USB device failure

USB device failure overview

For nearly two decades we have been designing and manufacturing USB loopback plugs to be used in testing USB ports. Often our devices are used to validate prototype PC motherboards or used in mass manufacturing.

During this period we have seen customers accidentially blow up a number of our plugs (and other 3rd party USB devices as well). This page summarizes the common failure causes that we have come across. We hope that by documenting the failure modes we can help people avoid them.

Update: We have now released the USB S.O.S. short circuit tester. Using this device before connecting more expensive devices can avoid a lot of the issues detailed below.

Common failure modes

Case 1: Customer removed the USB plug from the protective plastic case and put the PCB into a grounded metal case. Making the device more likely to get fried. Which it eventually did. Solution: Leave the device in it's case.

Case 2: Customer was testing a newly designed custom USB hub in a computer monitor. They had one unit of their monitor which was wired incorrectly, which in turn fried one of our USB3 plugs during testing. Rather than investigate their monitor, they connected another plug. Which also got fried. They repeated this until they fried 7 plugs on the same monitor. Then contacted us but never admitted to the exact nature of the electrical fault, so we don't know the details. Moral: If you destroy one device, investigate the possible cause before blowing up the next one.

Case 3: We had a customer performing ESD testing. Deliberately shocking the plugs with static & they eventually failed. Solution: Don't deliberately shock your USB device and expect it to last.

Case 4: Customer was testing a point of sale system (PoS). The system had special 20Volt USB sockets. We suspect bad wiring on one of the sockets. e.g. 20V on the 5V lines. Was hard to confirm as we didn't have access to the system. Solution: Make sure your voltages are correct.

Case 5: Customer admitted they had equipment with a shorted lines (+5V line shorted on the USB3 data lines). The data lines run at a lower voltage (just 1.2V). So they eventually blew up. Solution: Check for shorts with the USB S.O.S. tester first.

Case 6: We recently discovered during ESD testing that we can blow up the main USB chip with static shocks to the LCD screen. It wasn't a failure mode we anticipated. We might cover the screen with the plastic case for hardware revision 3 to fix this. Solution: Take precautions against static.

Case 7: Customer was testing a new motherboard design. USB devices were already connected to the motherboard before the motherboard was powered on. Customer connected power to the motherboard via a multi-pin cable (not via a switch on the PSU). While connecting the power cable the 5V and 3.3V pins could sometimes make connection before the ground pin. Without a ground connection there is unregulated current flow from 5V circuits to 3.3V circuits. Including via USB devices (circuit flow from 5V+ to the lower voltage USB3 data lines). Which eventually destroys the USB device. Solution: Make sure there is always a good ground connection before connecting any USB device. None of the protection circuits can function without a ground connection.

Case 8: During testing of a mass produced motherboard the USB3 connector became worn and eventually made a bad electrical connection. The insertion cycle count on the connector is around 10,000, but the connector manufacturer measures this in ideal conditions and it only an average in any case. The USB3 connectors are all very small connectors and it only takes a small amount of rough treatment to bend it out of shape. Solution: Treat the connector gently. Another solution is to leave the cable connected to the plug permanently, as this can remove the insertions entirely. So there is no physical wear on the device connector.

Additional background information

Our USB3.0 Loopback plugs has a collection of added features to protect against incorrect voltages and static. See the bottom of the USB3 FAQ pagefor details. But nevertheless is it impossible to protect them against all possible external faults.

We use TVS (transient voltage suppressor) components on the board to protect the plug from some types of ESD/EFT events. But these (by design) are sacrificial components. So they degrade over time after repeatedly being shocked with a voltage level over their specs. The higher the voltage of the static shock, the fewer the number of shocks they protect against. For hardware revision 1 this level was between 6KV and 20KV depending on the line.

There are 4 wires (made up of 2 differential pairs) for USB3.0. The lines connect to the main Chip (a Micro-controller in the case of our USB3 plug) via an ESD protection suppressor chip. So failure tends to occur when you get high voltage ESD above the levels that can be suppressed by the protection, or when there is longer duration, lower voltage, short circuit.

In addition to the 4 wires for USB3 there are also 2 seperate wires for USB2. In some failure modes it is possible to blow up one or more of the USB3 lines, but still have the USB2 lines functioning correctly. From a user's point of view this can result in the device functioning at USB2 speeds instead of USB3 speeds (there are other causes for this as well, such as bad cabling or bad device drivers however).

The Electrostatic discharge (ESD) protection suppressor chip in use in our USB3 plugs is the best we could find, but it completely passes all voltages from 0V to 6V into the Micro-controller. The Micro-controller itself is rated for only 1.5V. So a longish duration short between 5V and one of the data lines will blow up part the input circuit of the Micro-controller.

The obvious design question is why isn't there better low voltage protection against shorts in USB3 devices? The answer is a bit complex. For a dead short (0 ohm) you get a lot of current flowing (USB Type-C even does up to 100W). It is near impossible to dissipate the power without something burning. Componets & circuits are available to detect a short but with 9 pins in a USB3 conenction there are lots of different permunations for bad wiring to protect against. At least 4 circuits are required, one per dataline. Adding this type of protection on the data lines can interfere with data transfer, as the high speeds lines are very sensitive to corruption at USB3 speeds. In short, there is no reasonably priced solution. The best solution would be to have a dumb device to detect shorts. (a dumb device that never needs to do high speed data transfers, e.g. a multi-meter).

Repairs of USB3.0 plugs

If the USB3.0 cable is bad, this is an easy fix, as standard off the shelf cables are used (excluding the Type-C cable, which are a bit special as they are rated as different wattages)

If the connector is bad / worn or the ESD protection circuits are dead the device can likely be repaired.

If the main chip (the Microcontroller) is dead it generally isn't cost effective to repair it. This component is the majority of the cost of the device and it requires specialized equipment and people to remove the chip and replace it (not to mention the shipping costs).