The set-up shown in fig. 3 differs mechanically somewhat from that used in ref. : for compactness, the total length has been reduced, while the compression cavity has a larger bore (7 mm) for ease of beam alignment. This new set-up has been simulated in full detail with GPT, using a distance from compression cavity to sample of 40 cm. In practice, such a distance should allow attaching the beam line to fair-sized sample chambers. Given the good agreement found in  between the measured bunch characteristics and GPT-predictions, the GPT-performance figures shown in Table 1 for the set-up of fig. 3 are considered reliable.
Table 1. Bunch characteristics at sample*
Bunch charge 100 fC
Bunch length 100 fs (rms)
Spot size 250 μm (rms)
Transverse coherence length 3 nm
* see Annex 1 for more details
The version of GPT containing the field maps of all electron-optical components (except the x,y-deflectors), which was used for the above simulation, is available from AccTec B.V. This version readily allows an accurate prediction of the bunch parameters to be expected, for distances between compression cavity and sample different from that in Table 1, or for a different compromise between spot size on the sample and transverse coherence length. It is to be noted that spot size and coherence length can be exchanged to a considerable extent, but that space charge starts to play a role at too small spot size. Again, GPT is helpful in surveying the possibilities in this regard.
In ref. the bunch length was measured using an RF deflector cavity –at the position where samples are to be placed- to streak the bunches across a slit in front of the detector . Given the good agreement between measurement and GPT-prediction, it would appear that the GPT-values can be used reliably to deconvolute measured lifetimes in pump-probe experiments. If, however, an independent measurement of the bunch length is desired, a deflector cavity for streaking the bunch is available from AccTec.
For detection of diffraction patterns the sample chamber should provide a second port opposite the beamline port. It is desirable to apply a third focusing coil downstream of the sample, in order to focus parallel rays from the sample into a point on the detector and thus obtain the sharpest possible diffraction pattern. The detector is to be placed at a distance from the sample which is determined by the setting of the third focusing coil and by the resolution of the detection system.
Details of the simple detector system used in the first experiments are given in ref. . This detector provided a detection efficiency in the 10-20% range; complete –and more expensive- cameras with 90% efficiency are however commercially available.