Do we start with de-Gauss cycle and then move to nominal. Nominal implies 20 minute wait to avoid bulk of persistent current decay?
General feeling is that, given the values of the full persistent current effect, the decay and the geometric, it's actually probably not worth using the de-Gauss cycle initially. It would however be a useful backup and an interesting cross-reference (say, to establish the geometric) when the machine is well established.
Following Oliver's talk at http://ab-div.web.cern.ch/ab-div/Conferences/Chamonix/chamx2003/PAPERS/7_3_BO.pdf
Thanks to Stephane for the updated figures
Nominal persistent current b3: -7.5 units
Decay: +1.7 to 2.3 units (45%)
Leaving, after 20 minutes wait: -6.5 units b3 persistent.
Now add in geometric
New cross section : Geometric +2 units
leaving total of around -4
b3 See +3.8 units Q'x/Q'y +176/-176 check figure.
Correct 90% leaving 0.4 units:
b7: 0.35 units on nominal, 0.8 units on de-Gauss (-0.45 persistent, geometric +0.8)
b1 will have to be re-done
However De-Gauss:
Oliver..
I think most of your arguments for the de-Gauss cycle apply also for the nominal cycle provided:
-stable in time: one waits for more than 30 minutes after the magnet cycle. This should not bee too stringent a condition since one will probably spend most of the time at 450 GeV without cycling the magnets (we should not quench with low intensities during the early commissioning).
-No dependence on powering history:
I thought the field errors do not depend on the powering history provided the magnets spend more than 30 minutes at nominal current.
This should again not be a strong limitation to spend ca 30 minutes at top field during the magnet cycle. Perhaps one could even imagine a second cycle that should remove any dependence on the previous powering history.
-Saving of 20 minutes: if we assume that we will not cycle the magnets much during the early commissioning at 450 GeV this is a weak advantage. For HERA
we stayed essentially more than one week at 450 GeV without cycling the machine (if I remember correctly).
The main advantage for proposing the de-Gauss cycle was in my opinion the reduced b3 error. With the new MB cross section the b3 is almost the
same for both cases (nominal and de-Gauss). I think it would therefore be better to start with the nominal cycle right away and to learn the proper corrector
settings for the 'real' cycle and to use the de-Gauss cycle only as a fall back solution in case one encounters too many problems.
Stephane...
I am almost OK with all your argument (stability and so on). For the particular case of b3, I also agree while your number is not correct for the latest cross-section. By memory (I am at home), the persistent induced b3 is around -7.5 units, the geometric b3 (average over the machine and for all X-sections and including beam-screen contribution) should be around +2units, and the decay should brange between +1.7 and 2.4 units depending oncycle and so on. Therefore if you wait an infinite time you will get a systematic b3 of about +2 units instead of -5/-6 units with the nominal cycle. I think that what Oliver had in mind is for Xs2 where we had about -4 units for the nominal cycle and +4 units for the degaussing cycle... so no gain except stability and reproducibility.
Another argument for the degaussing cycle is b5: 0 units of geometric and about 1.1 units of persistant so very small value for the degauss cycle (but the MCD spools will have to be commissionned somehow). However, what, to my point of view, may kill the degauss cycleor reduce his usefullness for a full commissionning of the LHC is 1) b7 as soon as we will have a circulating beam: about 0.35 units with nominal cycle (at the real thresolds for DA) and 0.8 units for the degauss cycle (the persistant is -0.45 and the geometric is +0.8). 2) b1: if the beam is RF captured with the Deguass cycle, the RF adjustment will have to be redone with the nominal cycle due to a persistent b1 of about 8 units.