T962 TPC Bias Voltage Filter Description ---------------------------------------------- Original Rev. 13-JAN-2008 Current Rev. 25-JAN-2008 The TPC Bias Voltage Filters, which are external to the cryostat, serve at lease three important purposes: - These filters break the ground loops caused by the cables that carry the Bias Voltage from the power supplies to the cryostat. - They block noise on the Bias Voltage supplies from entering the cryostat. - They provide over voltage protection so that an improper adjustment of the Bias Voltage supply will not result in break down of the TPC wire signal DC blocking capacitors which would both destroy the preamplifiers and may require opening the cryostat to replace damaged DC blocking capacitors. The noise that is being blocked from entering the cryostat can come from either the Bias Voltage power supplies themselves, from other equipment that these supplies are connected to, or this noise could just be picked up from other sources by the relatively long SHV cables that connect the supplies to the cryostat (antenna effect). Noise on the Bias Voltage leads must be blocked from entering the cryostat. Once any noise is allowed to enter the cryostat it is almost impossible to prevent it from interfering with the very small TPC wire signals. Mounting and Grounding the Bias Voltage Filters ----------------------------------------------- To be effective the Bias Voltage Filters must be mounted very close to the point where the Bias Voltage enters the cryostat. The top flange plate of the cryostat is the best object to define as the common ground point. Thus the Bias Voltage filters need to be mounted to bracket(s) that are welded to the top flange plate. The purpose of these bracket(s) is both to provide mechanical support for the filters and to provide the main ground connection to the filters. You absolutely can not use the shield on the SHV output cable from the filter to ground the filter. Mechanical Construction of the Bias Voltage Filters --------------------------------------------------- The Bias Voltage Filters need to be housed in metal box(es). It is fine to use either 3 separate boxes or one common box for all 3 TPC Bias Voltage Filters. I would not mount the filters for other supply Voltages in the same box with the TPC Bias Voltage Filters, e.g. the Purity Monitor should have its own separate filters. You could use either die-cast aluminum boxes or NEMA steel boxes to hold the filters. These are available from manufacturers like Bud and Hammond through distributors like Newark. In either case, where ground connections are made to these boxes, it is important to first prepare a clean metal surface, to use external tooth lock washers on both sides of the metal surface, and to use big enough steel hardware to make a solid stable connection to the box. Number 10 steel hardware is convenient to use and it mechanically strong enough to make a good connection. The SHV connector at the input to the filter must be isolated from the metal box that holds the filter. This is most easily done by mounting the SHV connector to a sheet of G10 a few inches on a side and thick enough (1/8" is fine) to give good mechanical support. This sheet of G10 is mounted to the wall of the box with screws. The hole in the metal box that the SHV connector passes through should provide about 1/10 inch of clearance around the shell of the SHV connector. You can look at the Bias Voltage input to the electronics box on the PAB cryostat to see an example of this construction. Do not mount the SHV connector directly in the wall of the metal box using mica or nylon washers or something like that. That type of construction would prevent making a solid connection to the shell of the SHV connector and it has other problems. Note that the SHV output connector will most likely also need to be mounted in this way which isolates it from the metal filter box. Filter Component Selection -------------------------- This section presents the reasons for selecting the various component values used in the Bias Voltage Filters. Before that there are some general comments about the type of components used in the filters. For the reference designators please see the schematic drawing of the Bias Voltage Filter in this web directory. - Resistors All of the resistors are wire wound and may appear to be over size for used in this filter application. The reasons to use this type of resistor are: . minimize the chance of an open resistor We do not want any of the resistors to open under a fault conditions. An open resistor in one of the resistors that break the ground loops could cause a safety problem. An open resistor in the "high voltage" line could leave the capacitors charged at 500 V when the power supply is turned off - also a safety problem. . minimize the chance that any of the resistors that are in the "high voltage" lines will flash over under any conditions . minimize the excess Johnson noise generated by the resistors in the filter - Capacitors The capacitors used in the Bias Voltage filters should have a working voltage rating that provides a generous safety margin. These capacitors should use a good dielectric that does not make noise when it is polarized or have a failure mechanism. The capacitors must have a small effective series resistance up into the MHz range. R1 The purpose of this resistor is to give the varistor something to work against when the power supply voltage is set too high and under normal conditions it provides part of the series resistance for the one pole RC filter. If the power supply voltage is set too high then the varistor (VR1) will start to conduct and cause current to flow through R1 and thus a voltage drop will appear across R1. This resistor must have enough resistance so that a nominal amount of conduction by VR1 will cause a voltage drop across it but have a small enough resistance so that the normal TPC detector current will cause almost no voltage drop across it. 20k Ohms is a good compromise value. Digi-Key No: RSB-20KCT-ND This is a 20k Ohm 3 Watt wire wound resistor about $1.56 each. R2, R4, & R5 These resistors are used to break the ground loops. They must have enough resistance so that any induced ground loop potential will appear across the resistor (and thus not induce a significant ground loop current) and they must have a small enough resistance so that the shields of the SHV cables always remain near ground potential even under fault conditions (e.g. a short from the center conductor to the shield in the SHV cable from the power supply). 200 Ohms is a good compromise value. Digi-Key No: RSB-200CT-ND This is a 200 Ohm 3 Watt wire wound resistor about 90 cents each. R3 This is the main part of the R in the RC filter. It must be big enough to give the RC filter a long time constant (using a practical value of C) and it must be small enough so that there is very little voltage drop across it when a nominal value of current is drawn by the TPC detector wires. 50k Ohms is a good compromise. Digi-Key No: RSC-50KRCT-ND This is a 50k Ohm 5 Watt wire wound resistor about $1.69 each. R6 This is a bleeder resistor to discharge the energy stored in the filter capacitors C1 & C2 when the Bias Voltage Filter is not in use, e.g. after the wire plane bias voltage supplies have been turned off. The value of this resistor must be big enough so that it draws very little current during normal operation yet small enough so that it can discharge the filter capacitors in finite time. A value of 100 Meg Ohms is OK. Note that it will take this resistor some minutes to discharge the filter capacitors. An Ohmite type Slim-Mox 104 resistor will work well in this application. C1 & C2 These capacitors are the C in the RC filter. They must be big enough to get an RC time constant many times longer than one over 60 Hz. They must have a working voltage at least two times the normal operating voltage. Polypropylene film dielectric is used to prevent excess noise when the capacitor is polarized. Two of these 0.47 uFd are shown in the schematic but using 4 of them would be of some advantage if there is room for them in the box that is selected to hold the filters. The ground side of these capacitors must have a solid low resistance low inductance connection to the metal box that holds the filter and thus to the mounting bracket that provides the main ground connection to the filter. Illinois Capacitor PPB series 0.47 uFd 1000 Volt Newark stock number: 30K6751 about $4.36 each. VR1 This varistor provides the over voltage protection. The voltage rating of this varistor should be just comfortably above the highest voltage that we may want to apply to the TPC wires. Note that the turn on of the varistor is not an absolutely square corner so that you must have some margin between the rated voltage of the varistor and the maximum TPC Wire Bias Voltage that you want to use. A 600 V DC varistor would be a good pick. Digi-Key No: P7258-ND The price of this part is about 50 cents each. Notes: ------ - With two 0.47 uFd capacitors the time constant of this one pole RC filter is about 70k Ohm x 0.94 uFd = 66 msec or about 4 times period of the power line frequency. - There is a separate R1 and R3 because you do not want the varistor right up against the C in the RC filter because you want the filter to be effective against any noise made by the varistor when it is operating close to its breakdown voltage. R1 is made big enough for the varistor to work against when the power supply is set too high. The bulk of the R in the RC filter is put down stream of the varistor, i.e. R3, so that the filter can be effective against any noise from the varistor. - Unless the SHV cable run from the output of the filter to the Bias Voltage Feedthrough can be made very short and with minimum loop area then resistors R4 and R5 should be included and the filter's SHV output connector should be insulated from the metal filter box in the same way that the SHV input connector must be mounted. Answers to Questions: --------------------- > could I just use a single capacitor ... Although capacitors are available in larger sizes (in the 1000V rating the Illinois Capacitor PPB series is available up to the 0.68 uFd size) there are advantages in using multiple smaller capacitors in parallel. The ultimate attenuation that the filter achieves in its "stop band" will be limited by the physical layout of the filter and by the parasitic characteristics of the component used in the filter. A realistic model of a capacitor includes a parasitic series resistance. By using multiple smaller capacitors in parallel you reduce the ill effect that this series resistance has on the final attenuation that the filter will achieve. You can easily see this technique used even in consumer grade equipment. Next time you are looking at a pc mother board look at the banks of capacitors (both aluminum electrolytic and ceramic) that are used in parallel to make the power supply output for the cpu stiff enough. Another important reason to use multiple capacitors in parallel is so that you can have multiple ground runs and connections to the metal box that holds the filter. Any voltage drop across these ground runs (due either to resistance or inductance in these connections) will appear as noise at the output of the filter. Using multiple runs in parallel will reduce this problem and increase the ultimate attenuation that the filter will achieve. It is also because any noise that appears on the filter's ground connection will also appear on the filter's output that you need to have a solid ground connection to these filters. Their ground must exactly matches the ground that the preamps see. That is why the filters must be mounted to brackets that are welded to the flange plate of the cryostat. The filter can not do any better than the ground that it has. These polypropylene dielectric capacitors were selected for the filters because: they have a low parasitic series self inductance and resistance (all of the "plates" in one of these capacitors are in parallel) and because the polypropylene film does not make excess noise with it is polarized. Other types of capacitors are not as good in these respects. > (more capacitors) in parallel to C1 and C2 ... There are two ways to think about this: You could just double the number of capacitors and put them all in parallel (all with their own private ground connection to the metal box that holds the filter). That would increase the RC time constant and thus increase the attenuation of the filter in its "stop band", i.e. in this case the stop band is all frequencies above 1 over 2 pi R C. The other solution would be to add a 2nd stage to the filter, i.e. put in another 20k or 50k Ohm series resistor and put the second pair of capacitors after it. - You can calculate which of these options will give you more attenuation. Pick some frequency where there is likely to be a lot of noise (e.g. 720 Hz from the TransRex power supply making thousands of Amps for the Muon magnet or 50 kHz from the various switching power supplies around the T962 detector) and calculate the attenuation at that frequency for each option. - The physical layout of the two stage option will have less parasitic coupling from its input to its output (because they are further apart in the two stage option and because the parasitic capacitance of the two series resistors will be in series and thus couple less noise directly from the input to the output). - If we needed to carefully filter out noise up into the region of 10's or 100's of MHz then we absolutely would need a second stage built with components that work will in that frequency range. E.G. if this filter was for use with the Atlas LAr Calorimeter or something like that which has high speed detector signals we would need a second stage build with physically smaller capacitors that can work well in that frequency range. - The reason that we can consider another stage (with its additional series resistance) is because there is so little current flowing through this filter and thus even with 70k Ohms of series resistance there is little Voltage drop across it. Think how different the design of this filter would need to be if it had to supply 1 Amp of current. You could not stand a 70 KVolt drop across the filter. See the discussion about R3 above. To really design this we need to know: How much current will be flowing through this filter ? How much ionization current is there in the TPC detector ? How much parasitic leakage current flows in the rest of the Bias Voltage wiring after the filter ? How much Voltage drop across the filter is OK ? How stiff does the output from this filter need to be ? > ... putting together the HV filters for ArgoNeuT ... It is a small point but it is better if you call these the Bias Voltage filters. The HV in the LArTPC detector system is the potential applied to the TPC cathode. There is typically a different filter for the cathode HV.