"Wall Warts" (the colorful name for our beloved device Power Supplies) may be a thing of the past for some consumer electronics devices thanks to the ubiquitous and versatile Universal Serial Bus (USB) port on your Desktop or Notebook PC. Traditionally, power hungry devices that connect over the USB cable also required a Wall-Wart for power. The market (all of us consumers), however, is demanding that we jettison Wall-Warts, and that USB connected devices use the power sourced through the USB cable to supply the power needed by our favorite mobile devices. The truth is, there are several design considerations that must be taken into account when designing a consumer electronics device needing to operate without a Wall Wart.
Many people don’t realize that the USB cable attached to their PC is actually a power cable as well. In fact, in any given USB cable, there are only 4 wires: 2 for representing and carrying the data, one for power, and one for ground (see Figure 1)
Figure 1. Cross Section: White & Green carry Data, Black for Ground, Red for Power (VBUS).
Click to Enlarge
Because of this fact, devices are able to use this power connection as a means of receiving the power necessary for proper operation. The key, however is that the USB bus is strictly regulated to ensure proper power management. To use a real word example: You get out of the shower and towel dry. You pick up your hair dryer and turn it on. Normally, it works fine…until you try to turn on the curling iron at the same time. Then, the lights go out. Why? Every system that supplies power must incorporate some kind of power limits. In your house, it’s the circuit breakers between you and the power company. In a system connecting a peripheral device to your PC, it’s the USB bus power specifications. You don’t want your computer to go dark when you plug in a mouse or pair of speakers, do you? As a way of controlling potential misuses of power over the USB bus, power is limited by the specification and governed by software.
So what does a designer need to understand to take this into consideration?
According to the USB specification, USB devices can either be “Bus-powered”, powered through the USB cable, or “Self-powered”, powered by a battery or plugged into the wall (the Wall Wart!). In addition, USB has defined two potential power levels available to Bus-powered devices through a USB port: High Power Ports, and Low Power Ports. High power ports are capable of supplying 500mA to a downstream device. Low power ports can only supply 100mA to a downstream device. Why are there two different levels? The original intent of low-power ports was to enable bus-powered hubs. These devices draw 500mA from the upstream port, use 100mA internally, and distribute 100mA to each downstream port (See Figure 2).
Figure 2. Power configuration provided by Host PC with Hub providing port expansion.
Click to Enlarge
So how does power control work in USB?
The very first power regulation a USB connected device encounters is the “enumeration” power consumption limit. When a device initially connects to a USB port, it must limit its current usage to 100mA so that it can identify itself to the host (enumerate) on either a high-power or low-power port. Provided the device connects at less than the 100mA, the host reads one or more configurations from the device using the GET_CONFIGURATION command. In the world of USB, each device will respond to the host with a listing of its required configurations. Generally, a USB device has three possibilities:
1. The device has only one configuration for low-power
2. The device has only one configuration for high-power
3. The device has two or more configurations with a mix of high and low-power
In case 1, the host’s decision is easy: issue the SET_CONFIGURATION command, confirming proper power configuration, to turn on the device. In case 2, the host will only issue the SET_CONFIGURATION command if the port is a high-power port. If it is not a high-power port, the host will notify the user with a message (See Graphic 1)
Graphic 1. Port error by Windows XP when High Power port is unavailable.
Click to Enlarge
In case 3, the host picks the configuration that is appropriate for the port’s power level -- provided there are both High and Low power ports available. Once the port configuration has been negotiated and properly configured, it is up to the device to now behave properly on the port. If it does, the device is considered “USB compliant”. If it doesn’t, the device is “non-compliant” and can either cause system error issues or may not work properly at all under certain circumstances.
So why is this problem so difficult to solve?
Bus-powered USB devices have actually been around since the beginning of USB. However, when the USB 2.0 specification added High-Speed (480Mbps) data transfer, no USB silicon vendors could make a high-speed USB controller chip that would enumerate under the critical 100mA limit.
In other words, when peripherals incorporating the new High-Speed USB devices attempted to connect, they exceeded the limit of 100mA and were not properly configured. Initially, this was not a huge problem since most of the early USB 2.0 devices were mass storage (external CD burners and external hard drives) which suffer from a separate power issue unique to rotating media (discussed below). As a result, bus power operation of high power devices was rarely a consideration. The times, however, have changed. In June 2004, Cypress Semiconductor introduced the first true high-speed USB 2.0 silicon that enabled bus-powered devices. The EZ-USB FX2LP chip draws 50mA in typical operation, leaving 50mA for the other pieces of the USB system even before the SET_CONFIGURATION message is received…the beginning of the demise of the Wall Wart?...
…Not so fast. Just when you think you have a start to solving the Wall Wart problem, unique power issues must be considered for some applications. For example, some of today’s most popular USB mass-storage devices use rotating media to store data. In CD-ROMs and DVDs that connect over USB, the media is the familiar shiny plastic silver disc. In the case of external hard drives like those made by Seagate, Maxtor, or Western Digital, the media is polished aluminum discs. For each of these devices, the media is rotated until the desired area is under the read or write head and then the electronics access the data. For USB power purposes, there are two main problem areas: the motor and the head. Both are issues for the same reason: they consume more power than USB allows – especially as they begin their start up process. In fact, Figure 3 shows a typical power profile for the startup sequence for a disc drive product. Note the problem area where, as the disc spins up, the drive consumes too much power.
Figure 3. Disc Drive power consumption curve.
Click to Enlarge
Several companies have successfully solved the puzzle of building a bus-powered device with rotating media. The first of such products was introduced by Iomega. Iomega’s strategy was to re-engineer the entire drive mechanism so that it reduced the peak power used by the device. For example, they did not spin up the drive as fast as possible, but opted to limit the power used during spin-up. This sacrificed performance, but the bus-power feature made the device very portable and easy to use.
In-System Design took a different approach creating their bus-powered hard disk. They used an off the shelf disk drive and engineered a clever solution to limit the power drawn from USB. When the disk drive was idle, they stored power in a set of eight AA batteries. When the disk was in use, they drew down the batteries to power the drive. This approach had the advantage of allowing the use of any drive in the design, not just a specifically designed USB drive.
Graphic 2. ISD-315 bus powered hard drive.
Click to Enlarge
Additionally, as more drive manufacturers become aware of the desire to develop bus powered applications, and as more consumer product companies become USB spec savvy, they are employing creative solutions to solving this problem. In fact, Apricorn introduced the first USB compliant High Speed USB 2.0 Bus Powered Hard Drive by using a 1.8-inch Hitachi hard drive and a Cypress Semiconductor low power USB chip. The combination of a lower power drive, with slower spin rates along with a power saving USB device, allowed the company to introduce true “Bus-power” 20GB and 40GB Hard Drives. As the USB external storage market continues to grow and drive power continues to shrink, it seems like it’s only a matter of time before your external hard drive backup system and your external DVD drive lose their wall wart.
Even the most common “bus-powered” USB 2.0 devices today suffer potential non-compliance because of power. These are the ubiquitous “thumb drives” that combine a USB controller with one or two NAND flash chips in a package the size of a stick of gum. The USB NAND market has exploded from next to nothing in 2000 to 63 million units shipped in 2004. All USB thumb drives appear to be bus powered, but many will not work properly in USB 2.0. The following graph shows laboratory measurements of commercially available “thumb drives”. Many of these devices are well above the limit for a low-power port. Some of these devices are subject to failure the moment they are plugged into a USB hub (with low power downstream ports). The wise designer will anticipate power issues and make device choices that allow them to be USB compliant.
Figure 4. Thumb Drive operation above 100mA shows probability of choosing non-compliant device.
Click to Enlarge
Reset and power issues for bus-powered devices
Bus-powered devices don’t have the luxury of a stable wall-power supply or battery. As a result, designing them for fail-safe operation can be challenging. One area of particular concern is when power is quickly cycled on and off of a device (as is the case when you unplug and then re-plug a USB device quickly). According to the spec, bus powered devices can be hot-plugged and hot unplugged at any time. This means that VBUS (USB power) can be removed and reappear with very little delay when the user power cycles the device by pulling the plug. Unfortunately, this unplug and re-plug can confuse the system unless properly designed for.
In Figure 5 below, we will describe the power down sequence that can create system confusion. Trace 1 shows the RC RESET# line and 3.3v line during an unplug/re-plug event. VBUS (Trace 1) starts to drop as soon as the device is unplugged. Over 100ms later, the 3.3v line (Trace 2, red trace) begins to drop. The RESET# line (Trace 2, purple) tracks the 3.3v line’s drop towards ground. The unit under test is below minimum operating voltage at 300ms after the unplug, but the RESET# line is not below the Vil (TTL Input Level for Logic Low) threshold until almost 2 seconds after the unplug. Unfortunately, a Logic Low MUST be reached on the RESET# line in order for the system to recognize a bus reset. If you re-plug the device in the area of the curve depicted below as the 2 second Danger Zone, there is a very good chance that the Host PC will become confused as the device was not properly reset, and it will not recognize the re-plug.
Figure 5. Bus Power reset issues.
Click to Enlarge
Unfortunately, self-powered hubs will create even shorter pulses on VBUS when they are plugged into a host or when the host resets them – so if your product is plugged into a USB Hub, your design had better be prepared to contend with the short reset pulse. Reset sounds like an infrequent event until you hear that your product doesn’t work after the PC reboots!
The best solution to these problems is to use an external POR (Power On Reset) chip. If economic constraints force your design to use an RC reset circuit, however, follow these hints:
- Limit large capacitors on your board. This will allow VBUS and 3.3v to drop more quickly when your board is unplugged.
- Add bleed resistors so your board is at the 500μA suspend current limit.
- Delay suspend until the 10mSec limit.
RC reset circuits may be safely used in some self powered designs. Designers should make sure that the reset properly holds RESET# below Vil (800mV) for 5ms after Vcc has risen enough to supply Vcc(min) to your chips. Test your reset circuit in the following conditions:
- Cold power-up, plugged into USB
- Cold power-up, unplugged from USB
- Hibernate/resume, plugged into USB
- Power cycle, plugged into USB
- Power cycle, unplugged from USB
- Power cycle, plugged into 5 tiers of hubs (connect 5 hubs together and plug into the furthest one from the host)
- Unplug/replug the 5 tiers of hubs
- Repeat the above two tests with one tier of hub.
So Wall Warts may not be gone forever any time soon, but there sure are pressures from the mobile computing consumer base to get rid of as many as possible. Unfortunately, there are many consumer electronics companies not savvy enough (or creative enough) to solve the problem AND remain USB compliant. USB compliance assures end users are not confused, frustrated, or worse, damaging their PC or the peripheral device due to poor power management. It is the wise designer that considers the power constraints of USB, designs to meet the specification, and recognizes that the end market will reward them with greater sales due to reliable, effective, useful bus powered components. So get creative – and get rid of those Wall Warts!