Cold amplifier @ Milano Bicocca 12 dic 2019 C. Brizzolari, P. - - PowerPoint PPT Presentation

Cold amplifier @ Milano Bicocca

12 dic 2019

• C. Brizzolari, P. Carniti, C. Cattadori, M. Citterio, A. Falcone, G. Franchi, N. Gallice, C. Gotti,
• A. Lucchini, M. Mellinato, G. Pessina, S. Riboldi, P. Sala, M. Tenti, F. Terranova, M. Torti

Original amplifier (TDR)

2

THS4131

Work of G. Cancelo (FNAL) and others

• Series (voltage) noise gives the dominant contribution
• THS4131 at 77K: we measured 1.2 nV/√Hz, similar to room T
• Low frequency noise 2x-3x larger w.r.t. room T
• Power consumption: 3 mA at cold (+/-2.5 V -> 15 mW)
• Limited dynamic range when driving 50 ohm lines

Original amplifier (TDR)

3

THS4131 measured at 77K

1.2 nV/√Hz at 100 kHz 10 nV/√Hz at 100 Hz

THS4131 at roomT

1.3 nV/√Hz at 100 kHz 3 nV/√Hz at 100 Hz

First tests at MiB

FBK array: 6x 4x4mm2 SiPMs

4

Cold amplifier

• Q1: SiGe bipolar transistor (BFP640) for lower noise
• U2: Fully differential opamp THS4531 for high open loop gain and differential outputs

(same family as the THS4131, see TDR)

• Flexibility in tuning gain-bandwidth product
• Lower power consumption
• Almost rail-to-rail outputs on 50 ohm load
• For further details: arXiv:1911.06562

Modified amp: schematic

5

Dynamic performance

6

• Gain-bandwidth product ≈ 7 GHz
• Flexibility to compensate the open loop gain by changing resistor values,

to maximize bandwidth depending on actual value of SiPM capacitance and closed loop gain

• Drives 12-meter 50-ohm lines (AC-terminated) with edges in the 20 ns range
• Output dynamic range almost rail-to-rail
• Total power consumption (Q1+U2+etc): 2.5 mW

12 m output cables 26-27 ns differential rising edges ≈15 MHz closed loop bandwidth

• Dominant contribution from series white noise
• Spectra at different operating currents of Q1
• At 77 K:

0.4 nV/√Hz above 100 kHz (1 nV/√Hz at 10 kHz)

Noise

7

0.37 nV/√Hz

• S/N measured at 3V OV (gain 2.4x106) with one SiPM + added capacitance to simulate ganging
• Yellow/red dots: measured for SiPMs with tau ≈ 800 ns/100 ns, 300 kHz low pass filter
• Yellow/red curves: calculated S/N from noise spectra and SiPM gain, 300 kHz low pass filter

→ curves agree with measurement (dots)

• Green/blue: S/N with optimum (matched) filter →

(For large capacitance the optimal S/N does not depend on tau)

Signal to noise ratio (1)

8

1 tile, ≈1 cm2, ≈ 4.8 nF

Screenshot unfiltered (BW ≈3 MHz, risetime ≈100 ns)

At the values of capacitance corresponding to 48x 6x6 mm2:

• S/N ≥ 4 with 300 kHz low pass, depending on tau
• S/N ≥ 20 with optimum (matched) filter, independent of tau

Signal to noise ratio (2)

9

48x 6x6 mm2 ≈ 86 nF (assuming 50 pF/mm2) 48x 6x6 mm2 ≈ 60 nF (assuming 35 pF/mm2)

1 tile, ≈1 cm2, ≈ 4.8 nF

Screenshot unfiltered (BW ≈3 MHz, risetime ≈100 ns)

• Same calculation, now with values expected from the next FBK batch of cryo SiPMs:

tau=400 ns, cap=53 pF/mm2, gain=2.4x106 (3 V overvoltage)

• Added the case of a low pass at 3 MHz (essentially unfiltered)

Signal to noise ratio (3)

10

48x 6x6 mm2 x 53 pF/mm2 ≈ 91 nF

Plans for the near future:

• Use a single power supply instead of dual +3 V / -1V [already tested with +3V, OK]
• Test H1164NL transformers for differential to single-ended conversion

(used in DAPHNE input stage) – in progress

• Fix the value of the supply voltage from the DAPHNE board (+3 V? +5 V?)

→ 3 V preferred from our side, lower power consumption, same performance

• Input voltage of DAPHNE is 1 V maximum (AFE5808)

→ set amplifier gain to have 2000 p.e. dynamic range in 1V (also depends on SiPM gain and overvoltage)

• Ensure compliance with grounding rules

Next steps

11