Dear Fritz: I think this document is excellent, not only as a statement of requirements, but also as a general reference for people on CMS. I was wondering if a short synopsis of requirements at the top might help to underline some of the points. Here is a suggestion for such a summary: Summary of compatibility standards for CMS electronics: All signals transmitted differentially. All cables have shields either around the individual pairs or the cable bundle. All power supplies have electrostatic shielding. All requirements for uninterruptible power must be clearly stated. Digital and Analog signals are carried on separate cables. Each electronic system is responsible for removing its heat. Cooling systems must be designed to carry away over 90% of the generated heat. If water is used, cooling systems must plan on water at a temperature of less than 55 degrees. You might want to add more. I also notice that there is a mix of noise and safety issues. Should the two be separated? I should also add that Dick Loveless thinks the field in the electronics barracks will be about 100 gauss...not too comfortable. Should we consider some shielding of the barracks? An iron shell would be expensive. Would loading the concrete in the walls with iron be effective? I am not sure about this. Below I have some edits for the document indicated by **: Regards, Wesley Proposal on compatibility standards on CMS-electronics 1) Introduction Industrial electronics productions follow certain engineering guidelines, the most unknown in physics being Electromagnetic compatibility (EMC). However, EMC is a major ingredient for avionics, spacecraft electronics, medical electronics, automotive electronics and many others. No systematic approach is known for physics experiments although some more recent experiments tried to implement certain standards. Knowing about the difficulty to implement such rules we will limit our standards to details that may be assessed using commonly available equipment. More difficult measurements could be done within a small group of specialists who have access to more elaborated equipment or EMC-laboratories. All other standards might refer to internationally known publications such as VDE, IEC, EN. 2) Noise propagation and definition Any signal that is not intentionally present in a specific circuit is considered noise. Note: Intrinsic noise such as shot noise or thermal noise is not under consideration in this document because these types of noise are inherent in electronic devices. The noise source transmits noise via a coupling path to the noise receiver. These three elements, noise source, coupling path, receiver, are treated separately throughout the document. 3) Noise sources The general assumption is that all electric circuits (and certain other things too) creates a ** ** noise level known as emission. Mechanical dimensions, shielding, current balancing, power, rise-times and many other factors will determine the emission types and levels in dependance of the emission spectrum. Limits are given as national or international standards for certain types of equipment and cables runs, all dependent on the spectrum and separated for emission and susceptibility level. Generally speaking emissions inside the detector should be limited to the smallest possible levels which means all frontend electronics would be subject to the most stringent requirements. Cable runs: Attenuation capabilities of shielded cables reach into the 80dB-range. However, the attenuation may virtually disappear under certain conditions. One of the worst combinations are single-ended cables that carry additional currents due to ground potential differences. In addition all single-ended cables will need some sort of ground return which results in potential shifting causing the so-called common-ground interference. It is therefore generally recommended that all cables runs to and from the detector be true **(not faked!!) I SUGGEST TO BE MORE SPECIFIC HERE ** twisted pair twin lead with proper shielding that does not carry any operational currents. Proper shielding will limit crosstalk and cable emissions in general. ** Cables should either have the individual pairs shielded or the cable bundles shielded. ** Power cables radiate rather strong magnetic fields in addition to radiation caused by ripple currents riding on top of the fundamental frequency. Static fields usually do not have much of an effect. Dynamic currents such as mains AC plus its many harmonics and the ripple currents of mediocre DC-supplies radiate a low-to-medium frequency magnetic spectrum that generates standing voltages on all cables cutting through the cable's magnetic field lines. A limit on both ripple and rise-time of alternating or ripple power currents on power lines will be required. Clock circuits, digital signal transmission Transmissions must be done via ** differential copper ** or via optical links. *** Any high speed link, including computer links, ** must be run via differential shielded cables and ** employ data-framing to balance the currents if at all possible.** Conversions Signal conversion from common mode into differential mode occurs when cable drivers and/or receivers are incapable of absorbing standing voltages that the cable picks up from a noise source. This type of conversion is also widely known for coaxial cables where standing voltages have very different effects and hence cause different currents on shield and inner conductor. The connection impedances differ. So do the produced voltages. This is one of the reasons for preferring balanced lines. Monitoring links, detector control Control and monitoring circuits constitute an additional noise path. Emission levels for the detector control must be low because control and monitoring circuits often use DC-coupled devices. Power supplies In the low to medium frequency spectrum the main emissions will originate in power supplies. Makers of power supplies will limit their efforts on emissions ** to ** what is required by standards. Noise suppression on power supplies is a particularly expensive undertaking. In general linear power supplies (no regulation, magnetic regulation, series regulation via drop transistor) will not radiate noise except some low frequency magnetic near-field. Switching power supplies, thyristor phase-controlled power supplies, DC-motors, AC-contactors etc. will radiate strong noise if nothing is done ** to prevent it. ** It is therefore mandatary to know roughly which device is installed where in order to be able to trace interference ** ** where power supplies are involved. These emissions will, due to the impossibility of putting any power supplies near or inside the detector, be present mainly in areas at some distance from the detector as well as in the counting room. It is recommended to consult an expert prior to purchasing or choosing power supplies. Ground Grounding is a direct DC-connection to a very powerful and widely distributed noise source. Systems must be grounded for safety reasons but the ground connection should not carry any operational current. All operational currents must flow elsewhere, ** which distinguishes the ground clearly from the common. The ground is established by a grid in the floor of the experimental hall and electronics house. Electronics must be properly grounded for electrical safety. ** Beam pipe The beam pipe is a potential noise source. Near field levels ** produced ** by the beam itself are presently being worked out. The field levels obtained will be a requirement for electronics noise susceptibility levels on tracker and pixel detectors. 4) Noise paths Noise paths listing are as long as cabling layout of an experiment but there are obvious noise paths. Knowledge about potential paths helps avoiding false cable routing or a bad choice of the cable itself. Ground connections All connections to ground make a system vulnerable to common ground interference. As no system needs to be grounded in order to work the number of ground connections should be kept as small as possible. Systems connected to long cable runs should not accept ground transfer currents on a routine basis. It is bad practice to buy even more copper mesh in order to keep the ground potential between to points to a minimum. **Such practices can result in large currents, which can cause hazards and inadvertent potential differences between other locations. ** In general ground potential differences are inherent and ** may ** not be altered by copper links. Ground loops The so-called ground loop is not strictly speaking a noise path. It is, however, a way of generating noise from ** stray ground ** currents which happens mainly by converting stray currents into differential mode on the cable. Ground loops may also be closed by capacitances but this leads into the high frequency region where loops do not exist any more but antennas or antenna parts have to be introduced into the modelling. Cable leakage (crosstalk) Strict obedience to the use of shielded twisted pair lines should keep crosstalk in the 60 - 80 dB range. Crosstalk varies with frequency and shield connection so the only recommendation is the proper use and installation of cable shields. Path mains to DC power rails Coupling from the mains to the DC power rails is small provided the power supplies have electrostatic shie