## The Art Of Noise June 1, 2009

Posted by gordonwatts in science.

We like to say that science is deterministic. You do x,y, and z, and then a will happen. I 100% agree with this. Except when it comes to noise.

What is noise? This is part of the problem – it comes in so many forms. Noise is like a weed. We call a plant a weed when it doesn’t belong. Got grass in your garden? It is a weed. Got grass in your lawn? Definitely now a weed. Is your cell phone making buzz-buzz-buzz sounds on your computer speakers? Noise! Are you using the cell phone signal to talk? Not noise!

But electronic noise is more insidious than the weed analogy implies. If you have an iPhone you can do the following experiment. Sit down at a computer that has separate speakers – the key is that an exposed wire connects the speakers to the computer. Place the iPhone near the speakers*. Wait for a few minutes. Now you’ll hear the buzz-buzz-buzz. So you know the source – you put the iPhone near the computer and it makes that buzzing sound. Lets say you didn’t know that. You had the iPhone in your pocket, and you walked up to your computer – it would start buzzing. You’d notice that every time you were near you’d get the buzzing. You might conclude you were the source of noise (talk about a loud heartbeat!!). You’d be both right and wrong.

The beauty of science is that if you do a, b, and c, then x will happen. Electronic noise is very much about science – if you do x, y, and z then you’ll get noise. The problem is it is very very difficult to determine what x, y, and z are. Above x might be “walk near your computer”, and y might be “have iPhone in your pocket”. You might get x right away, and then give up on fixing the speaker. A computer is no good if you can’t be near it! If you figured out y you might be able to do something different (like leaving your iPhone across the room, or better shield your speakers).

This is why finding and diagnosing electronic noise is an art.

I was on shift for two nights last week. On Friday night we saw a strange noise pattern in D0’s calorimeter. The pattern of noise is called high noon. And it was in four crates of electronics – typical high noon noise is never seen in more than one crate at a time. Experts were here until 2am trying to figure out what the source was. No clue. Eventually they went home – have to investigate the next morning when everyone was awake. Impact on data seemed to be minimal, and the automatic noise detection algorithms seemed perfectly capable of detecting this type of noise.

There were other problems with the detector that night too. A muon phototube kept looking synchronization with the readout. Each time this happens an automatic system would rap its knuckles and bring it back into line. Finally, after it had lost sync about 20 times, we decided to hit it with a hammer: reset it. And lo… the noise in the calorimeter, which had been mocking us for the last 6 hours disappeared.

Keep in mind – the muon system is meters away from the calorimeter system. They share no common electronics – not even racks of electronics are shared by the two. But the muon system was acting like the cell phone, transmitting electronic noise and the calorimeter was acting like an antenna – the speaker wire – and picking up the noise. D0 has been running since 2001. As far as I know, no one had seen this particular failure mode before.

If anyone tells you science isn’t an art – that there is no room for creativity – they are full of it. There are a lot of unknowns in unsolved problems – and it takes some creativity to guess what those unknowns are. And that is also the point behind reproducibility – to make sure you found x, y, and z that produce effect a.

Long live art!

* BTW, the iPhone isn’t the only one that will do this, it is just the worst offender of any of the smart phones I’ve seen. And easy way to fix it is place the iPhone on some tin foil. I’m not sure if that is because the tin foil becomes an antenna and so the iPhone needs less power to communicate with the tower or because it provides some shielding. But it works. See? Art!

1. Nick - June 1, 2009

Aww, man. I hate bugs like this. They’re always in the spots you never think about.

I figure there would be more shielding with something as large and complex as D0. How much effort do you put into keeping this kind of noise out of detectors in the first place?

2. gordonwatts - June 1, 2009

Years and years. Some of it is unavoidable – for example, if you have readout electronics right on the detector (which is often the case because the amount of charge you are measuring is so small), you have to make srue it is “off” while other detectors are collecting their signal. This is what I suspect happened here: the muon chamber’s firmware got corrupted and so was still running while the calorimeter was collecting its signal.

Then there was the time we discovered that if a welding machine was running in the building our detector was a mess…

3. Freeney - June 2, 2009

It’s called GSM buzz. The signal Inherint in ATT, TMobile, and Nextel phones. It can do some weird things. If the souce is close it will Buzz speakers, change volume on an ipod, modify an electronic timer, and I am sure there are others. Foil is good but you can also place your phone on an anti-static bag. Or better yet place it inside the bag. This will slightly reduce your signal but the phone still functions. I believe the foil or metalized anti-static bag dispurses the signal. I learned about this at http://www.stopthebuzzin.com.

4. gordonwatts - June 2, 2009

Nice! I’d not seen that web site before.