jump to navigation

There is a cable on the floor! May 23, 2007

Posted by dorigo in computers, humor, physics, science.

The Level-2 upgrade folks kept me and my shift crew company in the CDF control room during the first part of the night shift. Anadi, Laura, Dan and Ted had a long brainstorming session in the middle of our command central to figure out what would be the best course of action, to try and solve an intermittent problem we have been experiencing lately with a thing called “trigger supervisor”.

Basically, what is happening is that a cable communicating trigger information between two crates started giving errors recently, and the cause of the problem is not totally clear. These errors create downtime of our data acquisition system, which means reducing our overall data taking efficiency – and in these days of Higgs boson hunting, every lost event has the potential to bite off a tenth of our final significance on a discovery plot!

They have not solved the problem yet, despite discussing in depth several possible tests. We are now running one of these tests with a cable strung temporarily between two racks and laying over the floor tiles of our the counting room (these cables usually run under the tiles). They are leaving to get some sleep now, but they were concerned that somebody might either trip on the cable, or stomp it, or who knows what… So, quite in line with the spirit of the Idiot’s guide I discussed the other day , they designed a “professor-proof” protection, consisting of yellow tape strung around the danger area, plus a sign hanging from it. It reads “There is a cable on the floor!”. See picture below…

I find this hilarious on several levels. When they first showed me the sign, it reminded me of one of those “basic English course” sentences, like “The black cat is on the mat”. But it is especially funny to think of what the purpose of the sign is: to thwart the menace from the many Ph.D.’s roaming around here during daytime, who may be well able to compute Feynman diagrams by heart, but are likely to trip on their own shadow if given a chance. Also funny is the fact that our counting room is one place where cables are not a uncommon sight: they are simply everywhere, including, as I noted above, under the floor: taking the pains to write a sign to warn about one cable seems really unnatural…



1. Chase - May 23, 2007

Thanks for the picture — it’s great! Let me ask you a question about something you alluded to. The cable problem caused downtime of the data acquisition (DAQ). How does CDF record and analyze downtime? I am sure I could figure that out if I spent some time on the CDF website, but I would be interested in your big-picture view of how it works (is it automated somehow? is the downtime categorized by which system failed?).

I noticed in the log book that at the end of the shift there are numbers for delivered luminosity and acquired luminosity, where do they come from? I don’t expect a very detailed answer of course (or even one at all depending on how busy things are at the moment!), but I am curious about how you see it working. I am somewhat responsible for analyzing the downtime at my facility (PITZ in Zeuthen, Germany).

Thanks again for the blog — it’s fun to see “behind the scenes”!

2. Andrea Giammanco - May 23, 2007

I would have stomped on it.

3. Quantoken - May 24, 2007

The sign states an obvious fact but fails to convoy the message its author intended. Suppose I am looking for a suitable cable and see the sign: OK, here I go, there IS a cable on the floor, so take it because that’s just what I was looking for 🙂

4. Kea - May 24, 2007

For short-sighted people such as myself, this sign would require closer inspection, resulting in tripping over the cable.

5. dorigo - May 24, 2007

Hi Chase,

our DAQ works with a three-level trigger system. We have a collision rate of 1.7 MHz, and we cannot read all the detector components at that rate, let alone save to disk. So we give a quick look with our level 1, while we pipeline information for several consecutive collisions. The level 1 decision discards most of the events, and sends the 20-30kHz most promising to Level 2. Level 2 has 20 microseconds or so to make a better use of all the detector data, and can thus reconstruct physics objects. If L2 finds out that the event satisfies some tighter requirements, the whole information is sent to a farm of PC (our Level 3 trigger), which receives in input up to 600Hz of events, and outputs 150 or so at most to tape.

Downtime results from the normal operation of the system, which tries to maximize the physics output, thus taking a bit longer to analyze more promising events, while others may be waiting in the pipeline and sometimes get lost. It also results from occasional failures of several of the hardware components. We have a monitoring of the deadtime which can tell what is causing trouble, so that we can keep things under control.

As for the luminosity, the delivered one is computed with our luminosity monitors, which are scintillator bars close to the beampipe. They record the activity of particles scattered from the collision, and from their occupancy (or rather, their lack of occupancy) and the proton-antiproton inelastic cross section one derives the instantaneous luminosity measurement. Then, the acquired luminosity is delivered multiplied by efficiency, that is one minus the deadtime.


6. dorigo - May 24, 2007

Andrea, me too..

Quantoken, that was another side of why the sign was amusing. In fact, it states the obvious without explaining why.

And yes, Kea, drawing attention is not always the best strategy to avoid harm.


7. Chase - May 24, 2007

Tommaso — thanks for the detailed response. I now understand where those figures are coming from. You guys have a different definition of downtime than what I am working on… I am only after what I think you are calling deadtime, and analyzing which system failures are causing most stoppages. I will keep these two different concepts in mind when I try to compare my numbers to how it’s done at the big physics beam experiments.

8. dorigo - May 24, 2007

The difference between downtime and deadtime is, for what I know, purely a matter of definition. Down means that the system is not active, dead means that it is overflown with information and unable to process it fully. During deadtime a device is nominally on, but cannot perform its task because of intrinsic limitations.


9. Damon and Jolie in the CDF control room « A Quantum Diaries Survivor - November 20, 2007

[…] CDF control room is the place where the detector is operated during data taking. It is a place full of monitors of […]

Sorry comments are closed for this entry

%d bloggers like this: