A few ideas about detector grounding and shielding D. Breton ETD parallel session – Frascati Meeting September 2010 ETD/Online - SuperB Frascati Workshop – September 2010
Introduction The problem was evoked in the architecture document which was the base of the ETD chapter of the white paper: “Ground has to be as equipotential as possible. This begins as soon as when building the experiment hall, were it would be wise to foresee the possibility to connect the iron bars sunken in the concrete ground to the experiment electrical ground, in order to provide the most perfect possible mesh.” About power supplies: “Power supplies should if possible remain outside of the radiation area, because radiation tolerant power supplies are expensive and less reliable. Therefore, power consumption on the detector has to remain reasonable, in order to limit the power loss along the cables.”
Grounding The location where noise is “dangerous” is the front-end analog electronics => that’s where we need to have the best possible ground This is actually in/on the detector One has to avoid the potential high currents flowing back to the power supplies through unforeseen paths, especially the other subdetectors ground connections have to be strong everywhere but also have to carry small currents Subdetectors have to be fully equipotential. This is less stringent between different subdetectors because there is no analog link between them The choice of having a “perfect” equipotential ground has to be made very early in the detector design
Powering the front-end electronics Ground is not supposed to be the return path for supply currents Power cables have to offer a very low impedance path for current returns The best solution is to leave power supplies float and to fix the ground potential on the detector side of the power cables Even better to regulate the power near to its use => get rid of voltage drop fluctuations => permits using cheaper DC/DC main supplies Radiation tolerant supplies are expensive and less reliable Let’s try to avoid them as much as possible
Powering: example 1 Analog front-end electronics located on the detector Remote power supply + regulators on the detector
Powering: example 2 Mixed front-end electronics located on the detector Remote power supply + regulators on the detector
Design rules & cables DC coupling requires perfect grounding Neither DC or AC current AC coupling is only sensitive to AC currents Separate or common analog and digital grounds on boards ? My own experience pleads for strong common ground (this allowed us for instance to use 2mV thresholds or BABAR’s DIRC PMT signals) Cables have to be carefully defined They have to be in perfect adequation with the current requested and the noise level tolerated Custom defined. Shielded twisted pair. 3 pairs (4mm²) Custom defined. Shielded twisted pair. 6 pairs (0.5mm²) Both are LS – 0H – FR. Being studied by CERN safety group.
Shielding Shielding mainly concerns the crate to crate and rack to rack interconnections. Whenever possible, use Ethernet-like shielded Cat6 cables with shielded RJ45 plugs at both ends. This also works for analog data up to ~20m. High rate serialized digital data can also be transmitted on this kind of cables up to a distance of 15 meters. Shields must be connected to ground on both sides. In a general way, try to avoid any consumption local to the receiver side Favour differential signaling because it offers a constant power consumption If unipolar, use serial adaptation at the source in order to have the signal return path same as forward path.
Conclusion Grounding and shielding is a key element in the success of the design. it is usually considered as a boring item. but spending some time thinking of it from the beginning can save you so much time and trouble eventually ... Grounding mainly consists in having strong equipotential grounds for subdetectors while ensuring good and independent current return paths to the power supplies Supplies can be floating DC/DC blocks, feeding regulators located on the detector next to the front-end electronics, and cleanly referenced to ground there. Shielding concerns all cables, especially those transporting perturbative signals or those sensitive to it. In order to be effective, cable shield has to be connected at both ends.
Exactly like for radiation, it is very important to take grounding into account at the very beginning of the design, because building a perfectly equipotential system eventually becomes very difficult !