September 16, 2004

Lessons of Experimental Physics

Posted by Arcane Gazebo at September 16, 2004 8:47 PM

This morning I came in to lab and looked at the previous night's data. And the data were pretty bad—certainly not usable for their intended purpose, and also bad in a way that pointed to a larger problem that had mysteriously developed over the past few weeks.

So I started changing things in the measurement, making it progressively simpler to reduce the space of potential sources for this problem. No matter what I took out, the oscillations in the data persisted, even after I was down just to the connections between the instruments, even after I was down to an electrical system isolated from the sample, even after I was down to a particular group of cables that were completely disconnected from the experiment.

In other words, I kept isolating the problem until I had it in the most improbable source I could think of. But all the other sources were now impossible. I simplified the wiring, and it worked. (Actually, it first caused a different problem, which I understood and fixed, and then it worked.)

So I don't understand what the problem was, and I don't know why my solution worked. But I fixed it!


Ain't that a kick in the head? I had one that I fought with all night and all I ended up changing was the number of scans... There's no reason it should work then, the S/N was fine to begin with... But it did. Right now, I am chalking mine up to the spec knowing it's master as my boss actually plugged in the scan increase as I was taking my Excedrin.

Posted by: Mylanda | September 17, 2004 3:44 PM

Just curious, what frequency range were the oscillations in?

Posted by: Lemming | September 18, 2004 2:47 PM

Somewhere between 9 and 10 MHz.

Posted by: Arcane Gazebo | September 18, 2004 11:11 PM
Post a comment