Mitt liv som Tobin

I spent Thanksgiving-Eve at the lab. There had been a power outage the day before, and it's always a huge chore to get things running again. A 20-second power outage routinely requires 48 hours of recovery time; this happens a couple times a year.

I went in around 6 pm and was very pleased to, by midnight, diagnose and correct a showstopper problem that had so far eluded solution, setting the demodulation phase on a wavefront sensor. But then I encountered more mundane problems, giving up with the detector still inoperable. Well, it was operating just fine, but our low-noise readout system wasn't. So, no data for Science. I got home a little after 5 am.

Woke up after noon on Thanksgiving day, checked the lab's electronic lab notebook. In the morning other scientists were able to get the remaining issues cleared up. Teamwork, I guess.

You can spend an infinite amount of time just keeping the thing running. That leaves no time for Thesis.

One of our current efforts at LIGO is to increase the laser power.

The sensitivity of our gravitational wave detector is ultimately limited by the granularity of light*. The light coming out of our detector, like any light, is packaged up in little bundles of energy, photons. In any given length of time, the number of photons coming out of the detector is an integer. We can't measure half a photon. It's the same problem as doing a survey where you don't survey enough people. If you ask only ten people who's going to win the election, your survey results can never be better than 10%. The solution is to survey more people. To measure more photons. As we put more laser power into our machine, the granularity of light matters less and less.

The noise caused by the granularity of light is called "shot noise." The name sticks because it makes us think of shot, like the light is a spray of BB's out of a shotgun. But it's really named after Walter Schottky.

We've replaced our ~ 8 Watt laser with a new 35 Watt one. But putting more light into the machine is not so simple as getting a more powerful laser, for at least two reasons.

* at frequencies above ~ 100 Hz

( 1. Things heat up )

( 2. Light pushes on things )

On Friday night we put 12 W into the machine for 12 minutes.

Late nights at the lab. We barbecued chicken (yes, from Whole Foods—it's a vice I'm cultivating) for dinner.

At dinner, talking about the current economic meltdown, Valera took a drag on his cigarette and smiled.

"Oh, I've been through a crash before..."

He's Russian.

Work at the lab lately has been quite satisfying. Working with Stefan and Hartmut, our current visitors, has been quite a pleasure. I feel more confident, and have had more time to think about things on my own. Also, we've made a lot of progress.

Part of today's excitement was in replacing a photodiode.



Our gravitational wave detector consists mainly of vacuum, enclosed by a steel tube and bounded by a collection of mirrors, and full of photons. The photons, little packets of light, carry all of the information. When the photons come out of the machine, we detect them with photodiodes, which turn streams of photons into streams of electrons, which feed through electrical circuits.

We use many photodiodes, several at each place where light comes out of the machine. A photodiode itself is a tiny piece of semiconductor, carefully mounted and connected via tiny wires to some bigger wires that poke out of its little can, and connect the photodiode to a collection of electronic circuits who are sensitive to variations in the intensity of the light at radio frequency, who gather this information, and send it out over a wire to more electronics, who feed the signal into computers, and servo systems, and data analysis algorithms. Servo systems that hold all our mirrors in exactly the right places. Data analysis algorithms that will tell us when we see a gravitational wave.

I don't know how much our photodiodes cost, or where we get them, but it is clear that they are very precious.



One of our photodiodes went bad recently, and today we replaced it. The new photodiode came as all new photodiodes do, mounted in a little metal can with a glass cover on top.

The glass cover protects the delicate photodiode and its delicate wires. But for us it is a problem; the small amount of light that gets reflected by the glass window instead of going through it corrupts our measurements. We have to take off the window, exposing the delicate photodiode.

This is done with a tool known as the "can opener," because it is used for opening photodiode cans. It is from ThorLabs, an company that makes scientific optical equipment, and as far as I can tell, it is specifically for this purpose.

I strongly appreciated Stefan's tutelage in this matter; he described how to perform the delicate operation, and then promptly left the room, leaving me to perform it. Trust leads to confidence. He's soloed an airplane; maybe the philosophy comes from there.

With the new photodiode, our gravitational wave detector is able to see 8.5 megaparsecs into space.

There is something liberating about being at the bottom of the work-heirarchy, something refreshing about the total singleness of purpose, the ability to work full-bore while the machinations of the organization complete seemingly useless 9-5 cycles of bureaucracy.

I camped out at the lab for the first time on Sunday night. At 5am the interferometer wasn't working and I decided to crawl into one of the 'sleeping rooms' to snooze before the 10am staff meeting. Only it turned out that the sleeping rooms (established during Katrina) had been turned into offices in the meantime. I camped out on the floor of Rana's office but got too cold and slept fitfully. I gave up and did some more work. Went to the staff meeting. Then got the key to the lab's "Kamp Katrina" (no joke), a trailer out back next to the chiller farm with a bare mattress where I slept out the rest of the work day. Ready to begin again when nearly everyone else had gone and the lab was again quiet and ready for work.

36 hours at the lab.

Late night soldering.