Photomultiplier Circuits (A2088)

© 2019, Kevan Hashemi, Brandeis University HEP


Amplifier-Shaper, Version One
High-Voltage Low-Pass Filter
SiPMT Head, Version One
Amplifier-Shaper, Version Two


We use assembly number A2088 for a family of amplifiers, shapers, high-voltage filters and photomultiplier (PMT) mounting circuits.

Amplifier-Shaper, Version One

[13-FEB-19] The A2088A is our first amplifier-shaper. Printed circuit board files in A208801B. The A208801A we aborted because we omitted C7 and C8.

We load two ERA-1SM into the circuit as shown. We measure gain versus frequency for various input powers.

Figure: Gain of A2088A with 2 of ERA-1SM, 3-dB Attenuator, and Increasing Input Amplitude. Plots are labelled in mVpp input amplitude.

We expect the gain to be +12 dBm for each amplifier, −3 dB for the attenuator, is +21 dB. We see up to 19 dB. The amplifier does not produce the +12 dBm output power we expect. We apply a 50-mVpp, 200 kHz input and observe the following output.

Figure: Output of Unmodified A2088A for 50-mVpp 200 kHz input. For 100-mVpp 40 kHz input see here.

For inputs greater than 20 mVpp the amplifier oscillates and its gain at the signal frequency is compromised. Our maximum output amplitude is around 300 mVpp when we expect 3 Vpp.

The shaper we configure after consulting our treatise on pulse shapers. We set L1 = 3.3 μH and C3 = 1 nF. We find that the 100-nF capacitors cause the output baseline to shift. When combined with a 50-&Ohm; signal impedance, thes capacitors introduce a 5-μs time constant. During a 100-ns pulse, we expect the baseline voltage to move up by 2% of the pulse height. We observe something closer to 10%. We place 1-μF capacitors in parallel with the 100-nF coupling capacitors and get the following pulse.

Figure: 40-mV 12-ns Pulse Shaped and Amplified. We have 1-μF coupling capacitors to minimize baseline shift.

[14-MAR-19] We work on the A2088A until it stops oscillating. We have 100-pF coupling capacitors, and for decoupling the power supply we have 100 pF, 100 nF and 1 μF capacitors scattered about. We have top-side ground planes made of copper tape. We have U1 as ERA-3SM and U2 as ERA-1SM, the attenuator is −3 dB, and L1 = 0 H, while C3 is omitted.

Figure=: The A2088A Modified Until It Stops Oscillating.

We apply 100 MHz input and obtain the following plot of output power and gain versus input power in dBm (decibels above 1 mW into 50 Ω).

Figure: Gain and Output Power verus Input Power for A2088A at 100 MHz After Modification.

Each of our four 100-pF coupling capacitors have impedance 15 Ω at 100 MHz, so each of them introduces a loss of 0.4 dB. The ERA-3SM and ERA-1SM should provide gain of around 34 dB. Our attenuator is −3 dB. We expect a total gain of around 29 dB. We observe 23 dB. We expect a maximum output of +12 dBm and that's what we see: 2.2 Vpp into our 50-Ω oscilloscope input.

We add 1-μF capacitors in parallel with all four 100-pF coupling capacitors, and two 10-μF capacitors in parallel with our existing decouplers. At 100 MHz with 70 mVpp input, our output is 1.9 Vpp for gain 29 dB. We measure gain versus frequency.

Figure: Gain versus Frequency for −20 dBm Input. Modified A2088A with ERA-3SM, 3-dB Attenuator, ERA-1SM, 100 pF and 1 μF couplers.

We load 3.3 μH for L1 and 1.0 nF for C3 to make our shaper. We obtain the following for a 200-mV input pulse.

Figure: 200-mV 12-ns Pulse (Top) and Shaped Output (Bottom). Output pulse is 600 mVpp.

The exact shape of the output pulse, and the amplitude of the small pre-pulse spike, depend upon the arrangement of our coaxial cables near the oscilloscope.

High-Voltage Low-Pass Filter

[01-FEB-19] The A2088B is a low-pass filter for a PMT high-voltage power supply. The high-voltage input, up to 3 kV, is low-pass filtered to remove noise, with corner frequency 4 kHz. SHV plugs receive and transmit the high-voltage power. The 10-kΩ series resistor will drop the high-voltage output by 10 V per 1 mA.

A side port with BNC socket provides a high-pass filtered view of the output high-voltage, with corner frequency 400 Hz. The filter proves effective in removing 100-kHz pulses from the high-voltage, and reducing PMT output noise by around 80%.

SiPMT Head, Version One

[11-FEB-19] The A2088C is a mounting circuit for a 6 mm × 6 mm silicon photomultiplier (SiPMT), part number MICROFC-60035. Printed circuit board A208801C.

The A2088C follows the SiPMT Breadboard, shown below, which provides connection to the SiPMT anode and cathode to measure dark current.

We use the above breadboard to measure dark current versus bias voltage.

Amplifier-Shaper, Version Two

[20-MAR-19] Our second amplifier-shaper has four copper layers. On the top layer are signals. On the first middle layer is the ground plane. On the second middle layer is a power plane. The bottom layer is for test points.

We add more attenuators to give us more freedom to alter the total gain of the circuit, and optimise its dynamic range for large and small inputs. Dual coupling capacitors expand the bandwidth down to 3 kHz, while at the same time providing reliable coupling to the attenuators at frequencies 1-10 GHz, where oscillations might otherwise arise. The resistors R9 and R10 allow us to match the input of the shaper more exactly to the PMT signal cable, so we can reduce the reflections that come back from the high-impedance PMT output. Inductors in the bias impedance of the two amplifiers further restrict the penetration of 1-10 GHz oscillating currents into the power supplies, reducing coupling between the two stages at these frequencies.

[25-MAR-19] The A208801B printed circuit board has four layers: top, ground, power, and bottom. The bottom layer we use only for pads and vias, and a token track to make identify the layer as a positive image. Printed circuit board A208801D, shown here.

[24-APR-19] We have our first assembled Amplifier-Shaper A2088D. We have U1 = U2 = ERA-3SM, R1-R4 3dB, L3 = 3.3 μH and C9 = 1.0 nF. Looking at the response to a 20-mV 12-ns pulse, We see signs of oscillation on OUT, and gain varies by a factor of two as we touch the circuit board. We replace L2 = 100 nH with L2 = 0.0 nH. The oscillations stop and the gain is stable.

[10-JUN-19] We report on the linearity and dynamic range of the V2 shaper in V2 Shaper-Amplifier Performance.

[26-JUN-19] The A2088D with C8 = 1.0 nF and C7 = 1.0 μF produces a sustained oscillation of 200-300 MHz at its output with amplitude of order 10 mVpp. We remove C8 and the output noise drops to 2 mVpp, with no obvious oscillations. The 1.0 μF ceramic capacitor in place of C7 is the CGA4J3X7R1H105K125AB with self-resonant frequency 5 MHz and equivalent series resistance 1 Ω at 100 MHz.

We mount the A2088D in a metal enclosure with plastic end-caps. Note the hole for delivering power with the two-way cable-mounting socket. Connecting the power the wrong way around will not damage the circuit, thanks to diode D1.

We define the following versions of the A2088D, each named after the length of the pulse it emits for an ideal input pulse of 1 ns.

A2088D-0No Shaper, L3 = 0 nH, C9 = 0 nF
A2088D-2020-ns Shaper, L3 = 100 nH, C9 = 1 nF
A2088D-5050-ns Shaper, L3 = 3.3 μH, C9 = 1 nF
Table: Versions of the A2088D

We measure gain versus frequency for 10 mVpp sinusoidal input for four A2088D-0 and compare to the ZHL-3A, which should provide flat gain up to 150 MHz. We are using a 300-MHz analog scope to measure output amplitude, and a 160-MHz function generator to provide the input.

Figure: Gain of A2088D-0 with versus Frequency. Measurements by Xinfei Huang.

We apply 12-ns, negative-going pulses to the input. We increase their height and record output pulse height.

Figure: Output Amplitude versus Input Amplitude for A2088D-0. Measurements by Xinfei Huang.