Video: Advanced Topics in High-Current Filtering | Presented by NexTek | Duration: 1776s | Summary: Advanced Topics in High-Current Filtering | Presented by NexTek | Chapters: Welcome and Introduction (94.99s), Filter Fundamentals (157.115s), Heat Sources Analysis (290.9s), Voltage Drop Measurement (430.13s), Temperature Rise Control (573.25s), Overcurrent Thermal Response (698.58s), Overcurrent Modeling (834.03s), Thermal Management Rules (976.78s), Failure Mode Management (1128.325s), Fail Safe Filters (1235.585s), Power Line Filtering (1364.965s), Isolated Chamber Measurement (1586.255s), Closing Remarks (1745.52s)
Transcript for "Advanced Topics in High-Current Filtering | Presented by NexTek": Hi, and welcome back to EMC Live. So far, we've heard from two industry leaders and experts, and now we're gonna finish today's sessions with yet another. Thanks again for being with us today. In this presentation, we'll hear about some of the unique challenges that are prevalent when filtering high current circuits. We'll hear recommendations to improve performance through filtering and some additional insights in the world of high current filtering. Our speaker is George Kaufman, the chief technical officer at Nextech. Over his decades in the field, George has secured several patents in the filtering and suppression field and is a senior technical contributor to Nextech product design, manufacturing, and application support. Nextech designs and manufacturers coaxial surge protectors and high current EMI filter products. Please ask your questions in the q and a tab to the right of the chat. We will send them to George and his team after today's session. Now I'd like to welcome George Kaufman to EMC Live. Take it away, George. Hello. This is George Kaufman from Nextech in Billerica, Massachusetts. We're gonna be talking about some application topics with high current feed through filtering. The first thing we're gonna define is what is a high current low pass filter. We're gonna discuss thermal performance and managing overcurrent pulses. We're gonna discuss failure modes and reliability improvements. We're gonna discuss, power filter, circuits, and we're gonna discuss, measurement of insertion loss. For this discussion, a high current low pass or feed through filter is rated at over 50 amps or more, is a feed through configuration. So its filtering is in intact to, over one gigahertz. These particular devices are compact and very rugged, and their job is to reduce RF voltages, whether it's, susceptibility or an emissions, control, application. This particular, is a 400 amp filter. They usually have a feed through electrode with an electrode in one side and as well as the other side, and they have a mounting to a bulkhead usually. This particular one is about 90 millimeters or about four inches long, and they're usually available in a variety of capacitance, values and in a variety of voltage values. Their insertion loss starts at, the cutoff frequency, which is defined by the capacitance, and it continues up at about 20 dB per decade and then reaches a plateau. Some important parameters that we have are the total voltage drop from wire to wire, which we're going to be using more in a few slides from now, a temperature rise, which is typical, and then, a heat rise constant, which is the delta t would be equal to that to that c one value times the total watts dissipated to the point 85 power. We're gonna be using this formula in a few minutes. This would be typical specification. When we're talking about temperature rise, it's important to discuss the heat sources and the heat balancer. This has got the electrical circuit here for a typical feed through capacitor. It's got the thermal where the power is, originating from and where it's going or being dissipated and the physical. We start with the physical. We have an input terminal, and this is a wire connected to a lug, goes to the electrode. Through the electrode, there is a grounding and mounting panel here, and there's an output, in this case, a bus bar. And the total voltage drop, we're gonna be using this later, is the voltage drop from the wire to the wire, or in this case, it would be from the wire to the bus bar adjacent to the connection. The circuit, the rated through car, which is going through various resistances associated with the wire lugs, the the lug to electrode connection, the through conductor, and then out and connecting to the bus bar and so forth. There's also a shunt or a bypass current that has there there's a rated RF shunt current that is going through the capacitors to the ground or to the mounting shield effectively. And the ESR here is causes power dissipation, due to that shut current, and, of course, the through current is an I squared r generator here. If we take a look at the thermal dissipation, heat sources, we have a kind of a constant dissipation associated with the through conductor. There's a highly variable heat dissipation due to the shunt current, and there is a considerable considerable variation on the connection and the lug itself. So, for example, the wire lug, you'll see it later on, has a lot of, variation in terms of the lug design and the, and the and the wire, used. So these application variables, the attachment on each end and also the shunt current are the highest variable heat generator. And, of course, we take that delta t and we apply it up to the local ambient, which might include, nearby hot objects. We've got an HPR feed through filter here, and it's in a low voltage high current loop. We're putting we're putting the current with a copper lug into the into one end of the electrode, and we're coming out of a copper bus bar on the office's side. And we're transitioning from the copper bus bar through an aluminum lug back to the return loop. We've got a equivalent bulkhead cooling surface. We're just using an aluminum washer here. It could be larger than that in some applications. But we're gonna be measuring the voltage drop at various points of connection to find out what's going on with the voltage drop and where the where the power dissipation is coming from. So if we measure from the one side, it's not the wire, but it's the conductor adjacent to the electrode to the other copper wire, we're getting a drop of around 15 mill millivolts. So 15 millivolts times the 300 amps would be the approximate power dissipation within this particular, unit as a whole. And if we divide this up, we wanna know what each part is. We go from the same point on the busbar to the electrode. We're getting about three and a half millivolts of drop across the end of the busbar to the electrode interface. And on the other side, we're getting from from the wire to the electrode. We're getting around nine millivolts of drop. So we're seeing substantially more drop in this particular conductor. We can compare this, con connector to this aluminum connector. So we're gonna go to the to the wire on the one side to the equivalent of the electrode if it was out in this location would be in in this location, approximately on on the bus bar. And we're getting a voltage drop of 2.6 millivolts approximately. And this would indicate that the power dissipation with this large aluminum lug is actually much lower than with this copper lug. So you can use lugs rated for the ampacity, and they have much different performance. This just indicates where the power how you can measure the power, dissipation, and troubleshoot which components might be causing a higher than normal dissipation. In controlling temperature rise, there are a couple of important factors. The primary driver is usually the true current I squared r, and you've seen the variety of voltage drops and therefore power that's associated with the connections. And that can cause a blocks dissipation and associated temperature rise. The second usually, the second most, highest cause of temperature rise is the harmonic or shunt current. This is a little bit harder to calculate and usually is not a critical factor, but can be in some cases. So for example, that would be calculated or could be cap calculated from the applied voltage, and then we use I equals c d v d p to get the pulse currents that are associated with that voltage waveform. And we take the pulse current train, and we could use FFT, for the for the harmonic currents and the ESR for every one of those, frequencies and calculate in sum to get the watts dissipated by the shunt. What kind of power are we usually talking? We're talking about for a 55 amp model, usually dissipation of eight watts or less to reach rated temperature. And for a 600 amp, it would perhaps increase to around 25 watts. So that's the kind of power we're dissipating totally between both of the sources here. The delta t is usually back to that formula is a constant times the watch dissipated to the point 85 power. So this is why measuring the total voltage drop and current are important because you can put it right into that equation and come up with an estimated delta t. We're going to try to estimate the thermal response for overcurrent conditions. And to stand in for the filter as well as the interconnecting wiring, we're just gonna use an eight AWG wire and we're going to see what happens with overcurrents. This eight w eight AWG wire or 10 millimeter squared is a has a 105 degrees C temperature rating rated insulation system. Its hookup single wire hookup rating is 90 amps at 30 degrees c, and all this testing was done at low altitude. If we take a look at the delta t versus time, if you just put a rated current of 90 amps in the wire, this asymptotically approaches delta t of about 54 degrees c. If you put higher currents like a 120, a 141%, or 200% of the current rating into the wire, this same delta t is a is attained in shorter and shorter times. For example, at a 141% of the current rating, which is about twice the power dissipation in the wire, this temperature rise of del delta t of 54 degrees c is reached in about three and a half minutes. The time constant for this a a AWG wire is about four and a half to five minutes. So what's happening here is the time constant is roughly the same. It depends on the geometry steady state temperature of each one of these overcurrents is higher and higher and higher. You can estimate the new temperature rise of the, higher currents as the ratio of the current of application to the rated current, the 1.8 power times the delta t at the full rated current. So each one of these currents are are going to plateau at a higher temperature, but we, of course, we've terminated the test when the insulation system is compromised. So you can have a two parameter model which predicts the temperature rise of overcurrents. So let's put that model to use for overcurrent situations. We're going to go from zero to a 100% of the delta t for a particular insulation system or and or product. And we're gonna have a time constant. It's time down here, and the one time constant is approximately right here. If we put in the rated current, the model is fairly trivial. We're just going to asymptotically approach the steady state rated temperature rise. If we put a very high or overcurrent, and we're gonna we're gonna increase at a much faster rate. And when we get to the rated temperature rise of the system, we're gonna stop the overcurrent and then transition to the full rated current. When we do that, the temperature of the insulation system will just be constant at that level, but this is not a repetitive event. In order to start this curve and do it again, we have to start down here at the lower level so that we can, again, approach this. So it's this can only be at one time on startup. What happens with repetitive pulsing conditions as shown by the yellow and the blue lines, we can not as much current as this particular case, some intermediate value, but we stop below the rated temperature, and then we have to have a cool down lower current period. When we do that and we cool down closer to this level, we can start the process again, and it will get a little bit closer to the rating. If we continue this, there'll be an upward trend of this, and if we if we've done it right, then this peak temperature will eventually be the same as the rated maximum temperature. However, there's a couple of conditions. The when the I is larger than the rated current, we we have this temperature rise, and then we're gonna have to have the current considerably lower sometimes than the rated current. So the average current will always generally tend to be less than the rated current if it's a cyclical phenomena. So some rules of thumb for thermal management of high current conductors. Free convection cooled conductive systems are nearly isothermal. They're all the same temperature within a just a few degrees c in a balanced design unless there's a problem component. So that means that you can usually measure the temperature on the conductor or on an insulated portion of the conductor, and it'll be fairly close in temperature as long as the power drops are balanced across the entire path. Recurring overcurrent situations needs a margin to restart at a low low initial to restart at a lower initial temperature. The average current, in this case, is less than the steady state current rating of the wire. You can refer to a similarly rated delay fuse or circuit breaker curve. Usually, these are lower have a lower transient response than the wire itself. You saw me eight AWG, case. The time constant was around four and a half minutes for the thermal response. Larger conductors, for example, a 300 amp wire, number two, gauge would have a time constant of, like, fifteen minutes approximately. So as you go up in current from that eight AWG, the response time gets slower, and usually the fast the the circuit breaker or fuse curves have a shorter response. So you can use this as a guideline to what might be safe to operate. It's best to measure the temperature rise of the insulation system and connectors and the filters and try to maintain a 10 or 20 degrees c margin. And this margin is needed for altitude, for uncontrolled, or unknown hot spots, and other situations of that nature. And if you measure the voltage drop along the conductor, you can try to minimize the higher voltage drops that are spaced closest together. The high current, conductors are the thermal dissipators that are very significant. And if you try to put too many of these close together, the cumulative effect of the power dissipation can be significant. And, controlling the voltage drops is the way to manage that. We don't like a transition to failure mode management. The common failure mode, for filters are overvoltage applications, severe over over voltage applications in some cases, random component failure, over temperature conditions, and very high inrush or charging our RF bypass current. And the capacitors used in filters can store significant amount, amounts of energy, and shorting releases this energy usually to cause thermal damage internally in the unit. Damaging follow on currents from the source or the generator can cause significant thermal damage, particularly if the DC voltage is greater than about 12 volts, which can sustain a long term arc. These catastrophic failures are very rare. We have very few cases of this, but the risk of that failure mode does exist. And in some critical applications, this risk of failure is not acceptable. We're gonna discuss ways that Nextech has to manage these failures. The first is to, use, environmental stress screening to remove infancy failures and provide higher higher reliability, to reduce the likelihood of these kinds of failures. The the burn ins that we use, and that's an e modifier in the part number or or suffix in the part number, is we do a burn in, and we use thermal shock, temperature cycling. And this is used in some applications, and these are done to, mill c forty nine four sixty seven type criteria. The second way that we manage these failures are are fail safe filters, and what we've done here is put series capacitors in for the bypass capacitors. And in this particular case, we have one capacitor in series with a parallel, group of two. So and we have an an array of these in the filter, so these themselves are redundant. If we take a look at this particular configuration, it's shown right here in the physical, product, and it would connect to the center conductor to the to the ground. These, are flexibly mounted to the center conductor, so it's very shock and vibration oriented, and you get an inherent double voltage rating. The shorting of any one of these connect, capacitors would not cause a failure to ground and not not cause cause a short. And in fact, the first time you get a failure, for example, if this one were to fail in the entire array, these two are not the most likely candidates. You'd get a second failure random failure of another, element. So we call this an n plus n redundancy. This can sustain several failures and still function at more than adequately. So what are the advantages of the fail safe filter? It's very difficult for the failures internally to propagate. Some capacitors have a series capacitance in one ceramic body or one polymer body, and you can imagine that those can propagate across all the capacitors in that structure. These are completely independent capacitors. There's a very significant reliability lifetime increase with this technology. Failures of short circuits are managed quite well, and this dramatically reduces the risk of high fault current or flow current damage. It does not help, however, with severe abuse such as overvoltages. We're gonna touch on some of the challenges of power line filtering care the characteristics. And if we first take a look at the system impedance, the system impedance of an AC line starts off very low and then increases at around a 100 kilohertz, one megahertz, somewhere in this range, and goes up to, exceed 10 ohms up until the 100 ohm range. This becomes very significant because 50 ohms is is a predictable filter impedance area, whereas the low impedances are quite different from traditional filter design. If we take a look at DC systems, we see the same characteristic. We see very low ohms for the power frequency or or the DC level. And as frequency goes up to one to 10 megahertz or, higher, the the impedance of the system is increasing. This happens to be a, test condition for auto automotive applications. So we see that, for low impedance systems, series l's are quite effective. And for high impedance systems above 30 ohms approximately, shunt capacitors can be very can be very effective and and actually more compact. So what we do is we usually use the feed through to do high frequency filtering, particularly above one to 10 megahertz. And we use for the switch mode power supply, pulse width modulation, tight noise sources, we use discrete l's and or c's, to control this device and usually closer to the source. The feed through is specialized in the highest frequency filtering for EMC compliance. But how do we handle this dramatic change in system impedance? 50 ohms is typically used for filter performance measurement. Most power systems transition to about 50 ohms approximately at around one megahertz. So a network analysis can be used successfully at modeling and trying to predict the outcome. Below about 200 kilohertz, the impedance is usually considerably less than 50 ohms, and we found that SPICE modeling in conjunction with your power circuit is the best way to actually model the effectiveness of a filtering solution. We mentioned that, filtering needs above 200 kilohertz or so can be challenging to filter, and the impedance of the line maybe is less than 50 ohms. In this case, we have pi filters, which are configured to boost the insertion loss in the particularly in the 500 kilohertz to, let's say, 100 megahertz region. This is substantially more. You can see the the capacity of action is starting and would continue on this slope. We've added 10 or 20 more dB in this lower frequency region, and this will help in compliance in the in the region below around thirty thirty megahertz down into the the few 100 megahertz range. These are compact as well and are available in high current configurations. Filtering measurements are usually done in the coaxial world. However, measurement limitations with high current filters usually exist because the wire is a single large conductor into each side of the filter, and this produces a measurement challenge. If you do the measurements in open air, you get a significant coupling capacitance. And this shows a coaxial wire on the ground plane, a grounded center conductors, and a dielectric shown, for, about, 50 millimeters or about two inches on both sides. And in this case, there is a ground floor that will exist. In this particular case, the wires are fairly long and fairly large diameter. We're getting a inability to get below about 30 dB in the upper in the upper region just due to coupling if you were measuring insertion loss. And this is with both conductors grounded, so they are perfectly filtered. So we need a a better solution to measure the performance rather than open air. So what we use is a two isolated chambers. We have a slider in the middle for the DUT to be mounted on, a first chamber with a coaxial connector, in contact with the electrode of the DUT, and we have a separate chamber, on the other side. So this chamber is isolated from the outside world and the other side, and this this one is also isolated, likewise. You see the you see the picture here of the slider, the DUT mounted on and going entering into the tube. So we've got the DUT mounted on the slider that goes in, in the middle, and we put the second cup on and we make contact, we get an insertion loss that it has a very low ground, floor over here. So we can see that this device is a very wide band. This scale is from 20 kilohertz to one gigahertz, and we're we're filtering from about a 100 megahertz very effectively out to a gigahertz. And that's made possible by the isolating chamber. Thank you, George, for sharing those insights and for helping us in on a high note today. And as I said earlier, we'll send any questions you have for George to him and his team, and they'll follow-up after we conclude today. And with that, we've come to the end of EMC Live fall twenty twenty six. On behalf of all of our incredible speakers and the entire interference technology team, a huge thank you to all of you for joining us today. Until next time. Take care.