Algorithms and a User Interface for Tuning a Multi-stage SQUID Amplification System Project Description Final Revision April 11, 2005 Computer Science Engineering Senior Benjamin Werner [email protected] Green Bank Telescope Penn Array Project Leader at Penn Dr. Simon Dicker [email protected] Experimental Cosmology Group Project Leader Professor Mark Devlin [email protected] Computer Information Science Department Senior Project Advisor Professor Insup Lee [email protected] The Experimental Cosmology Group at the University Pennsylvania is finishing the development of a telescope receiver that will operate at 90 GHz, a frequency in the far infrared band. This receiver will be commissioned on the 100 meter Green Bank Telescope in West Virginia as one of a few receivers used for different radiation bands. The scientific interests that lie in observations at this frequency include looking at high red-shift galaxies and answering the question of how common solar systems like ours are in our galaxy. The detector that is being used in this band is a Transition Edge Sensor (TES), specifically a bolometer operating at 450 mK (or -272.7 degrees Celsius). Optics and filters within the receiver as well as the telescope mirrors themselves direct sources of 90GHz radiation to an array of bolometers, sensors that detect heat incident on their surface. This array has eight by nine pixel elements. All bolometers in the array are read using an electronics system (Mark III) developed by the National Institute of Standards and Technology (NIST) and modified by the National Aeronautics and Space Administration (NASA). Part of the electronics system and the bolometers themselves are kept within a cryostat designed and built by the Experimental Cosmology Group at Penn that cools to about 250 mK in a single chamber. The goal of this computer science engineering project was to design and implement algorithms that can be used to tune a three stage amplification and multiplexing system, each stage comprising of one or more Superconducting QUantum Interference Devices (SQUIDs). SQUIDs are circuit elements that respond to a magnetic flux through the device by changing the flow of electrical current out of the device’s two terminals. These SQUIDs, working with inductors that produce magnetic flux, act to transfer signals from one stage to the next and also create the gain. The amplification chain of SQUIDs is what is used to read out the changes of the bolometer detectors as they sense incoming radiation. “Tuning” this system consists of properly setting voltages through digital to analog converters associated with each of the three stages of SQUIDs such that the output is as close to linear as possible with respect to the input. This is desirable because only after a response that is close to linear is achieved can digital feedback be applied so as to make the amplification system completely linear. A secondary requirement of the tuning algorithm is to place each of the TES detectors onto their transition from normal to superconducting. Figure 1: A Sea Squid, Histioteuthis hoylei (left), as opposed to the circuit element used in our amplification chain, a SQUID (right). Current flow changes from Y to Z as the magnetic field through the loop changes. This is due to effects explained by quantum mechanics. [Image credit (left): Hawaii Undersea Research Laboratory] The algorithm that has resulted from this project is necessary because the instrument, as a whole, is to be used by astronomers who are unfamiliar with the system, and may only have a basic knowledge of the technology involved. Thus, a user friendly interface that allows the observation of the tuning process as well as the control of parameters must also accompany the algorithm. Tuning the SQUID amplification system will result in different voltage settings to the stages due to changes in the overall system such as temperature of different components. Therefore, it must be done before any observation is attempted. Creating an algorithm to do this previously human task has sped up the process to limits that are dependent on the data acquisition and system commanding latencies.1. Background and Related Work 1.1 Project History The Mark III data acquisition system was developed by NIST and NASA. Hardware was developed long before the software package was created to acquire data. Firmware in the acquisition cards, however, is still in flux as new functionality has been requested, and bugs have been fixed. NASA’s Instrument Remote Control platform was chosen as a framework for the software, a decision that has been repeatedly questioned due to the fact that IRC is still in Alpha testing, even as commissioning approaches. With input from the Penn Experimental Cosmology Lab, the system is still evolving, mostly in the software acquisition area, but also to an extent, the electronics themselves. Effectively, Penn Cosmology is the first user of the Mark III system. 1.2 A Pre-existing SQUID Tuning Specification Algorithms for tuning the SQUID amplifiers present in the Mark III system have been specified at very high levels before; however, no implementation has ever been made. Due to the fact that the procedure needs to be done one time for each column of SQUIDs, the procedure tends to be tedious and time-consuming. In previous electronics systems, all of the voltage controls were physical analog dials and switches. Part of the reason why these controls are now controlled by a computer is so that tasks such as SQUID tuning could be implemented and hid from the user. An example snippet of the pre-existing procedure specification is shown below. These procedures are meant to be followed by experts who know the system, hence phrases such as “until satisfied” and other high level concepts appear. These conditions are able to be specified in a complete way, and that were a first step to making the algorithm. Figure 2: The current specification of the routine to tune SQUIDs written by National Radio Astronomy Observatory scientist Brian Mason. Scientists familiar with the Mark III electronics follow this routine to tune the first of three stages of amplification. The above routine describes how to find the optimum bias current for the third stage, also known as the series array for all columns of detectors. It states that the user should select a bias