4GLS Beam Loss initial brainstorm - 4th May 2007

Machine protection system has to be designed - we need to consider all likely eventualities.

General considerations: some diagnostics will cause loss (e.g. running into an FCup), so we will need different operating modes to allow this.
As with other facilities, we will need an interlock system
We need to obtain a time constant for the cryomodules - how long will they ring if the power goes off? About 0.1ms
Machine protection scales:
30-300 ns is as fast as signals can travel through the 4GLS; machine protection can't go as fast as this.
Typical switch off in machine protection is 10s of us (check).

Vacuum faults:
Slow leak - Gas bremsstrahlung (radiation monitors), ion trapping change (how much?), linac ices up,
Fast leak - triggers gate valve - need to turn off electrons before valve shuts. What happens if a gate valve shuts by itself? This is probably standard accelerator technology, but the beam power is very high in our case. e.g. 5 kJ can hit a faulty valve in 0.1 ms if the electrons are not turned off! Same goes for diagnostics, e.g. screens.

HACL:

Cathode:
Long time constant change (QE change)
Laser shut-off - probably not a problem as there are no electrons
Laser loses sync with RF system (outside bucket) - bunch is at different phase, so is bunched differently and comes out with different energy from injector. How fast can that happen?
VUV-FEL mode: low rep rate selected. Can the laser go CW (chopper fails)? How will the chopper work?
other failure modes of the laser/cathode?

Gun:
HV PSU trip-off (low SF6, PS fault) - injector energy will reduce over a time until we can trip off the laser. Particles will go in the wrong direction in BC1 - need to collimate?
HV PSU set wrong - what happens if we are mis-tuned, and how do protect against it?

Buncher:
Fails, beam is not bunched - what happens? Energy spread and length growth - will this be beyond
Buncher works at the wrong phase? Can the acceleration produced be significant

HACL Modules:
Modules at wrong phase
Wrong voltage - probably be to reduce the voltage
Everything here has a time constant
RF control fault - e.g. missing bunches or pulses - gives transient beam-loading

BTS:

Magnet failure:
Wrong setting/Trip - may for example have to arrange that inductance of magnets is high to make their intrinsic turn-off slow. May get this for free...
Some magnets are more critical than others (e.g. the ones upstream of the main linac)
Collimators are in the HACL arcs to take a short pulse of errant beam, and also to catch halo; main linac can easily change in 0.1ms.
Magnet failure can cause path length change - need to check how big this change can be.
Magnet failure can cause beam movement which can hit the IDs - need to do something similar to what we did for the SRS and DIAMOND, e.g. ray-tracing of possible beam routes through the IDs.

Main Linac:
Check - ILC linac failure modes.

VUV-FEL:
Lasing goes on and off - should be no problem as unlased energy range lies within lased energy range.
Steering from misaligned/faulty magnet arrays - can't go too fast, should be able to be

PLC:
PLC moves or there is a magnet failure - energy recovery is lost. Linac will pick this up.

Dump:
Energy too high - power load too high, irradiation. Beam can also hit crotch and spreader.
Energy too low - absorption depth in dump will be less.
Raster magnets fail - need to independently sense the raster scan failure, as there will be lots of radiation/signal in this part of the machine.

XUV-FEL:

Gun/Injector:
Missed pulses can occur - probably doesn't matter if there are no pulses, e.g. in main linac.
Overpulses (e.g. extra laser pulses) - then get extra electrons that are not being accelerated. Depends on the length of the RF pulse.

Linac:
Wrong relative phase of XUV and HACL bunches. Can the HACL bunches 'push' the linac phase around enough to change the XUV gun phase.
HHG laser can melt something directly?

2D-IR Meeting - 24th April 2007

No AP work has been done since the 2D-IR meeting. FEL work has reviewed differing options for the RF module and bunch repetition rate.

The latest agreed configuration (v1.2) is the original CDR configuration (2 FELs not operating together).

George Neil has made comments that JLab is considering some similar solutions (perhaps with Talahassee).
Perhaps this paper: http://www.bessy.de/fel2006/proceedings/PAPERS/TUPPH009.PDF

Having 2 FELs from one electron beam, the action of one FEL disturbs the electron beam for the next FEL. It is an open question whether these two FELs can output at the same wavelength (to be considered by Neil Thompson) - same-wavelength operation would allow 'on the diagonal' measurements to be made.

It could be possible to use only coherent emission from an undulator rather than an FEL - it is not clear whether an FEL would offer any pulse length advantage (i.e. can it generate shorter pulses than coherent emission alone?), or is this just a limit from the wavelength?
Post-meeting comment - of course, if we have coherent emission at a certain wavelength, then there is no point in having an FEL, as the cavity will not cause any microbunching. Recall that an FEL only works at wavelengths shorter than the bunch length; microbunching of a bunch can't occur at length scales bigger than the bunch (think about it - it doesn't make any sense!).

Christine Ramsdale has considered some of the controls/DAQ aspects.

Mark's notes:
-----------------------------

The full suite of FELs to provide complete coverage of the wavelength region requested by David Klug would not be available on day 1 but a phased introduction of facilities is envisaged.

Neil Thompson agreed to look at feasibility of Dual undulator / 2 FELs in one cavity before 25 May. Then to investigate if coherent undulator radiation could also be used, but significant longer pulse lengths expected to be a problem due to slippage. FMQ noted that George Neil may already have proposed a solution for 2 FEL outputs with one injector system- HO will speak to him.


UV-vis detection pulse would be provided by down converting the UV-FEL radiation. This has the advantage of maintaining high rep rate compared to table top laser system.

MB confirms that can transport FEL light with good efficiency and to provide spot size required.

It was agreed that this experiment would be an ideal system to develop data acquisition for. GRM agreed to initiate a flow diagram for this.

Seminar - A.Terekhov, 18th April 2007

4GLS photoinjector problems:
10ps response decreases QE
Low (30 meV) energy spread decreases QE
100 mA on cathode needs 50 W laser, needs cooling
100 mA gives short lifetime due to ion back-bombardment

Solve the first 3 using new cathode heterostructure design - the response time depends on the number of random walks through the thickness of the cathode. A cathode with a thin active later is therefore better. But QE drops because of incomplete light absorption, and mean transverse energy can increase.
In reflection mode (RM), can use Distributed Bragg Reflector
In transmission mode (TM), can use Graded Band Gap Layer - can increase QE by maybe 10x.

Band bending near surface can increase transverse energy spread via momentum scattering (and 'acceleration'), up to value of negative electron affinity. This is not the longitudinal energy spread, however, the longitudinal and transverse energy spreads are similar (difference between conduction band minimum and vacuum energy level, around 150 meV at 300 K and 250 meV at 100 K.

If active layer thickness is smaller than the characteristic thermalisation length (of the order of 100 nm) then you get bigger energy spread.

Notes from 4GLS Physics Meeting 11th April 2007

Forthcoming milestones:
Wakes
Diagnostics
Main linac quads position fixed
Magnet spreader specification fixed - apertures
Magnet spreader design
BC1 optimisation - report due
BC2 optimisation - Sara
Collimators in HACL arcs - Hugh
Beam loss brainstorming - David 27th April
Longitudinal cavity wakefields of 7-cell cavity design and HOM dampers - Emma/Carl? late June