Below are the questions asked during the live event, along with their respective answers.
A: Below 200-300 MHz or so, much of the radiation comes from the cabling and wiring to and from the equipment. The generated voltage in those wires comes from the induced currents into those wires from the noise source. Using a current probe on the wire or cable near the connector of the source of the emissions is typically a good representation of the radiated emissions observed. If you find a way to reduce those currents, you will likely find a way to reduce the radiated emissions.
Q: You say to be careful using near field probes. What is your concern about using near field probes?
A: Some sources, especially magnetic sources, do not transmit well. So you might see high fields with a near field probe, but don’t have a problem in radiated emissions because of it. One extreme example might be the fields between plates of a capacitor might be very high, but away from the capacitor there is no perceptible field.
Q: Is there any defined boundary between broadband signal and narrowband signal?
A: There are no set rules for these types of issues. Broadband tends to be lower frequencies – say below 200 MHz or so, because it is sourced from lower frequency emissions. Broadband noise is typically made up of closely spaced harmonics of narrowband noise. But closely spaced means the fundamental must be low frequency. Thus at high frequency broadband rolls off. But narrowband tends to be found from high frequency clocks and high-speed data. To maintain signal integrity, it is important to keep the first 5-10 harmonics unfiltered. Thus if you have a 16 MHz clock, filtering below 100 MHz or so may not be wise, so other mitigation must be used.
Q: Why does a magnetic field exist at a lower frequency, not at higher frequencies?
A: Well, magnetic fields do exist at a higher frequency. However, they do roll off with distance faster than electric fields because magnetic fields must complete a loop, whereas electric fields will radiate out from a point source. There are other factors as well (source and wave impedances, and so forth). I guess we should be thankful that we only need to worry about electric fields in general.
Q: How do you fix the resonant loop consisting of the vertical antenna and power cord coupling?
A: First I will try to make sure my power cord is not in contact with the ground plane. Or minimize contact as much as possible. But if you have energy on the power cord, then you need to couple it back to the source before it leaves the equipment and goes down the power cord. So the use of line to chassis capacitors (within safety limits of course), and common mode inductors is often the solution. Remember that there must be a path back from energy radiated from power cords. This can be to chassis or to other cables tied to the system. If cables, then those need the same type of filters.
Q: Why is it always difficult or a challenge to arrest Common Mode disturbances? Is there a Thumb rule?
A: Because common mode is such an effective radiation source. Since differential mode results in fields from closely paired lines, common mode must radiate back in some larger area. Often this is from one cable to another, or cables to the chassis. Dr. Clayton Paul in his book Introduction to Electromagnetic Compatibility shows that a 20 mA differential mode current with wires spaced 50 mil apart, will radiate as much as an 8 uA common mode current.
Q: How do you mitigate the human body effects when you grab cables or squeeze chassis?
A: I don’t, or I use them to my advantage. I either rely on the fact I am changing the fields from sensitive wires by grabbing them, or squeezing the cables, or to reduce my personal effect, I will minimize my movement. For example, hover my hand over the cable, then touch it with minimal movement to observe changes. For chasses, I can use my hands or use a wood or plastic stick to press on things. I once had a man bring in hockey sticks to manipulate cables so he wasn’t near them. It worked.
Q: What about vertical power cable and the emission in vertical polarization? As I have observed, it’s the widely-spread cause of 30-50 MHz emission. Even when measurements are conducted in FAC so there can’t be any coupling of the antenna with the GP as you told.
A: Actually even in a Fully Anechoic Chamber, I have seen coupling through the ferrite on the floor. However, you also have a source to the power filters to the chamber. The power to the chamber must be routed through line filters which have a significant amount of capacitance to the chamber. If you can have ferrites on the power lines after they leave the test chamber but before the filters, this may help reduce this effect.
Q: If a mobile phone can overload your receiver when inside the chamber, what is the difference setup-wise when you are performing a field strength or radiated spurious measurements on a mobile phone using a radio base station simulator? Because you can use the same measurement equipment to perform such tests. Thank you.
A: Two things: First, the test lab knows you have a transmitter inside the room, and can add attenuation and take precautions to protect their equipment. Second, often the transmitter is connected to a load that absorbs the energy and keeps it from transmitting. During normal radiated emission measurements, the analyzers and preamplifiers are wide open, and often unprotected – thus rather sensitive. It should be noted that cell phones will increase their power when they detect the closest tower is going out of range. Thus, when you close the door and have a cell phone in the chamber, it will transmit at its full field strength looking to make a connection.
Q: Have you seen emissions from light installations in chambers where the incandescent lamps have been replaced with LED lamps with PWMs?
A: OH YES. I have seen several labs who have to turn off the lights during radiated emissions testing to assure they are not measuring the emissions from the lights.
Q: You talk about being in a chamber during radiated emissions to try things. Is that too dangerous to do during immunity? Excellent presentation.
A: Thank you. During radiated immunity testing, I do not recommend being in the chamber. But if you look at the safety standards for radio frequency exposure, it appears you can be in the chamber during a 10 V/m test, at least for a limited time. But I still don’t feel it is wise – but that is my personal view.
Q: How do you cure the radiation from the board to wire connector junctions – i.e for LVDS cable or Camera cable assemblies? Observed by EMC probe that there is no radiation on cable assemblies.
A: Wow… I am not sure I know how to answer this one. I think I would need to see the layout and how the connectors, cables, and filters are put in and used. I am not sure this helps, but I recommend running internal cables (especially ribbon cables) along the chassis if possible, to keep them from cross-coupled energy. But be careful that if you have a noisy cable, it doesn’t generate new currents in the chassis you have to deal with.
Q: If you have long input/output cables, what is the best way to lay these cables out (in a coil, back and forth, etc.)?
A: Unfortunately, ANSI C63.4 is pretty clear how they have to be routed (off the back of the table, 40 cm from the ground plane, and bundled non-inductively). But when you are in troubleshooting mode, do anything you want to find out what is the radiation source and what is not. Replace cables with shorter or longer ones. Add ferrites. Coil them or lay them out flat. Drop them on the ground plane or pick them off of it. Or anything else that comes to mind.
Q: Troubleshooting book recommendations?
A: Can I recommend my own book? EMC Troubleshooting Cookbook for Product Designers.
Q: Is there a particular material you’d suggest to go between screens and chassis or is it better to design a specific geometry for dropping signal coupling or both? Specifically LCD displays with a touch screen.
A: I think both would be my answer. I try to avoid adding gaskets if I can since they can be another impedance between bonding surfaces. However, they can also help. Assure overlaps are adequate. Assure the chassis does not have paint or other overspray insulation but is cleaned of impurities. Touch screens can be a problem and may need capacitive coupling to chassis to quiet them. When using capacitors around the opening, try to keep them symmetric as possible (one in each corner for example).
The following three questions have been grouped together, as the answer given addresses all three issues.
Q: Are there any solutions for broadband noise to suppress totally? Found Balloon squeezing pattern here. Energy moved to different freq.
Q: How do you address noise from switching regulators used for localized power sources in a large machine?
Q: How do you attack noise from switching motor drives (H-Bridges)?
A: I try to go back to the source of the energy, to begin with. Say the source is the rectifier. Then around the rectifier, I try to add some line to line capacitance to quiet the high frequency being generated there. If there is capacitive coupling from the rectifier to the chassis (heat sink or something like that), then that needs to have a capacitive couple back to the lines as well, both input and output lines. Then add common mode inductance outside of that to make the path down the line more difficult for currents to flow – they would rather flow in the caps instead. For other noise sources, similar issues have to be used. You need to either soften switching transitions (which may reduce efficiency and not a good solution in that regard), or you have to dampen overshoot and ringing (less effective since the quick rise or fall time is a significant source of high-frequency energy). As soon as possible outside of these circuits, you will need to address the energy before it leaves the area. Remember that many of these problems are due to parasitic issues (inductance and capacitance). The coupled energy needs to be returned to the source.