USB Versions Explained: Speeds, Connectors and Industrial USB

USB versions are the generations of the Universal Serial Bus standard, from USB 1.0 in 1996 at 12 Mbps to USB4 Version 2.0 at up to 80 Gbps, and each one sets the data rate, the connectors a port can use and, with USB Power Delivery, how much power it can supply. USB is a standardised interface that lets a computer power and communicate with peripheral devices, and it has replaced PS/2, parallel and many serial ports with one compact connector.

USB has grown well beyond keyboards and printers. On the latest ports a single USB-C cable can carry data, video, power and touch input to an industrial display, and USB now competes with Ethernet for cameras and instruments close to the computer. Its popularity in embedded and industrial computing over the last decade follows from that progress and from the automation of many industries, where more connected devices need to talk over different interfaces, sometimes with the signal converted or isolated.

This guide sets out every USB version with its speed, name, connectors and cable length, untangles the USB 3.0, 3.1 and 3.2 naming, covers power delivery and connector types, then looks at USB in industrial use: locking connectors, isolation, bandwidth planning, hubs and serial converters. It should help you answer common questions when selecting, converting, isolating or expanding USB in an embedded system:

• Which USB version does the data bandwidth of my application need?
• Is power available for charging or powering my connected device?
• Will anything interfere with my signals, or should I isolate them?
• Can USB endure my application if the environment is harsh and prone to failures?
• Do I need to upgrade legacy devices, or can I convert them?

The table lists every USB version with the year its specification was published, its maximum data rate, the name the USB Implementers Forum (USB-IF) now uses on packaging, the connectors it appears on and a typical maximum passive cable length. Longer runs are possible with active cables, extenders and optical cables.

Every newer version is backwards compatible with the older ones, providing the right connector or adapter is used: a USB 2.0 sensor works in a USB 3.2 port, at USB 2.0 speed. Our explainer on types of USB ports shows how to recognise each port on a computer.

USB 1.0 and 1.1

The original USB 1.0 standard was introduced in 1996 with a maximum speed of 12 Mbps, designed to connect simple devices such as a mouse and keyboard to one computer. Adoption was slow at first, and it was USB 1.1 in 1998 that fixed the early problems and started USB’s universal adoption. Both used the Type-A and Type-B connectors.

USB 2.0 (Hi-Speed USB)

USB 2.0, released in April 2000, raised the transfer rate to 480 Mbps, 40 times faster than USB 1.1. It kept full compatibility with earlier devices and set the power that every port still supplies: 5 V at 500 mA, or 2.5 W. It also brought Mini-A, Mini-B, Micro-A and Micro-B connectors for smaller devices, and supported passive cables of up to 5 m. USB 2.0 remains the most common interface for touch controllers, keyboards, barcode scanners and serial adapters, because 480 Mbps is plenty for them.

USB 3.2 Gen 1 (USB 3.0): SuperSpeed USB 5Gbps

The demand for higher bandwidth and more ports, driven in part by IIoT devices, led to USB 3.0 in 2008, now called USB 3.2 Gen 1 or USB 5Gbps. It raised the rate to 5 Gbps and added a separate pair of wires for each direction, so data flows to and from the device at the same time, a full-duplex link that USB 2.0 lacked. Ports supply 900 mA. Type-A and Type-B connectors gained extra contacts, and a wider Micro-B SuperSpeed connector appeared. Passive cables are typically limited to about 2–3 m.

USB 3.2 Gen 2 (USB 3.1): SuperSpeed USB 10Gbps

Formerly USB 3.1, USB 3.2 Gen 2 doubled the bandwidth to 10 Gbps in 2013, and the USB-C connector followed in 2014. Twice the speed matters for vision applications transferring high-quality uncompressed video for inference. The USB Power Delivery specification, used with USB-C, allows up to 100 W from a port with a suitable cable.

USB 3.2 Gen 2×2: SuperSpeed USB 20Gbps

Released in 2017, USB 3.2 Gen 2×2 runs two 10 Gbps lanes over the USB-C connector for up to 20 Gbps. It works only on USB-C and has been less widely adopted than USB4, though some industrial boards carry it on USB-C ports.

USB4 and USB4 Version 2.0

USB4, launched in 2019, runs at up to 40 Gbps on USB-C only and borrows its design from Thunderbolt 3: it can tunnel USB, DisplayPort video and optionally PCI Express over one cable. USB4 Version 2.0, published in 2022, doubles the rate to 80 Gbps, or 120 Gbps in one direction for displays. The cost is cable length: passive 40 Gbps cables are usually limited to about 0.8 m. Thunderbolt 4 and Thunderbolt 5 are closely aligned with USB4, and our explainer on Thunderbolt vs USB-C sets out the differences.

In industrial computing, USB4 now appears on compact computers and small form factor boards, where it drives extra displays and fast external storage. For harsh environments, USB 2.0 and the 5 and 10 Gbps versions remain the norm, because they work over longer cables, on Type-A connectors that can be locked and with a wider choice of rugged peripherals. USB4 suits OEM equipment where the cable stays short and protected. Our guide to USB4 covers it in more detail.

Much of the confusion around USB versions comes from the USB-IF renaming the 3.x generations twice, once in 2013 and again in 2017, and then moving to speed-based names on packaging in 2022. The same port can appear under three or four names on different datasheets.

When you read a datasheet, look for the data rate rather than the version number. A port described as USB 3.2 may be 5, 10 or 20 Gbps. Our explainer on USB 3.2 Gen 1 vs Gen 2 shows the difference in practice.

A key advantage of USB is the ability to power connected devices without a separate supply. That reduces cabling, simplifies installation, helps portability and can cut hidden costs. How much power is available depends on the port and on what the two ends negotiate.

In practice, many computer ports supply far less than the maximum. A 4-port PCIe card or a compact computer may share one power budget between all its ports, so check the total as well as the per-port figure before relying on USB to power cameras or instruments. Cards such as the Axiomtek AX92321 series let each port be set to 900 mA or 1500 mA and switched off in software, which allows a hung camera to be reset remotely.

Power and display can share the cable too. A USB-C port with DisplayPort Alt Mode can drive a monitor and power it, which is how single-cable USB-C monitors work; our explainer on USB-C explains which features a USB-C port may or may not carry.

USB connectors on embedded computers and I/O devices fall into two groups: standard and industrial. Standard connectors such as Type-A, Type-B and USB-C are small, light and easy to use, and where ingress protection and resistance to shock and vibration are not needed, they suit most applications.

Locking, latching and circular M12 USB

Industrial USB connectors are designed for rugged applications where a loose plug means a lost device. Industrial and military users need USB connections that stay put, and the usual answers are a screw-lock, a latch or a circular connector.

Rugged PCs often use standard USB ports with screw-lock fixings for locking industrial cables; the Neousys POC-751VTC in-vehicle computer, for example, has four screw-lock USB 3.2 Gen 2 ports. Industrial hubs use latching Type-A and Type-B designs with a positive lock tab that sets them apart from standard ports. Some latching designs are aimed at the connector retention requirements of Class I, Division 2 hazardous locations, and some are quoted to hold against around 50 N (about 11 lb) of pull. Industrial connectors are also typically rated for more insertion cycles, which helps where devices are connected and disconnected often.

M12 connectors are another industry standard. Their threaded coupling secures the connection against shock and vibration, and their strong outer shell resists damage and the ingress of dust and water. Some vehicle computers carry an A-coded M12 USB port alongside M12 Ethernet.

There are several reasons industrial and medical sites isolate USB. Industrial environments involve electrical noise and equipment powered from different supplies, and when a device and the computer sit at different earth potentials, current flows along the USB cable’s ground and can corrupt data or damage a port. A USB isolator breaks that path with galvanic isolation, passing the data across a barrier while keeping the two grounds apart.

USB isolators are typically rated for isolation voltages of about 2.5 kV to 5 kV. Check the speed as well: many isolators run only at Full Speed (12 Mbps), which suits sensors and serial adapters, while Hi-Speed models cost more and isolating 5 Gbps USB is less common. Some industrial USB hubs build isolation into each port.

Medical equipment raises the bar. IEC 60601-1 defines the means of protection between a patient and mains-powered equipment, and USB connections to patient-connected devices are often specified with two means of patient protection, which is tested at 4 kV AC. Our guide to what medical grade means explains how this applies to computers.

ESD, cable length and interference

Exposed USB ports on kiosks and machines take static discharges from users. Many industrial computers quote ESD protection on their ports to IEC 61000-4-2, often ±8 kV contact and ±15 kV air discharge, which is worth checking for public-facing equipment. Keep within the passive cable lengths in the table above, or use active or optical extenders for longer runs. And be aware that USB 3.x signalling can raise the noise floor in the 2.4 GHz band, so keep shielded cables and Wi-Fi or Bluetooth antennas apart, a point covered by our explainer on electromagnetic interference.

The speed printed by a port is shared. All the ports on one USB host controller, and every device behind a hub, share that controller’s bandwidth, so four 5 Gbps cameras on one controller will not each get 5 Gbps. As more devices are connected, the data rate per device falls, which is why an industrial USB 3 hub has more headroom than a USB 2 hub, and why it pays to decide what goes on the hub before you buy it.

For multi-camera vision, cards with several independent controllers keep each camera on its own path. The Neousys PCIe-USB381F spreads eight USB 3.1 Gen 1 ports across four controllers and sustains up to 400 MB/s per port with four cameras running. When you have used every port on an IIoT gateway, an industrial USB hub is the simpler expansion; for cameras, a card or a USB frame grabber is usually the better route. Our guide to PCIe bandwidth shows how much a card’s slot can carry.

The limits are distance, bandwidth sharing and, for some devices, driver support. Where a device sits more than a few metres from the computer, Ethernet is usually the better interface; our guide to Power over Ethernet covers powering cameras over longer runs.

Industrial automation and test. Industrial PCs with high USB channel counts control equipment, automate systems and run test stations, and an industrial USB 3.0 hub adds channels where more are needed. Typically installed in cabinets close to the data source, these systems carry both power and data to cameras and sensors that feed predictions and commands in robotic automation and production lines.

Kiosks and vending machines. OEMs integrate board-level computers inside kiosks and vending machines, and USB is the natural interface for local peripherals such as card readers, printers, NFC and RFID readers and keypads. Industrial motherboards and single board computers with larger form factors carry more USB channels, with extra ports on internal headers.

Security and smart city surveillance. Surveillance computers use current USB versions to connect cameras and sensors, and for outdoor installations both the computer and any rugged USB hub must withstand the weather. Our industrial NVRs are built for that duty.

Medical. Medical OEMs value USB’s plug and play, hot swapping and broad peripheral support, but must factor in isolation where devices connect to patients.

In-vehicle, military and unmanned systems. Vehicles and unmanned systems bring shock, vibration and wide temperature swings, so latching or M12 USB connectors are a must for reliable connections. Typical devices include GPS receivers, vision cameras, handhelds and night vision equipment, with high-bandwidth ports for AI vision. Our in-vehicle computers carry screw-lock or M12 USB for this reason.

Serial ports are a simple and reliable way to connect one device to another, transmitting data a bit at a time at modest speeds. Industrial PCs with many serial ports are common in factories, connecting sensors and control equipment, and industrial automation has relied on serial communication for decades. A big advantage of serial is simple integration: most operating systems support serial devices natively. Our explainers on serial ports and RS-232 vs RS-485 cover the standards.

Serial devices can still be connected to computers without a serial port. A USB to serial converter adds a COM port when it is plugged into an industrial computer, which can save the cost of replacing legacy sensors and controllers when a site is modernised. For vehicle and machine networks, USB adapters exist for CAN too, such as the Sintrones EMUC-B202 with two isolated CAN ports. Where many ports are needed, serial I/O modules or a computer with native COM ports avoid a stack of adapters.