At the last commissioning meeting, J.-C. Denard passed out the IRFEL BPM requirements. This note considers the requirements as far as FEL commissioning and operation goes.

The optical mode size and divergence at 3 microns in the IR Demo are a 1/e^2 radius of 600 microns and a divergence of 1.5 mrad. George Neil has already pointed out the difficulty of matching the electron beam to this mode. The matching will be done using the wiggler viewers and by tweaking on the spontaneous signal as seen through a spectrometer. The BPMs on either side of the wiggler will be used to verify the electron beam stability during commissioning and will allow one to reproduce the beam position from day to day so that one does not have to start from scratch each time. The electron beam must be reproducibly place in the wiggler to a positional and angular accuracy at least a factor ot three smaller than the radius and divergence. Regarding the IRFEL BPM requirements, I therefore have a few comments.

1. As Dave Douglas pointed out, a measurement rate of only 1 Hz is quite difficult to deal with. It has been pointed out that the limitation on this rate is machine protection. I would propose threading the beam and doing the initial restore at a low micropulse repetition rate from the injector but at 10 Hz if desired by the operators. If we use full charge, we could still reduce the average beam current to less than 1.5 microamp by operating at 2.339 MHz with a 250 microsecond pulse at 10 Hz. The average current during the pulse would be 313 microamps. The long term average current would be 780 nA. This would increase the necessary dynamic range to 16 but this does not seem to be that large considering the large signal available. We could increase the safety of the system by threading the first few pulses at 1 Hz and then switching to 10 Hz when it is clear that the beam is not being lost along the way. The risk of burn through is negligible with such a pulse even at 10 Hz. This is one tenth the energy per macropulse used in the Vanderbilt and Duke accelerators (three vs. thirty Joules per macropulse). They have never burned through any chamber despite having a smaller beam emittance, a shorter macropulse, and no machine protection interlocks.
2. The beam must be reproducible to approximately 200 microns from day to day at any given current so that we don't have to use the viewers in the wiggler to set up the FEL for lasing each day. This reproducibility allows the gain to be reproduced to within a factor of two. This should be enough to get the laser started and then the performance can be tweaked from there.
3. It would be nice to have a current dependence less than 200 microns for the following reason. We would like to center the beam on the viewers at low current. One would then raise the current by increasing the repetition rate from the injector. If the indicated beam position moves by greater than 500 microns one would not know whether the beam had actually moved (and this much movement would prevent lasing) or whether the beam was actually in the same place but only indicated a different location. We cannot use the viewers at 5 mA since they keep the beam from being recirculated. Synchrotron viewers may settle the question of beam movement but on ly if the betatron phase advance is correct between the SR monitor and the BPM in question. Quads could also be used to check the position and, as suggested, adjust the calibration to match the current.
4. For slow orbit lock the beam position resolution must be as good as possible since the laser power will be sensitive to beam movement as small as 50 microns (Bill Colson can and probably should quantify the power change vs. position and angle). If some averaging or signal conditioning can be used to enhance the orbit lock stability by a factor of two, it would be quite nice.