Ultimate Screw-ups
Comment
Salio
I think the heading of this article should have been "Build in safety futures ...
one_armed_bandit
Jerry Weinberg tells a story where he was brought in to find a problem with an ...
It's never a good idea to maim your customer
Chuck Hill
10/14/2010 5:49 PM EDT
There was once a time when computers used something called a floppy disk for data storage, and those disks were 8 inches. These 8 -inch floppy disks used to come in single-sided and double-sided, single-density and double-density grades. Disks would be manufactured, individually tested and graded depending on the quality of the magnetic coating. This was typically done manually on a big production line.
I worked for a company that built an automated test system with four test heads, bins for incoming stock, and many bins for sorted disks. A robot shuttle whizzed up and down an 8-foot track with all the speed a 3kW motor could provide, moving disks from the feeder bins to the test heads and from the test heads to the sorting bins.
The day came for our first demo of the prototype system. Our customer (a VP of operations at the largest maker of floppy disks) came in to see a system busily moving disks to and fro with remarkable precision. He walked to the end of the machine and rested his hand on the end of the track. The static discharge that occurred glitched the system, turned the shuttle motor’s power amplifiers full on, and sent the shuttle hurling (you guessed it) in the direction of his hand. It slammed into the end of the rail with a horrifying bang
We all watched in silence for a full minute, as the color drained from his face. Eventually he removed his hand to see what was left of his digits. Fortunately, we had designed in a safety stop to prevent the shuttle from getting damaged in the event something like this occur. The little oil-filled piston gave its life, eventually stopping the shuttle with a space just smaller than the customer’s fingers. Our customer’s hand was bruised, but not seriously hurt.
He left, having experienced a wide range of emotions that day: initially impressed by our demo, horrified and then relieved at retaining all his digits, embarrassed by his role in the accident, and a bit unhappy that our product still had some serious flaws. We made a list of things to fix (a safety barrier, better grounding, and a much bigger safety stop) and went out to have many, many beers.
Just before completion, the project got cancelled. It seemed our customer managed to get the yield of the premium double-sided, double-density disks above 90% and no longer needed testing or sorting. At least that was the explanation. We hope it had nothing to do with the near maiming of a senior VP. That’s never a good idea.
Chuck Hill has 30 years design experience. He has a Master’s Degree in Engineering from Arizona State University, and a Master’s Degree in Business from DeVry University.
|
|
|
|
|
WireMan
10/15/2010 3:37 PM EDT
Mt brother had a similar experience while attending the National Computer Conference years ago. He walked up to a rack-size hard-drive system and touched the case. The resulting electrostatic discharge caused the hard-drive motor to stop spinning. When we went by the booth later that day technicians were still working on it. At least no one got hurt.
Sign in to Reply
WKetel
10/16/2010 9:08 PM EDT
This is an interesting case of what looks like inadequate connection between things that should all have been tied to some power common. It would indeed be very educational and possibly interesting to trace the exact mechanism of the failure
Sign in to Reply
agk
10/17/2010 7:10 AM EDT
Reading this article i recollect my college days. We made a lowpass and hugh pass filter and added a mic and a relay and simple logic to switch to bulbes from the output of the filter.The bulb connected to HPF named as woman and LPF as man. Most of the times during the exhibit of the project with voice from the students this was triggering properly . Except for one student. Whenever that student touches the microphone both the bulbs glow without uttering a single word. That student had a high static electricity always in her hands.Great natures gift and wonder !!!
Sign in to Reply
araasch
10/18/2010 4:44 PM EDT
Another thing that should be considered is a overspeed shutdown mode, assuming that in normal situations the drive motor is not expected to ran at maximum speed. This same type of test could have saved several lives in a radiation dosing machine which failed due to a software bug, thereby delivering a dose which was several orders in magnitude above the maximum therapedic dose (and therefore relatively easy to make a hardware test for). This basicly comes down to having a machine which has hardware sanity checks in it which cause it to go into a safe mode if something is detected that indicates a out-of-normal situation. Important whenever life or limb depends on the proper operation of a device.
Sign in to Reply
sharps_eng
10/20/2010 4:09 PM EDT
Too often a designer concentrates on the main performance arc, in this case the motor drive end-to-end. The safety features are then bolted on afterwards to constrain the motion to credible limits. Safer method is to define the safe envelope first, then shrink it until the remaining degrees of freedom allow only the arc actually needed for the function.
Sign in to Reply
kalpak
10/21/2010 12:56 AM EDT
"Member of upper management" syndrome strikes again! Next time, use your company's VP/ Chairman/ CEO for finding such bugs.
Better still use your CFO. After such a "demo", the CFO will never say no to a budget for test equipment.
Sign in to Reply
one_armed_bandit
10/25/2010 9:58 PM EDT
Jerry Weinberg tells a story where he was brought in to find a problem with an IBM mainframe nobody else could track down. The mainframe would glitch or reset at odd times.
He started watching. He noticed the problem occurred when an operator touched the keyboard, but only when she walked up and first sat down, and the first thing she touched was the keyboard.
She was wearing a skirt and nylons - conforming to the dress code. The dress code was modified for female operators.
Sign in to Reply
Salio
11/3/2010 1:03 AM EDT
I think the heading of this article should have been "Build in safety futures prior to customer demo". I mean how come someone from the plant didn't do what this VP had done? This could have been avoided had someone actually poked around in the system. The embarassment could have been prevented.
Sign in to Reply