Power supply lt3571

Simple SiPM/plastic scintillator based beam loss monitors

Introduction

This repository contains the hardware design for a kind of SiPM/plastic scintillator based beam loss monitor. The design is based on and borrows the name (CLAWS) from an earlier design employed during commissioning of the SuperKEKB collider. A related thesis can be found here. CLAWS was supposed to be a abbreviation for some longer name but it surely wasn't a very elegant one, neither can I seem to remember it. Anyhow, as a short name is quite practical, we kept on calling the system CLAWS.

The sensors later have been re-purposed to detect beam loss and trigger a beam abort if there is too much of it. This application does not require very precise sensors. Further, opposed to the original design, now a larger dynamic range was desirable. To also allow for measurement of beam losses which occur for longer than a few microseconds, DC coupled analog electronics are required. The design published here was created to implement the required changes outlined above and create a system better suited to the current application at the SuperKEKB collider.

Disclaimer

The circuits here have been designed by an at the time undergrad student (me) who had no real experience in electronics or electronics CAD. I am sure there are many aspects that I would do differently now, be it related to the circuit itself or how it is described in KiCad. I am however of course open for suggestions to improve the design (although I cannot guarantee I will find the time to implement them myself).

Overview

There are two parts to the detector system: The detectors, which are to be placed where the beam loss is expected to occur and the receiver board, which should be placed near the digitizer, which is usually positioned outside the radiation controlled area of the accelerator complex.

The following is an illustration of the complete system. The individual components are described in detail below.

Detector Board

The detector (sometimes also referred to as the sensor) is based on a plastic scintillator (BC-408 or similar material), light from which is detected using a Hamamatsu SiPM (or MPPC). The SiPM is located in a small dimple in the bottom of the scintillator. The scintillator is wrapped in an appropriately sized reflective film (M3 DF2000MA or similar) to make sure as much as possible light reaches the SiPM. This is the same as what was used with the earlier versions of CLAWS and can be traced back to what is described in this paper. The wrapped scintillator is fixed to the PCB using black aluminium tape, which also serves to shut out any light from the environment, which could overwhelm (and potentially damage) the SiPM (although I've also heard from people leaving SiPMs turned on under ambient light all the time without problems, so not sure how true that is...), but mainly would simply interfere with a beam loss measurement as the signal would be much larger than the beam loss induced one.

The signal obtained from the SiPM is turned into a differential signal using the LMH6553 fully differential amplifier. The gain of this amplifier may be set by choosing appropriate resistor values in the feedback loops of the circuit. This amplifier then drives a differential pair in an ordinary Ethernet cable. However, usually only CAT6A or CAT7 cables are used, as they usually provide superior shielding, which even though a differential pair is used, appeared to help with environmental noise picked up by the cable.

The required voltages to drive the amplifier circuit as well as the SiPM bias voltage are delivered to the sensor using the same Ethernet cable. The amplifier here is configured with a single 10V power supply. The voltage in the cable should be slightly above 10V, as it is regulated by a LDO regulator on the detector board. 10V is the maximum supply voltage supported by the LMH6553 and is chosen to maximize the dynamic range at the amplifier output (which depends on the supply voltage).

One twisted pair in the cable remains unused for now but could be used for some sort of double sensor setup which two sensors at the end of a single cable (could be interesting if one wishes to assert coincidence of a signal in two sensors to reject noise).

Receiver Board

The differential signal provided by the detector board is received by and properly terminated at the receiver board. It contains a circuit using the same LMH6553 fully differential amplifier, which provides further amplification of the received signal, as well as the option to convert it to a single ended signal. Both inverted and non-inverted signals are supported. The not used side must be terminated using a 50 Ohm resistor (which can however also be mounted directly on the board while leaving the corresponding output connector unpopulated). There also is the option to further use both outputs at the same time to obtain a differential signal.

The receiver amplifier circuit is, opposed to the one on the detector board, arranged in a symmetric configuration. This is done to avoid a large DC bias at the output, which poses no problem at the output of the detector board, as it is removed by the receiver amplification circuit anyways. The symmetric configuration necessitates +5V and -5V supplies (for both of which there are LDO regulators on the board).

Presumably due some unavoidable variations in the used components (resistance values etc.), the circuit is never perfectly symmetric and thus a small DC bias remains. The same bias appears at both outputs (inverted and non-inverted). To compensate this, the common mode output voltage control pin of the LMH6553 is used, which is accessible through a potentiometer on the board. As the bias is present at both outputs, one can only ever tune the circuit to have one of them centered perfectly around zero. This however should pose no significant problem, as in most use cases one probably only uses one of the two outputs, or uses both to get a differential signal, for which a small DC bias does not matter.

The Receiver board further also distributes the voltages required by the detector boards. These voltages as well as those needed by the receiver board itself are supplied externally through connectors on the side of the board. There further is the option to daisy-chain multiple receiver boards and share the supply voltage rails between them. The voltage routing on each board is to some degree configurable using a set of jumper pins. As regulators are used for all but the SiPM bias voltage, the low voltages provided can be chosen rather freely, as long as they fall within the ranges supported by the LDOs . However, testing showed that if they are on the high side, the LDOs do get rather hot. Whether this is still within the supported temperature range was not (yet) verified.

There is also an option to use the positive voltage (before the LDO) for the receiver board amplifier as the positive voltage for the detector board amplifier. This turned out to be the most convenient configuration as it allows the setup to be powered off a +/-12V double rail supply (next to a supply for the SiPM bias voltage).

To share voltages between receiver boards, a 10-pin ribbon cable is used. The boards may be arranged in a stacked configuration, in case of which the connector on the bottom must be installed and used, or a flat configuration with a single long ribbon cable with multiple connectors. There is also the option to add further boards of the same form factor which for example could contain a high voltage power supply to provide the SiPM bias voltage, or DC/DC converters to provide the other supply voltages.

SiPM Bias Voltage Supply

A power supply board as mentioned above for the SiPM bias voltage generation was designed based on the LT3571 (see kicad/power_supply_lt3571_single/) There is also a four channel version which however does not support board daisy-chaining but can be controlled using an Arduino-Nano/STM32-Nucleo form factor microcontroller board through a DAC. The single channel/manual voltage control version was found to be much easier to use in the end so the four channel option was not pursued much further (keep it simple and stupid).

The design appears to function alright, but does not provide an absolutely clean voltage. This to some extend is reflected in the signal form the SiPM. However, for the intended applications the signal does not have to be extremely precise to begin with and the introduced noise was found to be negligible.

The board also does indeed emit some EMI, so care must be taken on where to position it. There likely are possibilities to improve this design. The board can be run off the same +12V as used as an input to the receiver board. The output voltage for the single channel version is set using a potentiometer which is positioned in the same spot as the offset adjustments for the receiver boards so a universal enclosure can be used.

KiCad

The design is entirely drawn in KiCad. I like to use the newest version. The files in this repository are for version 9.

Libraries

There is are custom symbol and footprint library containing all the parts not present in the standard KiCad libraries. The provided 3D models are managed with git LFS.

Simulation

Some of the symbols reference SPICE models in the ${KIPRJMOD}/../lib/spice_models/ directory. Those are not present in this repository as I am probably not permitted to distribute them. If simulation is required, please acquire the models directly from the manufacturers.

KiBot

KiBot can be used to automate creation of gerber files, export circuit diagrams and assembly documents or similar. The included configs are those which I used when producing the boards at use for the SuperKEKB abort system. Adjust to taste.