Machine FAQ (NC Linac)
What is the photon energy range which LCLS can provide and how long does it take to switch?
(April 21, 2021) T. Maxwell
At present, the available photon energy range is ~ 200 eV up to 25 keV covered by the soft x-ray and hard x-ray FEL undulators. The soft x-ray system will provide access to the 200 eV to 2 keV range. The hard x-ray system will cover the range from 1 keV to 25 keV with the higher photon energy range being commissioned through Run 18.
It is important to understand that this is the range of FEL operation. The x-ray optics transmission and instrumentation must also be taken into consideration for the determination of the experimental energy ranges.
The new Variable Gap Undulators (VGU’s) provide a means to change photon energies by changing the undulator gap (variable strength). This is a continuing area of development expected to result in faster energy changes and photon energy scanning capabilities. To date, soft x-ray photon energy can be varied up to a factor of 2 without significant degradation of performance for typical parameters. Hard x-ray photon energy can be similarly varied 25% or more.
Energy changes based on accelerator beam energy change are more performant and require anywhere from 5 minutes (small energy adjustments of 5-50%) to 30 minutes (energy adjustments of a factor of 2-3) with some accelerator tuning required in to re-establish the full X-ray pulse energy. This retuning is faster when the energy is increased rather than decreased.
Please ask the ACR operator for a time estimate when requesting photon energy changes.
It is also possible to make very fast, but small photon energy changes of +-2% (relative photon wavelength) via accelerator beam energy without requiring any other adjustments (although some small X-ray power variations may accompany this). This fast energy control can be implemented directly by users in a programmed fashion. Please ask your LCLS point of contact in advance about preparing this in coordination with the ACR.
Can experiments be performed at the same time in HXR and SXR beamlines? Can the beam parameters be changed independently for the HXR and SXR beamlines?
(April 21, 2021) P. Krejcik, T. Maxwell
For normal conducting linac operation a fast kicker located at the end of the linac can now deflect the 120 Hz linac beam into either the HXR or SXR undulators, allowing interleaved operation of the two beamlines. See FAQ 13 for details on the rates available in each beamline.
The variable gap undulators in each beamline allow the x-ray wavelength to be set independently for each beamline. Since the electron bunches are coming from the same linac the program coordinators will discuss with the users in each beamline what the best choice of beam energy, bunch length and bunch charge will be for each combination of experiments.
The brightness of the x-rays from each undulator depends both on the gap setting and the electron energy, and the reason the exact choice of electron bunch parameters must be negotiated and worked out with the program coordinators.
For the special case of fully interleaved delivery to both beamlines, as described in FAQ 12, there exists the capability to independently vary the electron bunch energy over a small range in order to allow Vernier scans of the X-ray wavelength for each user.
We also continue to expand the capability to provide different average electron beam energies to both undulators to enable simultaneous operation at very different photon energies for secondary undulator operation. This includes operation of the soft x-ray undulator as low as 700 eV during most hard x-ray delivery, and as high as 7-8 keV for the hard x-ray program during primary delivery to most soft x-ray programs. Again, please discuss with your beamline physicist and program coordinators regarding this new capability.
What is the highest pulse energy available (number of photons in the pulse) and how does it vary with photon energy and pulse length?
(July 23, 2019) T. Maxwell
As noted in the LCLS Parameters table, the expected pulse energy for the new undulator systems is about 0.6-1.5 mJ for the hard x-ray line for a nominal FWHM pulse duration of 30 fs, and ~2 mJ in the soft x-ray undulator for a nominal 50 fs pulse duration. The number of photons in the pulse depends on the photon wavelength (energy) and duration.
These are equivalent to 0.2e12 to 9e12 photons/pulse for the hard line and 0.8e13 to 60e13 photons/pulse for soft.
Very roughly speaking, reducing(/increasing) the pulse duration comes with a proportional reduction(/increase) in the expected number of photons per pulse for a given photon energy.
If the electron bunch is shortened by increasing compression (generating a shorter X-ray pulse), generally the number of photons in the x-ray pulse is reduced since the pulse is shorter. However, the increased peak current of the electron beam in the undulator makes a more efficient FEL up to a limit. Therefore the number of photons in the pulse may not drop as fast as a simple pulse-length scaling suggests and in some cases even increases. Note that this pulse length reduction by increasing compression also generally includes a corresponding increase of the FEL SASE bandwidth.
Whom do we contact, both before and during the experiment, to request changes or special parameters, and who do we contact for questions on X-ray beam characteristics?
(July 23, 2019)
Please always ask your LCLS point of contact first and they will help you reach the appropriate expert(s) where needed.
What is the nominal LCLS X-ray pulse length and how is it typically varied?
What is the nominal LCLS X-ray pulse length, how long does it take to change it, and how does the pulse energy and peak power vary with pulse length?
(July 22, 2019) F.-J. Decker
As noted in the LCLS Parameters, the nominal pulse length is 30 fs for the hard x-ray line and 50 fs for the soft x-ray line. Electron beam compression is the most direct way to make small variations in the x-ray pulse length, as detailed here.
The x-ray pulse length is approximately a copy of the electron bunch length which is set by the linac bunch compression parameters. A shorter electron bunch length produces a higher peak current in the undulator allowing for one degree of pulse duration control.
At the shorter photon wavelengths (~8 keV) the peak current required to achieve FEL gain saturation is 3000 A or higher. However, at longer wavelengths (800 eV and below) a lower peak current is possible (down to 500 A). Therefore, the pulse duration at longer wavelengths can be varied at anywhere from 250 to 30 fs FWHM (i.e., 500 A to 4000 A at the nominal 125 pC bunch charge, t = q/I), but at the shortest wavelengths the pulse length must be no longer than 30-50 fs FWHM (i.e., 3000-5000 A) or the FEL power will be very low and quite unstable from shot to shot.
Adjusting the pulse length within compression limits can be accomplished in 1-2 minutes. In fact, the electron bunch length is under feedback control and the desired setting can easily be entered, with less than a minute required for full stabilization, though some minor reoptimization may be required for large changes.
Still-shorter pulses from < 10 fs to 30 fs are possible by cutting down the electron beam in the linac via other methods (e.g., the slotted foil). These methods truncate pulse duration while nearly preserving peak power. P = E/t, e.g.: 2 mJ / 40 fs = 50 GW.
What is the degree of coherence of the X-ray pulse in both transverse and longitudinal directions and how is this affected by the choice of photon energy?
(July 23, 2019) H.-D. Nuhn
The FEL x-ray pulse has good transverse coherence (M = 1 - 1.5), since for normal operation the transverse coherence area is larger than the spot size. However, systematic errors in the FEL may slightly degrade the transverse coherence and require additional tuning and correction where critical.
The FEL does not generally have full temporal coherence. In SASE mode, the frequency bandwidth can be 200 times larger than the transform limited bandwidth for very long pulse duriations (>30 fs) to 30 times the transform limited bandwidth for short pulse duriations (<10 fs).
Note: Longitudinal coherence is improved in Self Seeded Modes. See related FAQ.
How does the X-ray transverse beam size vary with photon energy and electron beam compression settings??
(July 23, 2019) H.-D. Nuhn
The X-ray transverse divergence and therefore beam size downstream of the undulator changes as a function of energy. The size at a given experiment will depend on the distance, source point, and divergence. See an example of size versus energy in the middle of the FEE (at the direct imager) in the below figure.
For details on aperture, final spot size, focusing, etc, please refer to the appropriate beamline and your LCLS point of contact.
Where is the approximate location of the X-ray source point along the undulator, how does this vary with photon energy and compression, and how can we move this source point?
Where is the approximate location of the X-ray source point along the undulator, how does this vary with photon energy and peak current, and how can we move this source point? In addition, what is the source size and divergence and how does this scale with photon energy and peak current?
(July 23, 2019) H.-D. Nuhn
The X-ray source point is located about 1-2 Rayleigh lengths upstream of the end of the last used undulator. The source point location can be moved further upstream by removing down-stream undulators (about 4.4 m per undulator for SXR line and about 4 m for HXR line). For a given undulator configuration, the relative source point location, i.e., the distance of the source point to the end of the last inserted undulator, as well as the rms source size and the far-field rms divergence depend on photon energy, electron beam charge, and peak current. See the figures for estimates of the source point and divergence at various energies for the HXR and SXR beamlines when driven by the copper (normal conducting) linac.
For further details on transport including aperture, focusing, final spot size, etc, please refer to the appropriate beamline and your LCLS point of contact.
Where are the direct imager and the “YAGXRAY” screens located with respect to the undulator exit?
(July 23, 2019) B. Jacobson
Each of the new undulator beamlines is instrumented with several X-ray profile screens. Historically the “YAGXRAY” camera and other imagers were used with one in the Electron Beam Dump (EBD) and one in the Front-End Enclosure (FEE), respectively. Each system uses a scintillator and an optical CCD camera to capture images of the X-ray profile.
For Run 18 onward, other imagers are designated “IM#H/S”, where # refers to the occurrence number and H/S refers to hard or soft beamline. Each unit has three scintillator options: YAG:Ce, and two diamonds. The YAGXRAY screen remains on the hard X-ray beamline and the new YAGXRAYB is the counterpart for the soft X-ray beamline.
Locations of the X-ray profile screens along the beam path are listed below for the soft and hard X-ray beamlines. In each case, the distance referenced is from the end of the last undulator – segment 21 (SXR cell 47) for the soft line; segment 32 (HXR cell 46) for the hard line.
What are the shortest pulse lengths possible and how do parameters vary in these modes?
What is the shortest pulse length possible, how does this vary with photon energy, and what pulse energy is available at the shortest pulse? In addition, how long does it take to establish this pulse length and also to go back to nominal conditions? Can it be changed any time of day or night?
(July 23, 2019) M. Gibbs
In normal operation (see LCLS Parameters), the FWHM duration of the X-ray pulses can be set between 10-50 fs for hard x-rays and 30-150 fs for soft x-rays. The pulse energy is roughly 0.5-2 mJ in normal operating mode depending on photon energy. Changing the pulse length in this mode takes seconds for small (~20%) changes. Large changes may require some tuning to optimize the pulse energy, and take around 10 minutes. See also FAQ #4.
There are methods that can produce significantly shorter pulses, at the expense of pulse energy:
A "slotted foil" in the electron bunch compressor can be inserted by operations. This foil only lets a small portion of the electron bunch through the slot while the rest of the bunch passes through the foil and is degraded. Only the electrons which traveled through the slot will lase. The width of the slot can be adjusted, producing an X-ray pulse duration from 5 to 20 fs FWHM. At the 5 fs end, the pulse energy is 50-100 uJ. At the 20 fs end, the pulse energy can be up to 0.5 mJ. Set-up for the slotted foil mode takes roughly ten minutes. Once setup is complete, changing the pulse length via the slot width takes seconds. See FAQ #19 for more details on this mode.
Low Charge Mode
The electron beam charge can be lowered from its typical value of 125 pC to 20 pC. In 20 pC "low charge mode", the fully compressed electron bunch can produce X-ray pulses with a minimum pulse duration around 10 fs. In this mode the pulse energy will be around 150 uJ. Setting up low charge mode takes roughly an hour. Once established, changing the pulse length in this mode is very similar to normal operating mode: small changes happen in seconds, large changes take minutes.
Of the two methods, the slotted foil approach typically yields better power, is easier to reproduce shift-to-shift, and is easier to tune. However, at low electron beam energies (less than 5 GeV), the electrons spoiled by the foil can cause spurious machine protection system faults, which may be frequent enough to have an impact on delivery. Reducing the bunch charge may be a better choice in this situation.
Neither slotted foil mode nor low charge mode can be independently used during simultaneous delivery to both HXR and SXR undulator lines.
Sub-femtosecond operating modes have been recommissioned in Run 18. This includes XLEAP for soft X-rays and non-linear compression for hard X-rays, each generating tens of microjoules in < 1 fs. Please see the LCLS Parameters page and discuss with your LCLS point of contact for details and availability.
The XTCAV diagnostic, located just after the undulator, can be used to measure x-ray pulse profiles on a shot-by-shot basis. At full power it has an intrinsic resolution of 1-4 fs, with better resolution for longer wavelengths (related FAQ below).
What are the temporal characteristics of the X-ray pulse?
What are the temporal characteristics of the X-ray pulse, including number of spikes, spike duration, peak power in each spike, and how does this vary with photon energy and peak current?
(April 21, 2021) A. Brachmann, T. Maxwell
The temporal characteristics of the LCLS X-ray pulse are chaotic due to the SASE process. The X-ray pulse profile consists of many random spikes that vary from shot to shot. Typically, the spike duration is about 1-2 fs for soft X-rays (200 eV to 2 keV) and down to 300 attoseconds for hard X-rays (up to 25 keV). The total number of spikes is approximately the ratio of the X-ray pulse length (see item 4) to the spike duration. The peak power in each spike can be anywhere between one to hundreds of GW, depending on electron peak current, and can fluctuate from spike to spike.
We have also demonstrated sub-femtosecond pulses in the SXR and HXR range using advanced accelerator beam system and operational modes: XLEAP for SXR and non-linear compression for HXR. XLEAP has been recommissioned for SXR and HXR available but with testing pending on the new HXR undulator systems.
What is the bandwidth of the X-ray pulse and to what extent can it be controlled?
What is the bandwidth of the X-ray pulse and to what extent can it be minimized or maximized and how long does it take to make such changes? Can the X-ray pulse be energy chirped and if so, what is the final bandwidth available (FWHM)?
(July 22, 2019) J. Welch
For the normal SASE mode of operation the minimum bandwidth (BW) of the x-ray pulses is determined by the SASE process. In the SXR beamline the minimum FWHM BW ranges between 1.8 eV, for 250 eV x-rays, to 5 for 1.3 keV x-rays. In the HXR beamline the minimum FWHM BW ranges between 2 eV for 1 keV x-rays, to 30 eV for 25 keV x-ray.
These may be slightly reduced by de-optimizing the undulator taper, so the the electron is just barely saturated at the expense of pulse intensity. Spectral brightness is highest when the undulator is tuned for the maximum energy per pulse.
By “over compressing” the electron bunch, which introduces a relatively large correlated energy spread on the electron bunch, the relative BW can be increased to roughly 1% for the HXR and 3% for the SXR beams. In this case there will be a chirp on the x-ray pulse. It is a relatively quick and reversible process to change from normal to over-compression, as it only involves change phases in the linac and minor FEL reoptimization.
A monochromator can be used to reduce the BW of the delivered x-rays at the expense of photons. Please refer to the appropriate beamline and your LCLS point of contact for further information.
In seeded modes the BW is comparable with that of a monochromator and the brightness can be higher than for SASE. See the FAQ on seeded operation.
What is the machine repetition rate at present? Is there a possibility to increase the rate beyond 120 Hz? Can a single pulse be triggered in a one-shot mode and can the experiment pull the trigger?
(July 22, 2019) P. Krejcik
The normal conducting linac can operate at rates up to a maximum of 120 Hz. All macro-pulses (typically containing one pulse) are selected from this continuous and equispaced train.
The 120 Hz rate can be further subdivided into lower rates between the HXR and SXR beamlines. The resulting maximum allowed rates in the two beamlines must add to a maximum of 120 Hz, so that if the HXR receives N Hz then the SXR may receive (120-N) Hz, where N can take on values of:
0, 1, 10, 30, 60, 90, 110, 119 or 120 Hz.
The maximum rate in each beamline is assigned by the program coordinators. The assigned maximum rates can be independently lowered on demand in each undulator to any lower integer values <N in the HXR and <(120-N) in the SXR, including single shot and burst mode. Burst mode delivers a prescribed number of bunches to the user with a spacing determined by the rate N or (120-N) for the beamline.
Single shots and burst mode can be programmed by the user, selected from the maximum rate assigned to the beamline.
The rates described above always refer to the macro pulse rate from the linac, derived from the macro-pulsed repetition rate of the RF klystrons in the NC linac. Normally only one electron bunch (and therefore X-ray pulse) is delivered per macro RF pulse. However, under special circumstances the linac can also be operated in a “multi-bunch mode" where two or more bunches in close succession are delivered per linac pulse. The separation of the "multi-bunches" has to be an integer multiple of 0.35 ns (a number determined by the linac RF frequency of 2856 MHz). Yet another mode allows for "twin bunches" very close together (femtoseconds) within the same RF pulse.
These modes affect beam delivered to all destinations and are not generally compatible with parallel running. Please refer to the LCLS Parameters page for further details on these advanced modes and speak with your LCLS point of contact regarding availability.
Can I trigger the X-ray beam from my own experiment when I want it rather than asking the main control center? If so, can I vary the trigger time in arbitrarily small increments without regard to the 120-Hz triggers (i.e., trigger the linac)?
(July 23, 2019) J. May
You're able to move your experiment's timing around, but not the beam itself. The arrival time of the x-rays to your experiment is determined by the bulk of the configuration of the accelerator. Things like the klystrons, the injector lasers and time domain diagnostics all depend on time (and phase) stable operation, as do systems involved in machine protection and beam containment. Therefore, the accelerator does not, by design, allow for this.
Related to this point, FAQ #13 speaks to the selection of beam rates. While beam is available at the assigned rate, an experiment can have delivery suppressed at any destination by enabling BYKIK.
Separate to the accelerator, and key to the question, users have access to the triggers for their experiments. Delaying your triggered acquisition systems for your diagnostics in most cases accomplishes the same thing that you would by moving the entire accelerator around. Further, with experiments running supplemental laser systems, laser timing, controlled by the same precision timing hardware as used on the injector lasers, can be controlled as desired, allowing experiment control of laser/xray separation. These systems typically have step sizes on the order of 20fs (open-loop) and with diagnostics like xray/optical cross-correlators (the "time tool"), can be operated closed loop. Always discuss necessary temporal structure with your SLAC contact prior to experiments, and the Program Deputy and EOIC during your run.
What is the shot-to-shot jitter I can expect of the pulse arrival time, pulse energy, photon wavelength, pulse length, and transverse pointing, and is there a set of parameters that can be used to minimize these?
(July 22, 2019) FJD
- The shot to shot jitter for the X-ray pulse arrival time (with respect to the RF reference phase) is about 50 fs rms over 1-2 minutes
- The photon pulse energy or intensity jitter is about 3-10% (for 0.8-8 keV photons, respectively)
- The relative rms photon wavelength jitter is about 0.10-0.05% (for 0.8-8 keV photons, respectively, and based on measured electron beam energy jitter of 0.05-0.025%)
- The photon pulse length jitter is about 7%.
- The transverse pointing jitter is about 5-10% rms (for 0.8-8 keV photons, respectively), expressed here as a fraction of the rms electron beam size. The transverse jitter is often dominated by electron beam dispersion (x, y versus E) in the undulator which can be reduced to zero.
These stability levels are reasonably accurate for the nominal 125 pC bunch charge.
A “slotted-foil” method has been described, capable of producing 2-fs X-ray pulses, or two similar pulses with variable timing delay. What range of the two-pulse timing delay is possible using this method...
A “slotted-foil” method has been described, capable of producing 2-fs X-ray pulses, or two similar pulses with variable timing delay. What range of the two-pulse timing delay is possible using this method and how many photons can be available in each pulse? Does it depend on photon energy?
(July 23, 2019) M. Gibbs
The slotted foil, first proposed in 2004 (Phys. Rev. Lett., 92, 074801, 2004), was installed at the LCLS in 2010, and it has been developed as a regular operation mode for short pulse generation and pulse length control. It is located at the middle of the second bunch compressor (BC2). Since no machine configuration changes are needed to set up slotted-foil mode, it provides a fast way (a few minutes to check the foil horizontal alignment) for X-ray pulse duration control.
Three slot arrays are widely used, one single-slot array and two double-slot arrays. By choosing the slot width or slot separation, the single pulse duration or double-pulse delay can be controlled. It also depends on the BC2 bunch current. A Matlab-based GUI is used to control the foil and to calculate the unspoiled electron bunch duration or delay, found to be consistent with measurement.
For 150 pC operation, the calculated achievable pulse duration and double pulse separation can be found in the following table (BC2 current 1-2 kA is mostly for SXR, and 3-4 kA for HXR operation):
single slot, pulse duration double slots, pulse separation BC2 min (fs) max (fs) min (fs) max (fs) 1 kA (SXU) 13 61 37 82 2 kA (SXU) 8 27 16 36 3 kA (HXU) 6.3 18 10 23 4 kA (HXU) 5.8 14 7 17
When additionally in 20 pC operation, these numbers can be reduced to the few-fs level, at the expense of pulse energy at the cost of additional configuration time.
Measurements with the X-band transverse cavity have been performed recently, and are in good agreement with expectation (Appl. Phys. Lett., 107, 191104, 2015). This can also be recorded during tune up and/or experiment together with user data recording.
In January 2015, a second double-slot foil was installed for double-bunch mode, but can also be used in single-bunch mode to extend the double pulse delay upper limit approaching to the full duration of the bunch.
This mode affects beam delivered from the normal conducting linac to both undulator lines and therefore generally not compatible with parallel experiments.
How much separation in time and energy is available over all methods of generating dual pulses?
(Apr. 21, 2021) T. Maxwell
The LCLS has developed several “two pulse” operating modes driven by the normal conducting linac, several of which were recommissioned over the course of Run 18. Appropriate selection of the FEL mode depends on the experiment’s energy/time separation and pulse intensity requirements.
The largest possible energy separation will be up to a factor of 2 between photon energies via the split undulator technique with up to a few tens of microjoules per pulse and fs-scale pulse delay. Other more advanced methods allow higher pulse intensities (fraction of a millijoule per pulse) only for energy separations of up to 2.5% over a wide range of pulse delays. High-level details are outlined in the multi-color section of the LCLS Parameters document (below).
Users are strongly encouraged to discuss requirements for two-color experiments with their point of contact and accelerator staff prior to submission of a proposal.
How much 2nd and 3rd FEL harmonic power usually accompany the fundamental FEL pulse and how does this vary with photon energy?
(4/08/15) LCLS Machine Phys.
Measurements show the FEL can contain a 3rd harmonic component as high as 2% of the power of the fundamental wavelength, and a 2nd harmonic component as high as 0.05% of the fundamental. Typical measurements of the 3rd harmonic component are 0.5-1.5%. A 2nd harmonic component has been measured at 0.03-0.05% of the fundamental power.
What should I know about the XTCAV and what can I do with it?
(July 22, 2019) P. Krejcik
An X-band transverse deflecting cavity (XTCAV) is installed downstream of the HXR undulator beamline and a second XTCAV is being installed in the SXR beamline to serve a similar purpose. Each device measures the electron bunch time-energy phase space distribution. Since they are located after the undulators, time-resolved FEL lasing effects (electron energy loss and energy spread increase) can be measured. By comparing images between FEL-on and FEL-off conditions, we can reconstruct the X-ray temporal profile for each lasing shot without interrupting FEL operation.
The best temporal resolution measured is about 1 fs rms for soft X-rays (800 eV), and about 4 fs rms for hard X-rays (8 keV). Experimental results have been published in Nature Communications, 5, 3762 (2014). To record XTCAV data from the beamline, in each machine configuration, ACR provides calibrations and also suppresses lasing for the user to record necessary baseline (non-lasing) images. This may take a few minutes. Once recorded, regular operation is resumed and XTCAV images are recorded at will.
A configuration for the LCLS DAQ system has been implemented. Python-based analysis scripts are available to the beamlines to perform single-shot reconstruction. These are compatible with both ordinary SASE mode, as well as a number of more advanced LCLS machine configurations.
The camera can record images up to the full 120 Hz beam rate.
With dual operation of the HXR and SXR beamlines with interleaved bunches (see question 12) it is also possible to simultaneously record XTCAV profiles in both beamlines on alternate bunches. The temporal resolving power of each XTCAV is a function of how strong the RF kick is in each XTCAV deflecting structure, but since the RF power comes from a single klystron the resolving power of each XTCAV will depend on how the mechanical power splitter is set (in the range 0 to 100%). The program coordinators will weigh up the XTCAV needs for each experiment and find the best compromise to allow XTCAV measurements on each beamline when needed.
Please speak with your LCLS point of contact for the most recent information on recording and analyzing XTCAV data.
What modes are available for seeded beams?
(May 14, 2021) A.Brachmann, T. Maxwell
Hard and Soft X-ray Self-seeding ("HXRSS" and "SXRSS") are planned for the corresponding undulator beamlines. The original LCLS seeding systems have been re-designed and re-built for the new Variable Gap Undulator beamlines and to be commissioned. HXRSS was commissioned in May 2021, SXRSS commissioning planned for the winter of 2021. Full functionality at the beginning of the run cannot be guaranteed at this time (May 14, 2021). Please discuss availability with your LCLS Point of Contact.
Setup time is approximately 1 hour for hard and 2 hours for soft. An initial machine configuration is generally built in advance of the first day of the experiment. Additional setup time is still required as part of the experiment from day to day through the remainder.
The actual performance parameter for a particular experiment depend on a variety of factors, for example the exact energy, required bandwidth and acceptable spectral pedestal.
Mode Energy Range Bandwidth Pulse Energy Pulse Length HXRSS > 4.5* keV 0.35-1.5 eV 0.25-0.5 mJ up to 40 fs SXRSS 0.4-1.2 keV
~ 100 meV @ 400 eV
~ 150 meV @ 530 eV
~ 200 meV @ 800 eV
< 50-100 μJ @ 20 fs
up to ~ 0.5 mJ with pedestal
* above ~ 9.5 keV the achievable pulse energy drops off.
For seeded dual energy / dual pulse beams, see the parameter table posted at question 20.