Stress test failure confirmation

[NetVoikGuy]

Hi,<p>New here, thought I'd ask a simple question.<p>I have built a circuit based around the 555's and
other items. I am trying to make it fail. In doing so, I need to know how long the circuit lasted, until failure. <p>I can't sit next to the test bed for days with a stop watch. Any ideas?<p>In other words, once fired up, I need to know total time until failure. I can tie to any part of the circuit, and the "timing piece" can be outside of the environment ... but the trick is for my "outside" clock to stop when the circuit fails, but retain the display, so I know when the failure occured. I can , of course check each day, an mark the days as they pass.<p>tommy

[jwax]

Simplest- have your circuit power a relay. When it fails, the relay opens, closing contacts that start a timer. Your "time since failure" clock has begun timing.

[haklesup]

How you detect a failure depends on the failure mode you are expecting. If you expect the device to develop a short or open on the supply you would do it differently that if you expected an input to become leaky (parametric failure detected by characterization or high speed functional test) or an output to stick (gross functional failure)<p>Assuming you are looking at a change in supply current as the indicator you have to consider detecting an increase (short) or decrease (open). Don't forget shorts often lead to opens a few moments later. <p>A current loop transformer normally has no output with a DC signal running through but an abrupt cahnge in supply current will cause a blip on the output which is detectable.<p>A series resistor on the supply wire can be used to create a voltage drop that can be watched with a comparitor which activates the timer.<p>I can think of more but it is just guessing without knowing what kind of input the clock needs to stop and run. A mechanical electric clock would be easiest but you could use a digital clock too if you could inhibit the count somehow.<p>Is your circuit under stress (temperature, humidity, vibration, chemical exposure etc.) if not you might be waiting years to failure.<p>Reliability of any circuit can be calculated if you know the reliability of all the components used but in short is will probably end up being equal to the shortest MTBF component (mean time between failure).<p>When semiconductor companies do High Temp Accelerated Stress(LIfe) Test (HALT, HAST) it is done at 80 C which considerably shortens the life of a device. The relaibility engineer then uses calculations to determine the life at room temp.

[Edd]

Quote:
I can't sit next to the test bed for days with a stop watch. Any ideas?<p>Suuure you can…. you just need to know the proper technique !<p>Considering that you have mastered the manner of detection of the failure. With that
fact being supplicated with an accompanying activation of a relay making a SPST contact closure.
What you will be utilizing for your timer element will be a cheeep solid state design of wristwatch.(Rolexes need not apply…….unless this is government contract work)
Not too cheap though…..as it also needs the date display capability as well as the time.
The internal timebase for the watch will use a micro sized 32,768~ crystal cylinder.
A plastic watch housing is even an advantage, as an entry through a micro hole will be required to pass thru two fine enamel/formvar insulated wires [~30 ga) to be micro soldered across the terminal pads of the previously mentioned crystal. Once outside the watch housing the leads can be terminated to a beefier terminal pad and a bit larger wires utilized. The watch is then used in the close proximity of the relay
apparatus.
The correct date/time of the watch is verified then the start of the test operation time is logged nearby. As time passes the watch keeps time……until the relay makes closure, and at that time the watch will go into sleep mode with no display, also the watches time keeping function will go into sleep mode . If the closed relay contacts are then opened the watch will come back to life and show the time that it was disrupted.
Now if you don’t experience a MTBF rate in excess of 2-3 yrs…[battery life].I believe that you might be covered.
This is a technique that I came up with for a private investigator “ Peter Gunn” type of acquaintance of mine. He often needs to be made aware of “events” time of occurrences at odd/extended times.<p>
73's de Edd
[email protected] ..….....(Interstellar~~~~Warp~~~Speed)
[email protected]….......(Firewalled-Spam*Cookies*Crumbs)

[jwax]

Interesting, Edd! So by shorting out the crystal, the timing stops, but the watch holds the time it was stopped at? That's kinda cool! Thanks!

[philba]

I'd probably use a PC to monitor the circuit under test. The parallel port can be used to read values or you can get USB or serial based interfaces. You might have to write a little code but VB is pretty easy. I've done this to monitor battery drain with some hw I cobbled together (pic with ADC and serial). The vb program just read values sent by the pic, formated them in decimal and saved them in a file. I then used excel to chart the battery drain.<p>A similar approach is if you have DMM with a serial interface - they come with software for logging the readings.

[jwax]

Use a PC to monitor an event that could just as well be monitored by a $5 watch? Or a relay and timer? Must have a lot of PC's around, or just keeping the power company busy! KISS works best.

[philba]

Sure, if all you want is the point where it failed. <p>However, the PC allows for more readings and more detailed measurements. I particularly liked the ability to measure battery voltage curve. Current draw is also a very nice thing to know. As a general test monitoring tool, the PC rules. Maybe I'm nuts but if I'm going to do a failure test, I'd like to get as much information as possible.