedia and Pericom Semiconductor.)
Similarly for HDMI, there’s no maximum cable length specified, but the cable must meet required electrical specifications for Standard (Category 1) and High Speed (Category 2). In the HDMI 1.4 specification, Category 1 includes 720p and 1080i resolutions and also accommodates Ethernet. Category 2 cables include 1080p, 4K (4096 x 2160), UHD (3840 x 1260), 3D and other color depths. Category 2 cables can also transport Ethernet. HDMI cables consist of four (4) shielded twisted pairs, plus additional wires for other functions such as Ethernet, Audio Return Channel (ARC) and others.
In the case of USB 2.0, 3.0, 3.1 and HDMI 1.4, the specifications define analog and digital parameters for impedance, skew, cross-talk, attenuation, differential impedance, bit error rate (BER), jitter, insertion loss and more. All of these parameters dramatically influence the received signals’ ability to present a useful (per-spec) signal at the receiving end. The cable has a huge impact on these results.
If low cost cables didn’t work—for USB or HDMI—they wouldn’t be sold. But the issue is how thick is the cable or how short? If a user wants a longer or thinner cable, signal integrity challenges crop up almost immediately. If the lengths or wire gauges shown in Table 1 are violated, it’s likely the device at the receiving end won’t work properly.
Figure 1 shows a lab test set-up for USB 3.0 (and similarly for PCI Express 2.0) run at 5.0 Gb/s.
A signal generator creates and measures a Rosenberger Cable Eye Test Configuration with the pattern PRBS 2E10-1 at 5 Gbps (clocking at 2.5 GHz). A longer-than-specified 5-meter passive cable is used for the test with a differential impedance of 96 – 104 ohm, and an insertion loss of -10dB at 2.5 GHz. USB 3.0 should not work well with this too-long cable.
Figure 1: 2.5 GHz test set-up with a 5 m cable. (Courtesy: Pericom Semiconductor.)
The eye diagram for this test with a passive cable is shown in Figure 2. Firstly, note the barely open eye regions that indicate that this cable would perform poorly in the application. Secondly, the colors show that for lower BER (fewer errors per unit time) the eye areas are small. Stated another way, this 5 m cable would have very high BER and would likely be outside of accepted specifications. USB 3.0 at 5 Gbps wouldn’t work well, if at all.
Figure 2: This eye diagram for the set-up in Figure 1 shows high bit error rates (BER)
and small eye openings at 5.0 Gbps. This cable is too long, too thin, or both.
Convert to Active Cables
The solution to signal integrity challenges in USB and HDMI cables is very simple. Adding a redriver at the source (Tx) or the receiving (Rx) end boosts signals. A redriver is a special analog amplifier that conditions high frequency (and differential) signals to open eyes and solve signal integrity challenges. Some redrivers also apply line impedance matching, programmable gain, and add other analog parameters to compensate for line losses, impedance mismatch and channel (cable) effects.
An example of a USB 3.0 redriver applicable to the use case shown in Figure 1 is Pericom Semiconductor’s PI1EQX512A two channel 5.0 Gbps redriver, shown in Figure 3. Designed for USB 3.0, the device supports programmable equalization, de-emphasis, and output swing controls into two 100 ohm differential CML data I/O’s.
The redriver can be used at either the source end or the receiving end of a USB 3.0 cable. Figure 4 shows two set-up scenarios for the too-long 5m cable use case from Figure 1. In the left-hand image, the redriver is added at the receiving end, after the cable. In the right-hand image, the redriver is added at the transmitting end, prior to the cable.
Figure 3: Pericom Semiconductor’s PI1EQX512A redriver “fixes”
signal integrity challenges in USB 3.0, allowing longer and thinner cables.
Figure 4: Example uses for a USB 3.0 redriver. On the left, the redriver is at the
receiving end. On the right, the redriver is at the transmitting end.
The results of each are shown in Figure 5. In both cases, the redriver dramatically—almost “magically”—cleans up the USB 3.0 5 Gbps signals, opens the eye and allows an out-of-spec cable to be used.
The significance is that a simple redriver turns a passive cable into an active cable with new possibilities.
Figure 5: The eye diagram results from the set-up shown in Figure 4. Note that the redriver
at the receiving end (left) shows an extremely clean USB 3.0 signal at 2.5 GHz (5.0 Gbps).
Benefits of Active Cables
As stated above, active cables with redrivers offer many benefits over passive ones. Active cables can be longer than passive ones, exceeding recommended specifications because the redriver compensates for signal loss due to attenuation, jitter, timing skew, cross-talk, EMI and all manner of SI challenges.
Smaller wire gauge (numerically higher number) can be used and for a longer distance. Thinner gauge cables are more flexible and can be routed more easily, not only on a desk, but also in a rack, wiring closet or raceway. Thinner cables support a tighter bending radius, easing installation as well as opening the possibility of reducing the system chassis or rack size. Thinner, smaller cables also improve airflow, potentially avoiding the need for system cooling—or allowing existing systems to run at a lower temperature.
And last, but not least, active thinner, longer cables can be made more cheaply because the redriver allows using less costly wire.