Grounding the Cryostat and Noise Control -------------------------------------------- Initial Rev. 1-NOV-2005 Current Rev. 8-NOV-2005 My understanding of this is simple - to keep noise out of the preamp inputs and to prevent inducing noise directly onto the wire planes, you must not allow currents to flow in the walls of the cryostat. If currents flow in the walls of the cryostat then there are potential differences between different points on the cryostat and you end up with noise either because it couples into the wires themselves or because it appears as a Voltage difference between the preamp input ground and the "shield ground" provided by the cryostat. Even small currents in the cryostat wall can make potential differences because at 1 MHz (or whatever the signal frequency of interest is) the cryostat wall has significant resistance and it is physically big enough that it has significant inductive reactance at these frequencies. In a really big cryostat like FLARE, the vessel may be large enough to have cavity resonance modes that are close enough to the 1 MHz signal frequency to cause a problem. This needs to be looked at but I will skip over this concern in this note. The way that you keep currents out of the cryostat walls is to have only a single ground connection to the vessel. No other metallic or conductive connections are allowed. The one ground connection is at the point where the analog signals (be they raw wire signals or preamp output signals) come out of the cryostat. With only one ground connection to the cryostat, currents can not flow in the walls of the cryostat. The problem in implementing the one ground rule is in all the details. Examples of these problems follow: HV Filter --------- The "only one ground connection" technique immediately fails because often you need the analog output from one end of the cryostat and the HV Filter and HV Input at the other end. Both of these must be grounded to the cryostat. This is a simple easy to picture failure of the one ground rule. The goal is to have no currents in the cryostat wall. The way that you can still achieve this it to "float" either the HV Supply or the DAQ system that receives the analog signals from the cryostat. - Picture the HV Supply being a set of batteries hanging by a thread from the ceiling. This removes the second ground connection (ground loop) and thus there is not a problem of ground currents in the cryostat wall. The entire cryostat remains at one potential. - Or picture the amplifiers and ADCs that process the analog signal output from the cryostat being battery powered and hanging by a thread and sending their ADC data out over fiber optic link to the rest of the DAQ system. This is another way to picture breaking the multiple grounds to the cryostat, i.e. breaking the ground loops that can cause currents to flow in the walls of the cryostat. Currents from AC Loops ---------------------- It is not just metallic connections that cause currents in the cryostat wall. Because we care about currents at 1 MHz then capacitors connecting to the cryostat can cause just as much trouble. - Cryostat feet sitting on a piece of G10 can make a capacitor to metal floor plates. - Pump and cryogenic lines, even lines broken with G10 sections (or whatever) form a capacitor to the pumps or cryogenic storage system and their grounds. - Instrumentation cables and sensors may all be "electrically isolated" from the cryostat but they all have capacitive coupling to it. This includes things like: thermometers, pressure sensors, liquid level sensors, position sensors, and Argon purity monitor. - In reality the power for the High Voltage system and the DAQ components can not come from floating batteries but must come from the AC power line. The best that you can do is to use power isolation transformers to break these ground loops. But the isolation transformer will have considerable primary to secondary capacitance and some ground loop current will result. - All of these capacitors that make AC ground loops and thus cause currents to flow in the cryostat wall can add up to a significant number. At D-Zero this is about 10 nFd. Other required Grounds ---------------------- - Safety regulations will require certain AC power Safety Ground connections to the DAQ system components that are directly connected to the cryostat. The best that we could do at D-Zero was to make these Safety Ground connections through large RF Chokes. - I assume that safety will require the HV cathode supply to be covered for its full run to the cryostat and thus grounded to the cryostat. As described above all that you can do is to put an isolation transformer in the primary of the HV supply to help break this ground loop. - For a large outside cryostat I assume that you need multiple earth ground connections to it for lightening protection. I assume that there are safety regulations or best practice "regulations" that cover this. Multiple DAQ Ports ------------------ - The design of a detector as large as FLARE must use multiple ports in the cryostat for routing out the analog signals. - The best way to handle this is to have the DAQ system components at each port firmly grounded to the cryostat right at their port but otherwise floating. - This can be accomplished by using a separate isolation transformer in the AC line power that supplies the equipment at each port and by using fiber optic links to carry the data from the equipment at the port to the down stream components of the DAQ system. - Reducing the power that is consumed by the equipment at each port will reduce the size of the isolation transformer and thus reduce its primary to secondary capacitance and thus reduce the AC currents that it can force into the cryostat wall. Instrumentation Connections --------------------------- - A large detector like detector like FLARE will have a large number of instrumentation sensors on or inside of its cryostat. These include: temperature, pressure, position, liquid level, Argon purity and voltage monitoring sensors. Any of these can cause problems for the low level wire plane signals. - To picture the problem, the easiest sensor to think about is the innocent temperature sensor of which there will be hundreds inside of a large cryostat. - These temperature sensors are typically 4 wire resistors that make no metallic electrical connection to the cryostat or the detector components inside the cryostat. But in total they make a large capacitive connection and thus form AC ground loops. In addition the leads running to these sensors can transport external electrical noise to the inside of the cryostat and thus to the detector elements. - We have seen noise from both of these effects at D-Zero and had to do work to correct this problem. - A partial solution to this is to plan ahead and put connectors right at the cryostat ports where these instrumentation cables come out so that they can be disconnected when they are not needed. - The temperature sensors that are routed through each connector can be organized into those that are needed for long term operation and those that are only needed to monitor the cool down and to verify stable operation during the first couple of months after the cool down. Once stable operation is verified then most of these connectors can be unplugged and their AC ground loops broken. - Special low noise circuits are available for use with things like resistive temperature sensors. Signal processing right at the cryostat and then an optical data connection from this equipment to the display, alarm, and monitoring equipment would help reduce the noise. Other HEP experiments may already have experience with such low noise equipment. Cryogenic Control System ------------------------ - I believe that the temperature/pressure inside of large liquid Argon vessels is typically controlled by using the pressure inside the vessel to regulate the flow of LN2 through cooling coils that are inside of the vessel. This "safety critical" control loop potentially includes a number of lines that run between the isolated cryostat ground and various other grounds. Asking the cryogenic engineers to work with the electrical noise prevention engineer right from the start may help minimize this issue. - I assume that there are serious safety regulations and best practice "regulations" about how this critical cryogenic control system is implemented, e.g. are breaks in these control lines allowed ? Test setup ---------- - It would be good to have, early in the design of this experiment, an electrical mock up of a "ports worth" of preamps and DAQ system components. The detector itself can be simulated by just capacitors. - The intent is to as early as possible determine what noise sources are causing the most trouble so that they can be corrected. - This test setup could also be used to verify the noise level, both within the signal bandwidth (e.g. 100 kHz to 1 MHz) and at low frequency (e.g. mechanical or power line up to 100 kHz). - Knowledge of the low frequency noise level would allow one to adjust such things as the dynamic range of the preamps which effects how much heat is dissipated inside of the cryostat. Grounding and Noise Control is not hard --------------------------------------- - I don't think there is anything new or outside of the current understanding provided by electrical engineering that needs to be developed to properly ground and control the noise in an experiment like FLARE. - Getting this right should just involve paying attention to a ton of details right from the beginning of the FLARE design. Reference about Grounding of the ICARUS Experiment "Common problems of grounding of the ICARUS detector", Baiboussinov B. 2000-06-06 www.pd.infn.it/~bagdat/group/data33.htm Standard Grounding and Low Noise References "AC Power Handbook of Problems and Solutions", second edition, Deltec Corporation, San Diego CA, 1975 "The Art of Electronics", second edition, Paul Horowitz and Winfield Hill, 1989, Chapter 7