1. Status of TTF HOM Project Aug 9, 2005
Project Scope: Instrument both couplers for all SC cavities in the TTF2. (80 channels).
System expected to operate as BPM system
Expect <~few micron resolution (train average)
Expect <~30 micron resolution (bunch to bunch)
Expect readout rate ~1 Hz for all BPMs (5Hz May be possible)
Want full integration into TTF2 control system.

5. Hardware Status – Downmix Chassis
First Prototype board completed / stuffed Aug 5.
Preliminary tests of Gain, Noise, linearity “look good”, should have numbers in a few days.
Board designed for machine assembly
Parts ordered
In principal, just ship parts to vendor for assembly
In practice, probably more involved that that, but no serious problems expected
Chasses now simplified due to lower board power consumption (~5 W / board).
5 Chasses + spare, to be assembled at SLAC
No problems anticipated

6. Other Hardware
VME crate / digitizers: All commercial parts, but need to decide on control hardware.
LO source: Synthesizer, amplifier, Fan-out: Straightforward.
Calibration reference: Must produce ~1700 MHz tone, locked to harmonic of beam with ~1 picosecond stability
Not designed yet (ideas only) – technically challenging.
Needs to use TTF triggers, timing signals to produce clean reference

7. Timing
The software / hardware must allow the HOM signals measurements to be correlated with specific conventional BPM measurement pulses.
HOM system probably will not run at full 5Hz beam rate, so need method to identify which pulses are measured

8. Software
This probably represents >3/4 of the entire project effort!!!
Software architecture determines which VME hardware to purchase
Architecture will depend on which group does which part of the work (software can have a steep learning curve).
Need to make decisions soon!

9. Software Components
Data Download: 10, 8 channel, 100Ms/s (12 or 14 bit) VME digitizers.
For 1 millisecond, get 8M samples / pulse, Can produce up to 80MByte/second data rate.
Can run slower, but then this is limits bpm update rate
Data Digital Downconversion, filtering to produce I,Q data for each bunch. Matrix multiply (small) to produce X,Y for each cavity.
Very approximately 1 Gflop maximum processing required. ~1Gbyte memory.
DOOCS Server: 1000 pulses, with X,Y, X’, Y’ for 40 cavities, at 5Hz beam rate.
Total possible rate 800KWords / second.
Calibration Software: Same concept as existing HOM system but much more complex.
Must move correctors, read conventional and HOM BPMs throughout machine.
Use MIA, and model dependant analysis to determine cavity BPM calibration coefficients, download to DOOCS server.
Does not need real time, but calculations can be complex.
Probably have 1 computer as front-end, separate computer for calibration.

10. Front End Data Download,Digital Down Conversion, DOOCS server.
2 Reasonable Options
Standard DOOCS, VME crate controller (Force SPARC-56 CPU)
Easy integration with TTF2
Easy Support, (need new driver for digitizers)
Low performance (250 specfp2k)
SIS VME-PCI bridge
Existing drivers for LINUX pc from SIS
80MB/sec transfer demonstrated by SIS
~1600 specfp2k for PC.
VME-2e
Unique hardware – difficult support

11. Calibration Computer
LINUX PC with Matlab probably the most reasonable solution.
Interface with front end computer through DOOCS.
Are large data transfers (1GB) in DOOCS OK?
If we use a PC (PCI-VME bridge) for front end, can probably use same computer for calibration.