Future FONT BPM ProcessorsC.Perry 25 June 2009

Types of Processor Two types of processor:a) present mixer typeb) baseband type

Both will be made because:- baseband type should be simpler to use and adjust- but it will have marginal resolution at low bunch charge- and machine may have to operate at low Q to help get 37nm spot size

General Details Both types will have a number of features in common:- physical form: '6U VME format' module to mount in a subrack- channels:each module will have two channels, for X and Y of one BPM- normally, the X sum output will be unused- monitoring:sum and diff signals will be duplicated on isolated outputs- stripline phasing: all stripline signals will be taken via mechanical phase adjusters

Mixer Processor(1) This will be similar to the present, but:- high-level mixer to improve linearity- improved filters (giving some reduction in delay)- low-noise amplifiers on outputs- LO phasing to each mixer separately adjustable on the module using IQ modulators controlled by DACs- low-level LO input, with all necessary amplifiers on the module- provision for monitoring and tracking the LO phase drifts- provision for monitoring and tracking the digital processor sampling clock drifts

Mixer Processor(2) Monitoring and tracking LO phase drifts:- objective: to allow drifts of LO phase relative to beam to be observed and corrected, without interference to normal operation- basic method:- the X-sum channel will be set in quadrature- output added into Y-sum output after a delay- read by Y-sum ADC 3 or 4 samples after peak- possible variant:- drive channel with split Y-sum signal as now - no advantages, and slightly messiernote: this does not extend to relative drifts within BPM and processor, these phases are assumed to be set by a separate calibration procedure and to be stable thereafter - the IQ modulator/DAC should be stable indefintely to < 1 deg

Mixer Processor(3)Monitoring and tracking sampling clock drifts:- narrow system bandwidth to widen the output pulse slightly- tweak filters to improve symmetry of output pulse- look at difference between Y-sum output samples before and after main sample- this slightly increases delay (by about 1.5ns)- less than has been saved by tweaking filters- could keep full bandwidth, but this reduces sensitivity

Baseband Processor (1) Concept:- pass bipolar pulse output from BPM through a LPF- this converts the output to a slower bipolar pulse that can be measured directly- penalty is that the amplitude is very much reduced

Advantages:- only one adjustment: to sample on the peak- no LO is required- it is inherently linear- it is simpler

Baseband Processor (2)Disadvantages:

- bandwidth must be low enough that:- output can be accurately sampled on the peak- a high-quality 180 degree hybrid can be obtained

- 120MHz (Bessel response) bandwidth has been demonstrated with good hybrid performance and looks safe for sampling

- this means that even with a low-noise amplifier, sufficient s/n for 1um resolution needs a bunch charge ~6E9 or more

- some increase in b/w and improved s/n may be possible:- going to 200MHz might be possible- this gains a factor of 2 in resolution, or in bunch charge- but it seems probable that this will start to compromise performance in other ways

Baseband Processor (3)Monitoring and tracking sampling clock drifts:- delay the X-sum signal, invert, and add into the Y-sum output- this adds a second bipolar sum pulse after the main one- with the correct delay, a sampling point will fall at the zero crossing- 3 samples after the peak sample- variants:- could split the Y-sum instead of using the X-sum- messier, and no benefit- could make the sum filter slower, and add-in without a delay- simpler, but less sensitive and less stable

Baseband Processor (4)Caution: - it is assumed that:- the X and Y sum and diff outputs are accurately aligned in time- the 3 ADCs in the group serving one BPM samplesimultaneously- this should be assured by construction- can supplement by (permanent) trimming of the hardware in the lab- only the overall sampling phase for each group of 3 ADCs is adjustable- this also applies to the mixer processor, but sampling is less critical