Let's go hump hunting!

The place to discuss particle physics
josch222
Posts: 78
Joined: Sat Dec 12, 2009 4:10 pm

Re: Remember the facts

Post by josch222 » Sat Apr 03, 2010 7:43 am

Kasuha wrote:
josch222 wrote:(besides 50Hz ;-)
The truth is, there is not 50 Hz in the mains, the frequency is slightly changing throughout the day. See here for instance. But I don't see how these slight changes might cause the hump, the interference by far cannot go to 3.5 kHz IMO.
Of course you can't keep exactly 50Hz, ever been in a power station of the GW class?
It is pretty impressive (esp. the vibrations of the whole floor in the machine house).
You can switch large electrical loads immediately but it is not possible to apply the
demanded power immediately so the generators speed will go down a bit until the regualtion valves opened and the higher water or gas or steam flow has established.
But I really didn't know that the average is not constant, about 25 years ago
I was told that lower speed during the peak loads would compensated with higher speed
at night so all the mains frequency based clocks would keep the time in the long run.
Maybe that has changed with the integration of eastern european grids and/or the availability of better time bases like internet, time transmitters and GPS.
Or it was never true :lol: I remember that I had to readjust my mains based alarm clock
every few weeks, but I thought at that time this were due to bad filtering inside, so
that interference on the mains would cause additional counts or swallow some.

josch222
Posts: 78
Joined: Sat Dec 12, 2009 4:10 pm

Re: Let's go hump hunting!

Post by josch222 » Sat Apr 03, 2010 8:03 am

JNW wrote:Electrical interference can be caused by an alternative ground path.
[...]
Has anyone jogged around the LHC tunnel with a detector while the LHC is running (but with no beam, of course)?
That's a possible source too.
At least some of the power supplies seems to have ground fault detectors:
Some time ago I read about such a fault in the commissioning news.
But I don't know if all possible sources have such a device.

One should not rule out a local loop structure (simple steel construction f.e.) too,
which focuses or leads magnetic fields from supply lines to somewhere near the
beam pipes. Afaik there are places where the pipes run "naked", without magnets
around, maybe one should look at that places first.

josch222
Posts: 78
Joined: Sat Dec 12, 2009 4:10 pm

Re: Let's go hump hunting!

Post by josch222 » Sat Apr 03, 2010 8:45 am

Back to my idea with the ELF/LF interference:

So I looked at the spectrograms of a few stations for some "suspicious" signals
and listened some time to a real time audio stream from a station in northern Italy.
I found nothing, besides some funny coincidence:
At the time of injection around 1:00 Cern local time, i heard a signal and initially thought
that where the kicker magnets or something, but it turns out it sounds more like a characteristic GSM interference and it never showed up again at the following injections.
Anyway, that is not the kind of stuff to look for.

One problem remains: I simply don't know what to look for exactly, it would
be better to have at least a table that says when the hump was there and when not.
When was it at the highest freq. and when at the lowest.

If somebody wants too look at the spectra of the last hours here are the links of two
stations in Italy:
http://www.vlf.it/livedata.html
http://www.webalice.it/rromero/pnts_observatory.htm
Watch the timezone!
Interesting would be any diagonal elements or better a line going
up and down in the minutes range.
The last one is the nearest to CERN and it has a link to archived sound files,
sadly it is all mp3, so I think it is useless for a serious search for correlations
with the hump.

Alexey Kruglov
Posts: 3
Joined: Wed Mar 31, 2010 4:34 pm

Re: Let's go hump hunting!

Post by Alexey Kruglov » Sun Apr 04, 2010 10:54 am

1) First of all it'd be good to understand what signal that hump is in. As far as I understand they measure (how?) the transverse position of the beam at one fixed point in LHC ring, take it as a function of revolution number, and then take Fourier transform of it. The "tune" seems to be the number of betatron ocsillations per one revoltion. (I've read lecture notes on http://cdsweb.cern.ch/record/1017689/fi ... 07-014.pdf up to the chapter about RF system.) Those betatron oscillations are transversal oscillations of the particles about some periodic beam trajectory in the ring. They are due to the focusing effect of the optics and it's beam's normal behaviour. Horizontal and vertical oscillations are nearly independent and have different periods, there is little coupling (they measure coupling |C-|, but I don't yet understand what exactly it is) between them that causes some crosstalk. Betatron oscillations are seen as a narrow rigid peak on those spectra.

I've tried to estimate the RSM (root mean squared) amplitude of hump in relation to amplitude of betatron oscillations for the graph in the first link (https://ab-dep-op-elogbook.web.cern.ch/ ... Id=1068503). The graph has 28 pixels/10 dB vertical and 47 pixels/0.05 frev horizontal (frev = revolution frequency, i.e. per revolution) -> ~0.36 dB/pixel and ~0.001 frev/pixel. Betatron oscillation peak level is A_0=-57.5 dB, hump peak level is A=-93.5 dB, hump full width on level -95.7 dB is 12 pixels = 0.013 frev. Assumin Gaussian line porfile, this gives hump line halfwidth on 1/e level (by power) of σ=0.0092 frev, power P=sqrt(π) σ A^2. Betatron oscillation peak width is estimated as 1 frequency bin, that is w_0=1/8192 frev, power P_0=w_0 A_0^2. If those oscillation are in fact monochromatic, their power would be underestimated 0.5 to 1 times.

The net result is thay it gives power ratio P/P_0 ~= 1...2.1, amplitude ratio sqrt(P/P_0) ~= 1...1.45. I.e. hump's amplitude is comparable to betatron oscillations amplitude or is even larger.

2) On https://ab-dep-op-elogbook.web.cern.ch/ ... Id=1067999 we see some hum. On low frequencies it's both odd and even harmonics (except 1st harmonic), then only even. The 48th harmonic (the one 2 pixels wide) falls almost right between pixels, f=0 also seems to be between vertical axis and pixel to the left of it -> 48th harmonic's freq. ~= 0.2156 frev +- 0.1% assuming the error of 0.2 pixels (there's 45 pixels / 0.05 frev).

So the hum base frequency is f_hum = 0.004491 frev +- 0.1% = 50.50 Hz +- 0.1%, assuming revolution frequency of 11245.5 Hz. What that could be? What frequency are the power lines in Europe? Even if they're 50 Hz, it's too far from 50 Hz, unless I've made a mistake somewhere.

Kasuha
Posts: 570
Joined: Tue Dec 15, 2009 1:22 pm

Re: Let's go hump hunting!

Post by Kasuha » Sun Apr 04, 2010 11:45 am

Alexey Kruglov wrote:So the hum base frequency is f_hum = 0.004491 frev +- 0.1% = 50.50 Hz +- 0.1%, assuming revolution frequency of 11245.5 Hz. What that could be? What frequency are the power lines in Europe? Even if they're 50 Hz, it's too far from 50 Hz, unless I've made a mistake somewhere.
If 50.50 Hz was grid frequency, it would cause many clocks in europe go wrong 15 minutes a day so I believe it indeed isn't grid frequency.

User avatar
tswsl1989
LHCPortal Guru
Posts: 166
Joined: Thu Dec 24, 2009 2:22 pm
Location: Swansea, Wales, UK
Contact:

Re: Let's go hump hunting!

Post by tswsl1989 » Sun Apr 04, 2010 12:59 pm

UK grid is 50Hz, +- 1% (By law, in practice it's closer)
50.50Hz is right on the upper bound of that range.

Kasuha
Posts: 570
Joined: Tue Dec 15, 2009 1:22 pm

Re: Let's go hump hunting!

Post by Kasuha » Sun Apr 04, 2010 5:57 pm

tswsl1989 wrote:UK grid is 50Hz, +- 1% (By law, in practice it's closer)
50.50Hz is right on the upper bound of that range.
Hump is going from 2.6 kHz to 3.5 kHz, changing slowly between these during 7~10 minutes. Center is around 3 kHz then. If you do the math, interference with grid frequency could generate hump with center around 2 kHz. You sure can have 50.5 Hz in the grid for a while but then it ultimately must slide to the other end (50 Hz and below) and that'd generate interference way below the minimum hump frequency.

Plus I believe compensating possible grid interference was one of first things they were considering during LHC design.

Alexey Kruglov
Posts: 3
Joined: Wed Mar 31, 2010 4:34 pm

Re: Let's go hump hunting!

Post by Alexey Kruglov » Sun Apr 04, 2010 10:23 pm

1) Ha-ha. I've found the bug. The interval between the grid lines on the graphs is not constant, it's not 45 pixels/0.05 frev, it's 45 to 46 pixels. On the graph with ~50 Hz hum there're 46 pixels between 0.1 and 0.15 frev, 0.2--0.25 frev, 0.35--0.4 frev, 0.45--0.5 frev. There're 454+-1 (that is +-0.2%) pixels from 0 to 0.5 frev, so 194+-0.2 (0.1%) pixels from 0 to 48th harmonic and frev=11245.5 Hz gives
f_hum = 0.004451 frev +- 0.3% = 50.06 Hz +- 0.3%,
that is consistent with 50 Hz hum.

2) The 50 Hz hum in horizontal... er... direction on 24th harmonic is comparable to betatron oscillations in that graph.

3) To Kasuha. I also thought they should've compensated for interference. What I was trying to say is that despite this there _is_ interference. Maybe that was a temporary problem on 30.03.2010, I didn't check other days yet.

4) The humP frequency isn't necessarily f=2.6 to 3.5 kHz. It could as well be frev-f or frev+f or 2frev-f or 2frev+v or n frev+-f because of aliasing. Also on graphs and in the log I've seen numbers like 0.31 to 0.38 frev = 3.5 to 4.3 kHz.

5) If humP is also a result of interference, there're 2 possibilities: interference in measurment circuits or in controlling circuits (like magnets or by external magnetic field, for example). The first doesn't explain beam lifetime--hump correlation UNLESS it is involved in some kind of feedback. The second could explain the hump if the interference is strong enough. It'd be good to estimate the field required to induce the hump, but I don't know how to convert decibels to the absolute units (like beam displacement per unit frequency or what).

Does anyone know what their dB mean?

6) It may be possible to gain some information about the nature of the hump by observing presence or absence of resonances as hump frequency drifts. I didn't think about it yet.

User avatar
Xymox
Site Admin
Posts: 1022
Joined: Wed Nov 11, 2009 12:55 pm
Location: Phoenix, Az USA
Contact:

Re: Let's go hump hunting!

Post by Xymox » Mon Apr 05, 2010 7:59 am

I sent this link over the Paul... Its got some interesting ideas for sure...

interference can just be very annoying to find..

So this hump, is it fundamentally / harmonically related to anything going on with the LHC ?

That would help narrow the issue down.. Maybe..

josch222
Posts: 78
Joined: Sat Dec 12, 2009 4:10 pm

Re: Let's go hump hunting!

Post by josch222 » Mon Apr 05, 2010 12:37 pm

After Alexeys posts I lookes at some screen shots of one LF monitoring station again
and found an interesting signal.
It is very doubtful that this has something to do with the hump because AFAIR it could
not be seen in the spectrum of other stations. It may be a poorly regulated off-line power generator or something close to the receiver (from the web-site it looks like it is near by
an Alpine hut). It appears to be randomly started and switched off. But who knows.
I found it interesting because the frequency is about 52-53Hz and it is slowly decreasing
over time.
Here are some pics:
Attachments
last-ponteseELF-ULF-03.04b.jpg
a longer period with visible drift
last-ponteseELF-ULF-03.04b.jpg (249.6 KiB) Viewed 19129 times
last-ponteseELF-ULF-03.04a.jpg
some short periods
last-ponteseELF-ULF-03.04a.jpg (246.24 KiB) Viewed 19129 times

User avatar
Xymox
Site Admin
Posts: 1022
Joined: Wed Nov 11, 2009 12:55 pm
Location: Phoenix, Az USA
Contact:

Re: Let's go hump hunting!

Post by Xymox » Mon Apr 05, 2010 9:06 pm

ULF/ELF is such a incredibly low power signal its hard to imagine it having enough power to induce anything into another system.. I made a elf receiver long ago. Lemme tell ya, thats one damn hard signal to pick up. 100,000 turns of wire on a iron core for example..

could see solar wind storm and its associated effects causing a problem, but thats not happening and we are years away from those effects..

I personally think its far more likely its a local interference source...

Kasuha
Posts: 570
Joined: Tue Dec 15, 2009 1:22 pm

Re: Let's go hump hunting!

Post by Kasuha » Mon Apr 05, 2010 9:43 pm

Since I read about tevatron being able to detect earthquake in Chile I believe accelerators can be very sensitive to non-local interference.

Anyway - I think there is no point in us watching ELF or any other phenomenon we propose to be potential source of the hump - it's now up to beam operators to watch the hump and any other set of possible sources and check if there is any correlation. Only after this the hump source can be identified.

User avatar
Xymox
Site Admin
Posts: 1022
Joined: Wed Nov 11, 2009 12:55 pm
Location: Phoenix, Az USA
Contact:

Re: Let's go hump hunting!

Post by Xymox » Tue Apr 06, 2010 8:00 am

Well those huge quakes produce surprising motions on the other side of the planet. Like 1mm.. Or more.. Thats enough motion at hz frequencies to move around magnets which really dont like to move as they are a tad heavy..

josch222
Posts: 78
Joined: Sat Dec 12, 2009 4:10 pm

Re: Let's go hump hunting!

Post by josch222 » Tue Apr 06, 2010 10:49 am

I don't think this thread is really about finding the cause of the hump.
I have too little information and background on accelerator physics and technology to
even make an educated guess.
It is more about brainstorming, giving some input in form of strange or even annoying ideas.

From my experience in troubleshooting complex systems:
From time to time I have to deal with some malfunction or strange behavior of the systems
I build. They are some orders of magnitude less complex than the LHC, but it happens
that I'm stuck despite I'm an expert. Sporadic errors are the worst, you sometime have to wait
days til it occurs and you are really lucky if it stays from then on, so you can make measurements.
In that cases my business partner usually panics because he has to deliver the system to the customer
in time.
At this point I get flooded with more or less stupid ideas (form my Pov) on how to fix things.
This can be really annoying because I always have to explain why his ideas are BS and
will do nothing to fix the problem.
But in this process you sometimes have to question your own fixed mind on things
that you take for granted, or you have a sudden idea for a test-setup to catch the flaw,
or you get a hint on how to provoke it.
The suggestions have rarely something to do with the real cause and are often simply
not useful. But in the end it helps you to think out of your routine and be creative in
finding solutions.

The interesting thing is:
In most cases it turns out to be a software bug, not an interference or the like.
So the chance that my LF-idea is wrong are pretty high :doh:

User avatar
Tau
LHCPortal Guru
Posts: 200
Joined: Tue Jan 05, 2010 10:16 pm
Location: Heemskerk, Netherlands

Re: Let's go hump hunting!

Post by Tau » Tue Apr 06, 2010 1:02 pm

Good point Josch.
I agree that it is important to allow all ideas to come through, even if they are wild. Even though I am pretty convinced the cause of the hump is in the tunnel (otherwise it would influence both beams, for example, not just one), I find it important to get as many ideas as possible.
Maybe someone at CERN reads this, and gets an idea; that's when we really succeeded, even if we didn't pinpoint the exact cause.
- Tau

Post Reply