An Ultra Low Noise AC Beam Transformer for

An Ultra Low Noise AC Beam Transformer for Low Intensity Beams – for AD (& Elena? ) For Diagnostics and Deceleration Flemming PEDERSEN [Carlota GONZALEZ VAZQUEZ ] Alan FINDLAY Maria Elena ANGOLETTA

Outline • • • CERN AD (Antiproton Decelerator) CERN ELENA (Extra Low ENergy Antiprotons) Beam Transformer and Head Amplifier Transformer and Amplifier Noise Sources Low Noise Amplifier Design Feedback Resistor and Circuits Low Frequency Version Transformer in Beam Transfer Line LTI Coax Cable developed Can a broad band electrostatic device (à la AD ring PU S signal) achieve similar or adequate performance? April 2012 Ultra Low Noise AC Beam Transformer Slide 2

CERN AD (Antiproton Decelerator) • • • Collection, Cooling, and Deceleration of antiprotons Momentum 3. 5 Ge. V/c down to 100 Me. V/c Intensity: few 107 charges: current 0. 2 - 15 m. A DC beam transformers (Unser-trons) are not sensitive enough Bunched : use RF current (at two harmonics) for intensity, f. RF = 0. 17 - 1. 6 MHz, Bunched: Use full bunch bandwith signal for bunch length and longitudinal emittance measurements (tomoscope) Bunched: Use fundamental component for RF system phase loop Unbunched: Use longitudinal Schottky scans for intensity and momentum distribution For more than 13 years, this device has been the workhorse of longitudinal diagnostics and intensity measurements in the AD April 2012 Ultra Low Noise AC Beam Transformer Slide 3

CERN ELENA (Extra Low ENergy Antiprotons) • • • Cooling, and Deceleration of antiprotons Momentum 100 Me. V/c down to 13. 7 Me. V/c Intensity: few 107 charges: currents comparable to AD: b and circumference lower by similar factor DC beam transformers (Unser-trons) are not sensitive enough [. . but SC DCCT developed in several labs outside CERN] Bunched : use full bandwith signal for RF based intensity measurements (integration of bunch areas, CERN BE/RF receivers now available) Bunched: Use full bunch bandwith signal for bunch length and longitudinal emittance measurements (tomoscope) Bunched: Use fundamental component for RF system phase loop Unbunched: Use longitudinal Schottky scans for intensity and momentum distribution as in AD The initial conceptual design (2007 ELENA report) was based on the use of a similar ultra low noise beam transformer in ELENA April 2012 Ultra Low Noise AC Beam Transformer Slide 4

Applications of Low Noise Transformer Applications of this new PU (0. 3 - 15 MHz) in AD: i) RF phase PU during deceleration for RF system ii) Intensity during deceleration (using two rev. harmonics) iii) Bunch lengths iv) Intensity and momentum distribution while debunched (Schottky) S/N ratios at optimum Schottky harmonic is about 10 – 50 d. B at different points in the AD cycle (measured and calculated) It is more than 30 d. B better than the old 42 MHz Schottky PU: Noise current density near 1. 5 MHz: v) Single shot bunch intensity measurement in transfer line using LF version (0. 02 – 3 MHz). Resolution few 104 charges; calculated but not achieved due to current analog signal integration. April 2012 Ultra Low Noise AC Beam Transformer Slide 5

Low Noise Longitudinal Beam Transformer April 2012 Ultra Low Noise AC Beam Transformer Slide 6

Doubly Shielded Copper Cavity April 2012 Ultra Low Noise AC Beam Transformer Slide 7

Cavity Assembly (LF) April 2012 Ultra Low Noise AC Beam Transformer Slide 8

The Low Noise Feedback Principle • Make the pick-up resonant with as high a Q as possible to reduce thermal noise current of the transducer itself • Use a low noise high impedance amplifier (Si JFET's) and regain broad band properties by strong active feedback WITHOUT deteriorating the signal to noise ratio (inject feedback current through a small capacitor) • The transimpedance amplifier made this way has an extremely low 'noise temperature' in a certain frequency range (< 0. 5 K) April 2012 Ultra Low Noise AC Beam Transformer Slide 9

Transformer and Amplifier Noise Sources • • • N is transformer step-up ratio Ib is the desired signal Vna [V/Hz 1/2] is the amplifier voltage noise Ina [A/Hz 1/2] is the amplifier noise current Inth = [A/Hz 1/2] is thermal noise of the pick-up shunt impedance Rp, so make Rp as large as possible (high Q) • Transform Ca, Vna and Ina to primary of transformer • Increasing N lowers voltage noise, but increases current noise and capacity April 2012 Ultra Low Noise AC Beam Transformer Slide 10

Transformer and Amplifier Noise Sources • The amplifier voltage noise Vna/N can be transformed into an equivalent current noise by the relation: Ivna = Vna/N Zpu • The total equivalent, input-noise current is then: if the 3 noise sources are uncorrelated (not quite true for Ina and Vna) • If the amplifier noise is sufficiently small, the signal to noise ratio is dominated by the noise from the shunt impedance and is constant over a wide frequency range April 2012 Ultra Low Noise AC Beam Transformer Slide 11

Equivalent, Input-noise Currents (gap) April 2012 Ultra Low Noise AC Beam Transformer Slide 12

Amplifier input stage considerations • Bipolar transistors may have low voltage noise, but current noise much larger than for FET transistors due to the shot (Schottky) noise: where Ibase is the base bias current • A FET transistor has a voltage noise: where gm is the transconductance [A/V] and a current noise induced in the input circuit by thermal noise of the channel through the distributed gate channel impedance: where Cc is the coupling capacitance: ~ 2/3 of total gate-channel capacity Cgs. • FET noise quality is thus characterised by the gm/Cgs which is also the high frequency figure of merit f. T: best ratio between current and voltage noise is found in FET transistors with high f. T April 2012 Ultra Low Noise AC Beam Transformer Slide 13

Choice of FET transistors • Ga. As FET: very high f. T, but also high 1/f (flicker noise) corner frequency: > 10 MHz • Si MOSFET: medium f. T but somewhat better 1/f corner frequency: 1 - 5 MHz • Si JFET: similar f. T to Si MOSFET, but much lower 1/f corner frequency: ~10 Hz for Philips BF 861 C. By far the best choice in the interesting frequency range: 0. 01 - 20 MHz • Monolithic amplifiers with JFET inputs are available, but not with very low voltage noise and a high gain-bandwidth product (feedback!) April 2012 Ultra Low Noise AC Beam Transformer Slide 14

Amplifier Design • • • 16 parallel Si JFET's with high gm and ID to obtain low voltage noise (0. 25 n. V/Hz 1/2) - bipolar cascode to obtain low miller capacity and high bandwidth. Total delay ~ 5 ns. Bipolar 2 nd stage to further amplify with increasing equivalent input noise and prevent the rather noisy 3 rd stage from adding noise Class B output current mode feedback video op-amp for high slew rate, high output swing and wide bandwidth (> 100 MHz, AD 811) April 2012 Ultra Low Noise AC Beam Transformer Slide 15

Feedback Resistor and Circuits • • Although the resonant high Q transformer has a good signal to noise ratio, it has drawbacks: i) Highly resonant response ii) Amplifier saturates at very low current near resonance The response can be made broad band if a negative current feedback resistor is introduced around the amplifier This converts the high input impedance amplifier into a low resistive impedance Rfb/G : the response becomes broad band If Rfb and G are sufficiently large, the equivalent input noise current due to Rfb can be made negligible compared with the Rp noise current: Rfb > 16 N Rp April 2012 Ultra Low Noise AC Beam Transformer Slide 16

Feedback Resistor and Circuits • • There are several disadvantages from such a high feedback resistor (400 k ): i) difficult to implement at high frequencies (stray capacity) ii) limited dynamic range: Iin, max = Vout, max. N/Rfb large amplifier gain (10'000!) required, short delay difficult Much better performance and easier implementation is obtained by injecting the feedback current through a small capacitor combined with a 90 degree phase shift (integrator) By making Rint small and Cint >> Cfb the effective noise temperature of the feedback resistor is dramatically reduced April 2012 Ultra Low Noise AC Beam Transformer Slide 17

Low Frequency Version • To extend the Bandwidth towards lower frequencies (limited by Vn and Lp): • Use higher m ferrites: 200 -> 1200 • Use higher step-up ratio: N = 2 -> 4, this lowers • Transformer installed adjacent to high frequency device and signals combined (crossover 1 MHz) April 2012 Ultra Low Noise AC Beam Transformer Slide 18

Equivalent, Input-noise Currents (gap) April 2012 Ultra Low Noise AC Beam Transformer Slide 19

Performance • Longitudinal Schottky S/N ratios are typically 10 – 50 d. B with N = 5 E 7 at best frequency (~ 1. 5 MHz) • At higher energies, 300 Me. V/c and above, we use much higher harmonics to reduce Schottky acquisition time and statistics • The massive EMI shielding works as intended, main EMI source is output cable April 2012 Ultra Low Noise AC Beam Transformer Slide 20

Calculated vs. Measured Noise levels • Measured noise currents are within 20% of calculated values (except low frequencies < 10 k. Hz for LF version) April 2012 Ultra Low Noise AC Beam Transformer Slide 21

Absolute Intensity Calibration • • • Absolute transconductance gain Y(w) = Vout/Ibeam = 100 , 500 , 4 k , 20 k (4 selectable gains, measured by network analyzer) Cross check of Schottky intensity (prop Y 2) vs. bunched beam intensity (prop Y) Intensities measured in injection line (TFA 7049) and ejection line, which can be charge calibrated by a capacitor discharge. April 2012 Ultra Low Noise AC Beam Transformer Slide 22

Single Pass Beam Transformer in Transfer Line • In CERN AD: Fast extracted antiproton bunch, few 107 charges, length ~100 to 300 ns. • Integrate current output during 1 ms • Bandwidth required: 3 k. Hz (18 k. Hz) to 3 MHz [to avoid errors from response limits]. Use LF version • Charge noise given by (graph next slide. P: • Expected RMS fluctuation: 1. 9× 104 charges (4 s is 7. 6× 104 charges) April 2012 Ultra Low Noise AC Beam Transformer Slide 23

Single Pass Beam Transformer in Transfer Line April 2012 Ultra Low Noise AC Beam Transformer Slide 24

Parasitic Noise Sources – The double Cu shield of the pick-up cavity and head amplifier worked as expected: no parasitic noise observed into input of head amplifier – Some un-explained 1/f noise observed below 10 k. Hz in the low frequency version: actually scales as 1/f 3 [AD 811? Voltage regulator noise? ] – Main parasitic noise is medium wave radio noise picked up on 100 m OUTPUT CABLE from ring to ACR due to Zt (transfer impedance) of CERN standard coax C 50 -6 -1 (Zt ~5 m /m @ 1 MHz). This in spite of 63 d. B head amplifier gain!!! – New cable C 50 -6 -2 developed in industry (Draka) as a consequence of this (Zt ~ 0. 06 m /m @ 1 MHz): worked as expected when installed in 2001. April 2012 Ultra Low Noise AC Beam Transformer Slide 25

Coax Transfer Impedance April 2012 Ultra Low Noise AC Beam Transformer Slide 26

New CERN (Draka) LTI construction – C 50 -6 -2 Coaxial following CERN Specification 461 REV. 5. 1 S-02 Y(St)C(m. S)CH 2. 6/7. 3 (C-50 -6 -2) April 2012 Ultra Low Noise AC Beam Transformer Slide 27

New CERN (Draka) LTI construction Construction S-02 Y(St)C(m. S)CH 2. 6/7. 3 Inner conductor copper wire, bare mm 2. 6 Foam-PE Insulation mm Outer conductor 7. 25 Al-PETP-Al-foil + copper braid, silvered Wrapping PETP-foil Magn. Screen 1 layer of iron tape, highly permeable Wrapping PETP-foil Static screen copper braid, silvered FRNC ( Flame Retardant Non Corrosive Copolymere) Sheath April 2012 mm 11. 5 Colour brown or black Ultra Low Noise AC Beam Transformer Slide 28

New CERN (Draka) LTI electrical properties Transfer impedance at 10 k. Hz m /m nom. 0. 8 1. 0 0. 1 MHz m /m nom. 0. 150 0. 180 1 MHz m /m nom. 0. 06 0. 08 10 MHz m /m nom. 0. 250 100 MHz m /m nom. 2. 0 2. 5 G xkm ³ 10 Test voltage KVrms 3. 0 Operating voltage KVrms 1. 2 Insulation resistance April 2012 Ultra Low Noise AC Beam Transformer Slide 29

Adaptations required for ELENA version of AD LPU • • • Few modifications required to permit baking at 200 degr C: higher temperature solder (Sn. Ag @ 220 degr. C), ceramic SMA feedthrough? Could perhaps be raised to 250 degrees if required: FFEP coax cable -> PTFE cable (100 x less radiation dose [but not an issue in Elena], lower b = 0. 7 vs. 0. 83 for FFEP) AD vacuum chamber ID is 146 mm, so either Conflat transition pieces (increased length) to Elena ID of 80 mm or a mechanical redesign will be required In case of mechanical redesign due to smaller diameter, a shorter ferrite length should be investigated to save space Should the HF & LF pick-ups be mounted back to reduce gap to gap separation? Electronics obsolescence: A study has been made, and only a few (non-critical) components need to be replaced when rebuilding, but new PCB of both HF and LF versions required. April 2012 Ultra Low Noise AC Beam Transformer Slide 30

Broadband Electrostatic Device Optimum noise matching • • Optimum FET amplifier noise matching for a capacitive current source Applies also to ferrite loaded beam transformer at high frequencies Signal source is a current source in parallel with a source capacitor Cs FET amplifier with n parallel amplifiers. Each amplifier: – Equivalent input noise current of Ina [A/Hz 1/2] – Input capacity Ca [F] – Equivalent input noise voltage density of Vna [V/Hz 1/2] April 2012 Ultra Low Noise AC Beam Transformer Slide 31

Broadband Electrostatic Device Optimum noise matching • Convert the noise voltage into an equivalent noise current: • Total noise current: • There is an optimum value for n: • which for zero noise current implies n. Ca = Cs April 2012 Ultra Low Noise AC Beam Transformer Slide 32

Broadband Electrostatic Device Optimum noise matching • For the specific case of FET, equivalent input voltage and current noise are given by: • Cc is coupling capacity, Ca is gate capacity, gm transconductance • optimum found when amplifier input capacity n. Ca is slightly smaller (83%) than the source capacity Cs and independent of frequency. April 2012 Ultra Low Noise AC Beam Transformer Slide 33

Electrostatic broadband PU elementary charge wave shape April 2012 Ultra Low Noise AC Beam Transformer Slide 34

Calculated Schottky S/N ratios for a broadband electrostatic longitudinal PU • • PU length limited by desire to measure shortest bunch length accurately (about 20 cm? ) With no amplifier connected, voltage wave shape DC value above baseline and low frequency Fourier components independent of length (. . but lower length reduces optimum noise matched voltage noise) Unlike low noise beam transformer, there is no strong dependence on particle velocity bc Optimum noise matching important to achieve useable S/N ratios Best S/N ratio at lowest harmonics due to width of Schottky lines. This results in long acquisition times, poor statistics. Typical S/N ratios for N = 1 E 7, and nominal ELENA longitudinal emittances are about 20 d. B. Marginal S/N ratios for N = 1 E 6 and/or larger longitudinal emittances Proposed parameters very close to existing electrostatic AD S PU (L. SØBY) April 2012 Ultra Low Noise AC Beam Transformer Slide 35

Calculated Schottky S/N ratios for an electrostatic long. PU (100 Me. V/c) April 2012 Ultra Low Noise AC Beam Transformer Slide 36

Calculated Schottky S/N ratios for an electrostatic long. PU (35. 7 Me. V/c) April 2012 Ultra Low Noise AC Beam Transformer Slide 37

Longitudinal PU options 1) 2) 3) Use existing AD LPU design modified as required, vacuum chamber diameter, bakeable, possibly shorter length. Solid operational performance and excellent S/N ratios Replace low noise beam transformer with optimized electrostatic broadband PU (1 k. Hz – 20 MHz) with improved EMI shielding. Drawback is sensitivity to lost particles, and poorer S/N ratios in some scenarios (large EL and/or low intensity) As 2) but supplemented by an low noise (normal conducting) resonant Schottky to improve S/N ratios and enable the use of higher harmonics for faster acquisitions and better Schottky statistics April 2012 Ultra Low Noise AC Beam Transformer Slide 38

What’s next? Choices? • Side by side calculated S/N ratios should be made for AD LPU (low noise beam transformer) and existing AD S PU • Calculated S/N ratios should be confirmed with existing electrostatic AD S PU in AD • The possible use of noise optimized broad band electrostatic PU’s in the transfer line for intensity (. . and position? ) • A longitudinal PU design report should be written during the summer of 2012 comparing the various options (FP + collaborators) April 2012 Ultra Low Noise AC Beam Transformer Slide 39