datasheets.com EBN.com EDN.com EETimes.com Embedded.com PlanetAnalog.com TechOnline.com  
Events
UBM Tech
UBM Tech

Design Article

Comment


cmathas

6/19/2012 10:32 AM EDT

And I'm looking forward to publishing them here.

More...



bwf

6/19/2012 9:34 AM EDT

Hi All - Thanks for all the great comments. B&B does make an USB isolator, but ...

More...

How to avoid a Hindenburg-level USB meltdown in harsh industrial environments

Brian Foster, B&B Electronics Manufacturing Co.

6/13/2012 11:01 AM EDT

On a rainy day in 1937 the German airship Hindenburg arrived in Lakehurst, NJ, after a trans-oceanic flight.  The ship dropped its mooring ropes, and a few minutes later it went up in flames.  Theories about the fire include exotic tales of sabotage, but most scientists agree that the true culprit was an electrostatic discharge (ESD).  And they agree that the potential for disaster was designed into the Hindenburg right from the very beginning.

The problem was a matter of simple physics.  Hindenburg’s fabric outer cover was connected to the duralumin frame with non-conductive ramie cords.  The cords were coated with a light metal covering that was intended to improve conductivity, but it wasn't very effective. As a result, the outer skin and the duralumin frame were able to develop very large differences in electrical potential. And on its approach to Lakehurst the Hindenburg passed through a weather front with a high electrical charge.

Upon its arrival, still in the rain, the airship dropped its mooring ropes.  Within minutes the ropes were wet enough to ground the airship.  When they did, an electrostatic discharge jumped between the Hindenburg's cover and its duralumin frame. Theorists have argued about the ensuing fire, and whether it was initially fueled by a leaking hydrogen cell or flammable paint on the fabric cover. But they don’t disagree about the end result.  The Hindenburg was consumed in less than one minute. 

Like the Hindenburg, USB has potential trouble designed in.  This article will discuss the problem, and what can be done to remedy it.

The USB Specification

Developed to be a universal serial bus, hence the name, USB was intended to simplify and standardize the connection of computer peripherals and to eliminate much of the need for specialized expansion cards, which were installed by opening the computer case and physically adding new parts to the machine.  And it has proven to be a great improvement upon earlier data communications standards.  USB not only simplifies installation, it allows you to connect up to 127 devices to a single port. Devices are also hot swappable, and it will supply 5 VDC power downstream.  We’ve come a long way since the days of airships, however the laws of physics have remained the same, and USB can’t get circumvent them.

The Hindenburg Revisited
USB was designed for safe office and IT environments.  When it moves off the desktop into industrial applications, however, it’s subject to serious electrical problems.  They’re a function of its basic design.

Note that a standard USB cable contains four wires.  Two are meant to carry the data, and two carry 5 VDC power for downstream devices.  (See Figure. 1)  Whereas other types of industrial communication such as RS-485, can use differential signal transmission with no ground connection, the ground connection in a USB cable is unavoidable.  And it means that USB can easily transfer spikes, surges and ESD strikes to connected devices and computers.

Electrostatic discharges, for example, happen all the time. We’ve all experienced them--just walking on some carpet and touching the family dog’s nose is enough to give both of you a sudden surprise.  Like the Hindenburg’s outer cover, we change our electrical potentials every time we move around.  USB is deliberately designed to be hot swappable.  We’re expected to connect and disconnect USB cables over and over again, and there’s a chance of an ESD strike every time we do. Your trip across the carpet can generate ESD levels higher than 15 kV, and the typical level of ESD protection in an IC chip is usually going to be closer to 2 kV. 

Given that USB carries power, and has a connecting ground wire, there is an ever-present risk for ground loops.  That’s rarely an issue in consumer applications, where distances are short and connected devices normally share a common ground via a nearby wall outlet.  But industrial applications are likely to be much more complex.  Various pieces of equipment will be powered from entirely different building ground references.  A process control system and the front panel may be separated by hundreds of meters.  They may not even be in the same building.  When you connect a PC to the panel via USB, you may also be creating a ground loop path for a remote device with a very different building ground reference. 

USB cables aren’t helplessly vulnerable to electromagnetic interference (EMI).  They do use twisted pair wire on the data lines and there should be some copper braid and aluminum shielding.  (See Figure 1)  In desktop applications any EMI issues will be insignificant.  It’s a different story when a length of USB cable is exposed to the strong magnetic fields associated with industrial motors.  Powerful currents will be induced on the USB cable and the results may include anything from data loss and downtime to burned circuitry in connected devices.

Arcing
One of USB’s great advantages is simultaneously one of its weaknesses.  USB hubs provide 5 VDC power to downstream devices at up to 500 mA.  Current is allocated in units of 100 mA, and devices may draw up to 500 mA per port.  If the hub is bus powered, it can’t provide more than four 100 mA units of current to downstream devices, as it needs one unit for itself.  And it can’t have more than four downstream ports.  But if a hub has its own power supply, it can provide the full 500 mA to every port. (Computers normally have a powered USB hub built in.)

This ability to power gadgetry like cameras and external hard drives via a USB port is quite convenient.  But conditions can get rough in the industrial world.  Shock and vibration are common occurrences and USB cables can work themselves loose.  They can also be dislodged accidentally, when workers are moving things about.  When that happens, the same 5 VDC that is so useful in benign environments becomes a serious fire hazard.  If it arcs, and flammable gases or materials are present, you’ve got trouble.

Solutions--Range Extension
USB doesn’t have great range.  Even with hubs, the best you can do is 30 meters.  As cable is normally laid above, below and around working spaces, and as industrial working spaces can be very large indeed, 30 meters isn’t much.  Your network will most likely involve various kinds of extension and conversion, and some of them automatically resolve USB’s electrical issues while they’re extending its range.

Fiber optics, for example, not only give data communications incredible range, they eliminate the risk for grounds loops, power surges and EMI.  As data travels on a beam of light, rather than a copper core, fiber optic cable is immune to electrical issues of any kind.   Single-mode fiber has the greatest bandwidth, but is also more expensive than multi-mode.  And in both cases the cabling, transceivers and receivers are more expensive than the materials used in a copper cable installation.  A large part of any installation, fiber or copper, is actually going to be labor cost.  Fiber’s immunity to damaging electrical currents, and the protection it provides to connected devices, can more than make up the difference.

Solutions--Wireless
Wi-Fi is rapidly becoming another viable option for industrial installations.  Wireless networking used to be plagued by issues like multipath propagation, the phenomenon that occurs when transmitted signals bounce off intervening objects.  Different parts of the signal would arrive at the receiver at different times and out of sequence.  If multipath propagation got bad enough, early Wi-Fi devices couldn’t distinguish between the signal and the noise floor.  The current 802.11n Wi-Fi standard provides for the use of multiple input multiple output (MIMO) technology. 

MIMO devices anticipate multipath propagation and turn it into an advantage.  They deploy multiple antennas at both the transmitting and receiving sides of the wireless connection, and they split the data into numerous spatial streams.  The streams are transmitted through separate antennas and collected by corresponding antennas in the receiving devices, where onboard software uses signal-processing algorithms to correct and interpret incoming data. The new Wi-Fi devices also use such precoding and postcoding techniques as spatial beamforming, and the 802.11n standard adds frame aggregation to the MAC layer.  These new techniques give Wi-Fi the range and bandwidth it needed to become a reliable option for M2M data communications and industrial networking.  And, as is the case with a fiber optic link, a radio link creates a barrier that damaging electrical events cannot cross.





skyship007

6/13/2012 3:38 PM EDT

If you like Zeppelins or blimps try my Gasbags lighter than air comedy web site:
www.hybridblimp.net

Sign in to Reply



Dr DSP

6/13/2012 5:03 PM EDT

Wi-Fi is a great solution even if a Hindenburg-Level meltdown isn't a worry. As long as the Wi-Fi radiation won't spark my hydrogen filled bean bag chair all should be good.

Sign in to Reply



elektryk321

6/14/2012 4:27 AM EDT

Is there any prove that those photos show modules destroyed by "typical" electrostatic discharge in industrial area? I doubt, they look like were destroyed by lightning. Of course in every industrial facility should be installed surge protection devices and probably if is not installed correctly won't help in case of lightnig. But is there a sense to think about near ligthning at level of module? Protection against this events should be installed at facility level? Maybe we should also think about volcano eruption and earthshakes, they can also destroy USB devices.

Sign in to Reply



Brad@B&B

6/15/2012 3:13 PM EDT

I can understand the commotion over the photographs. They are from B&B’s collection of “customer disasters” sent to us over the years – and they are not all related to USB. I have seen many communication boards which show absolutely no sign of damage after a lightning strike but are dead all the same. Brian purposely wanted to illustrate the worst of the worst to drive home the point that isolation can be life saver.

Sign in to Reply



jonnydoin

6/18/2012 6:49 AM EDT

Brad, the third photo (BBUSB12.gif)shows burnout of an isolated barrier. One can see the 3 opto isolators and the HALO transformer. That is a 300mils barrier with isolated DC/DC, that should withstand at least 2.5KV of isolation.
This is no ESD or ground loop. This kind of damage can only be produced by lightning strike-class discharges.

- Jonny

Sign in to Reply



tsimm

6/14/2012 5:32 AM EDT

We must also consider the possibility of a very strong EMP developed when a nearby lightning strike occurs.A lightning strike produces a very strong magnetic field that can consequently cross all isolation barriers and induce high voltages directly on board components.Differing ground potentials can create very large voltage potentials between separate grounds, which is one reason the NEC stresses all grounds be bonded to a common point.It is common in industry to find multiple non-bonded grounds, such as steel I Beams and separately derived power supplies, and transformers without a bonded neutral.
The damages shown here appears to be high power occurrences,rather from a 5 volt source.
Of course, there is the possibility of a low-power spike causing a failure of the main power switch mode power supply control which would have sufficient capacity for this level of damage.(IMHO)

Sign in to Reply



Brad@B&B

6/15/2012 3:17 PM EDT

All good points and thanks for commenting on Brian's article.

Sign in to Reply



one_armed_bandit

6/14/2012 10:47 PM EDT

I *love* the smell of burning FR4 in the morning!!

Really cool effects there. Should mount those on the wall.

Looks like the third set (with RJ45) has at least one picture flipped in relation to the others - the top and side pictures don't match otherwise.

Sign in to Reply



cmathas

6/15/2012 2:09 AM EDT

Most people are satisfied with the smell of fresh coffee in the morning. Whatever wakes you up...

Sign in to Reply



Brad@B&B

6/15/2012 3:21 PM EDT

I believe Brian may have them mounted on his wall at home. Thanks for commenting:)

Sign in to Reply



macgregor54

6/15/2012 7:42 AM EDT

Great article. No kidding... from the teaser line in the email to the pix, well chosen words, precisely assembled, good content, excellent illustration, and while not exhaustive, explanations/exploration of factory-specific USB installation issues were accurately presented and sufficiently described. I never sign in for comments, but took the time to do so to make sure you got patted on the back for good work. Thanks.

Sign in to Reply



cmathas

6/15/2012 11:01 AM EDT

Wow, now that you've commented once, I expect you'll return often and continue to speak up. I am so glad you enjoyed the article. My goal is to post content that is both informative and enjoyable to read! Thanks for commenting.

Sign in to Reply



Brad@B&B

6/15/2012 3:24 PM EDT

Brian is busy traveling but certainly appreciates the compliments. He lives and breathes this stuff as do all of us at B&B.

Sign in to Reply



peralta_mike

6/15/2012 8:34 AM EDT

I have been dealing with this for about 6 years or so.

One of my solutions is to optically isolate near where the usb cable connects to the computer.

However, USB speeds are too fast for most optical isolators, so I use a USB to some kind of Digital I/O chip and then Optically Isolate at the output of the Digital I/O chip onto my equipment.


Sign in to Reply



cmathas

6/15/2012 11:01 AM EDT

Good solution - anyone else have a work-around like this?

Sign in to Reply



I_B_GREEN

6/18/2012 12:32 PM EDT

Why not add a hot swap feature to USB power.

Even just a resistive leader pin.

USB 3.14?
:)
Is it safe?

USB isolator dongle?
Does anybody make one?

Sign in to Reply



bwf

6/19/2012 9:34 AM EDT

Hi All - Thanks for all the great comments. B&B does make an USB isolator, but as was pointed out, the speed is limited to 12 Mbps. This is because isolation chip technology has is not capable of high speed data rates provided by USB 2.0 and USB 3.0. Not to worry though, we are actively researching this and I'm confident our engineers will solve this problem. During my travels, I saw many great applicatons of USB "in the field" and the response to this article shows that the technology is moving out of the office. I'm looking forward to writing some applications based articles to demonstrate the unlimited potential for Industrial USB.

Sign in to Reply



cmathas

6/19/2012 10:32 AM EDT

And I'm looking forward to publishing them here.

Sign in to Reply



Please sign in to post comment

Navigate to related information

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)