Thursday, July 22, 2010

On the Importance of Intermediate Testing

I replaced a sensor in my machinery on Monday. It was quite an involved process that required tearing apart a good chunk of the equipment, because the sensor was integral to the function of the equipment. Both the hardware and the wiring had to come out.

I put in the new sensor, and laboriously hooked up all the wiring the same way the old sensor was wired. It was the same manufacturer, same product, just a newer model.

And of course, because this is a precision machine the screws are tiny and I kept dropping them, and the wiring was on the back panel of my electronics box, and I couldn't move the electronics box because OTHER stuff was in the way, and the tiny screwdriver was missing again, and WHO BUILT THIS THING? Oh, yes, I did.

Deep breath.

I finally had everything back together, and I fired up the laptop and ran a test program to collect some data and see how everything was working.

Um, no signal from the sensor.

I took another look at the wiring, and at the specification sheet, and oh guess what? You know how every sensor I have from this company has red, black, green, and white wiring? Well this new sensor has red, black, green, white, AND YELLOW. And you know what? They DON'T EVEN USE green and white anymore. Yeah. Now they just use the yellow. Just to mess with you.

So back into the electronics box I go, armed with the tiny screwdriver which I found hiding under the spec sheet. Which I should have read first, before I began.

Properly wired up, I ran the test program again. Now I get a signal, but the signal doesn't move or change. Not good.

So I whip out the oscilloscope, a measurement device which displays and graphs electrical signals, and I hook it directly to the output of the sensor.

Figure: This is our oscilloscope. We're old school like that. No really, it's very old.

I jiggle things around and mess with the sensor, but still no change in the signal.

Tech support, here I come!

Well, actually, to home I went. I gave up for the day and decided to sleep on it.

Tuesday bright and early I tried things again, double checking everything. Still having problems. I ring up tech support for the sensor company - oh wait, that's right, they are in California, they don't have engineers in until 9am which is noon here.

That's okay - break for lunch!

The end of this story is that the sensor isn't working properly even though I did everything correctly. I had to take apart the equipment AGAIN to get the sensor out, and then I boxed it up and shipped it back (still under warranty).

But I could have saved myself a day and a half if I had just hooked the new sensor up to the oscilloscope BEFORE I put it in the machine. So the lesson learned is, test test test, always test. Test everything you can, as early as you can. It never hurts to have more data, especially if it doesn't take long to collect. Then I would have known immediately the sensor wasn't working.

Figure: Should have tested that before they set it in stone. At least mine was under warranty.
(picture from failblog)

No comments:

Post a Comment