Going against the grain dust
Solving the mystery of a locked-up microcontroller-turned safety feature provides a challenge for one company owner.
Douglas Forst, CMC Industrial Electronics Inc -- EDN, August 25, 2011
As the owner of an industrial-controls company, I
frequently receive calls from customers. Recently, a
representative from our best customer called, asking
for help with a sensor on a monitoring system. The
company happens to be in the same city as ours, so I
was able to visit the site. Our company manufactures
an intrinsically safe monitoring system for conveyors and other
equipment in hazardous locations and other industrial and commercial
facilities. The principle of intrinsic safety is to restrict the
amount of available energy in the field wiring system to less than
the amount that could cause an explosion or a fire.This customer operates a grain elevator that has a high risk of fire and explosion from grain dust. Grain-dust explosions are responsible for a number of deaths each year worldwide. The customer’s large installation uses approximately 2500 sensors on the network. Our system is designed to detect problems, such as hot bearings or off-track belts, with the conveying equipment. One of the customer’s speed sensors was on a long belt, but the sensor was frequently failing after nearly a year in use.
When I reached the site, the electrician
had already done most of what I
would normally have done. Our speed
sensor uses a magnetic code wheel triggering
Hall-effect sensors to determine
the speed of the belt. The customer’s sensor
would work fine until the code wheel
began to turn. Then, for no apparent reason,
the sensor stopped operating. The
design includes a working and verified
watchdog; whatever was interfering with
the unit was strong enough to override
all of the protection. The sensor would
resume normal operation if an operator
powered it off and back on again.The electrician had more than once replaced the sensor, the controller communicating with the sensor, and even the junction box it connected to. Two temperature sensors that plugged into the same box worked fine. I repeated most of the electrician’s work and even moved the sensor to another belt, where it worked perfectly.
I had heard a strange snapping sound
when the conveyor was running and wanted
to see what it was. The electrician said it
was just grain falling or the belt splice, but
I wasn’t so sure. We started the belt, and,
sure enough, as the belt reached speed, you
could hear a loud, repeated snapping sound.
The interval between snaps changed as the
belt reached speed. The belt ran for about a
minute before the control system stopped it
due to an underspeed condition; the sensor
had failed once again. As the belt slowed
down, the snapping interval grew longer
until the belt, and the snapping, stopped.We again restarted the belt to determine the source of the noise: Inside the rigging of the conveyor, a half-inch-long, white-hot arc was jumping from a return roller to the frame of the conveyor and landing just a quarter-inch away from a pile of highly flammable grain dust. The roller and the belting material were rubbing together, creating static electricity and causing the arc, which could have destroyed a multimillion-dollar facility and killed or injured many people.
The electrician then remembered that this roller had been replaced just before the problem started. This roller was from a new supplier; it had thermoplastic end caps connecting the metal roller to the bearing shaft, effectively insulating the metal roller from the conveyor frame. The new roller had created the perfect Van de Graaff generator. All previous rollers had been of an all-metal construction that effectively grounded any static charge. Using some Category 5 cable, we rigged up a simple grounding wiper and restarted the conveyor. Everything worked perfectly. How the static discharge was getting into the sensor almost 10 feet away, I will never know.
Douglas Forst is president of CMC Industrial Electronics Inc (Burnaby, BC, Canada).
Talkback
-
Had a disaster resulted, root cause probably never would have been determined. The problem was elusive in the operational conveyor and the evidence probably would be destroyed in an explosion/fire. The "win" is far bigger than fixing the sensor.
kenish - 2011-7-9 17:04:16 PDT -
The key lesson I learn from this excellent story is that you shouldn't trust anyone to resolve a question in troubleshooting, you have to resolve it yourself. The electrican said the snapping noise was just falling grain, but in fact it was something completely different and digging further solved the problem.
Richard Hewitt - 2011-7-9 15:19:50 PDT -
Changing the belt to a new manufacturer in an explosion sensitive (life threatening) situation is just bad - bad engineering and safety control. Having no electric field sensors in an area full of dust and rotating machinery, and then testing the system with the dust running is also foolhardy.
When new electronics fail mysteriously it points to electrical overstress (EOS). This can be from conducted electrical noise, or picked up by electrical fields generated from such discharges. The designer of the sensor obviously did not add EMC protection, and probably shouldn't have expected to (though its a wise thing).
This whole article was frightening. The whole thing could have easily blown up. Everyone from the person who authorized the belt change without qualification, and everyone troubleshooting the active system should be fired for gross incompetence and negligence.
Bret W - 2011-7-9 14:21:20 PDT -
Classic plate-discharge effect. You can knock out any unprotected circuit from a distance by discharging into a floating (or grounded) plate. An ESD spark causes a sharp current pulse to travel down the plate (the plate being, in this case, the conveyer frame), generating a strong e-m field. The sensor's circuitry picks up this field just like any other antenna, turning the e-m field back into a current pulse. The most susceptible line is almost always the reset line, and the spike is often short enough to generate a "runt" reset condition, where only part of the board resets and the rest just hangs. It takes a power-cycle to recover. Had the sensor been properly ESD hardened, the first indication of a problem might very well have been a grain explosion.
Marc Stewart - 2011-2-9 00:12:03 PDT -
Excellent case study and troubleshooting. Thanks for the article and for sharing your insight. I hope this was reported by the plant electrician and adapted it to their maintenance MOC and safety program.
Jeffrey Nichols - 2011-31-8 07:41:48 PDT
