EMC FLEX BLOG A site dedicated to Automotive EMC Testing for Electronic Modules

Troubleshooting RF Noise and Fixing Ground Loops

Fixing Ground Loop Noise

Antenna Polarization (Vertical & Horizontal)

A requirement for CISPR 25 Radiated Emissions and ISO 11452-2 ALSE RF Immunity.

  • The 1.7m test harness runing parallel with the edge ground plane will generate horizontal polarized emissions.
  • Portions of 1.7m test harness reaching connectors positioned above the 5cm thick Styrofoam on DUT and Load Simulator would generate vertical polarized emissions requiring vertical antenna polarization to be captured.
  • LS support equipment cables running over the edge of the metalic table may generate a combination of horizontal and vertical emissions.
  • Folded LS support cables tend to cancel the field generating very low vertical emissions if the folding is very tight.

It is critical to eliminate the common mode currents on both 1.7m test harness and LS support cables for lowering the noise floor to minimum 6 dBuV/m under CISPR 25 limits.

In automotive EMC the DUT is normally remote grounded in one point via supply return line to the negtive pole of the 12V battery. Local grounding for DUT with metallic housing is not practical given the risk of grounding loops and rusty connections as the car is aging. Unwanted common mode currents may run along the outside of the cabe's shild: 

  • The cable's shield should be connected to non-current carrying parts of DUT. If the emissions noise is actually on the shield of the cable, ideally is to use connectors that have provisions for connecting or clamping the cable shield in a 360-degree bond. Using pigtail connections is a less efficient way to connect cable shields to their connector shield terminations. The longer the pigtail used, higher the expect emissions, thereore it’s recommended to use multiple short pigtails to the connector shield surrounding the internal wires. This will tend to cancel the resulting fields.
  • Bonding the cable's shield to DUT's shielded enclosure may work if local grounding is acceptable for that design. Most of the time the shielded enclosure or the heatsink is capacitively decoupled from supply return.
  • adding common mode chokes to DUT PCB design to minimize common mode noise sources.
  • istalling an external common mode choke around DUT's end of the I/O cable.
  • Expensive connectors have provisions for connecting or clamping the cable shield in a 360-degree bond, which is ideal. 

 

Ground Loop

A noise current sharing a common return impedance with a signal current.

 

Confined System

When connecting signal line cables within a confined system, the shield is connected at both ends in order to provide a signal return current path. 

  1. For high frequency digital signals above (10 to 100 kHz), proper magnetic field shielding requires a connection at both ends of the cable shield. This provides a return path for the high-frequency currents to flow back along the signal path.
  2. For frequencies greater than 10 to 100 kHz, the return current wants to travel the path of least impedance – that is back through the cable shield – due to mutual impedance coupling.
  3. For electric fields, connect only one side of the shield at the noise source (or sensitive analog) end.

Distributed System

For a system distributed across a larger area, with potential differences in the reference returns between one end of the cable and the other, the shield is connected only at the signal source end. The potential difference between the main controller digital return and and various sensor returns can be quite different. The result would be noise currents flowing in the shield. Such type of hybrid ground is used where a series capacitor is used to connect the non-source end of the shield to signal return (e.g. 300 feet long cables in aerospace industry). 

Opto-isolators, differential pairs, common-mode chokes are useful to “break” any noise currents in the shielded twisted pair of sensor cables.

Audio or power line frequencies

  1. For fixing a ground loop issue, grounding one end of the shield or blocking the low-frequency (or DC) component with a capacitor might work best. Isolation transformers may be used for both line and audio applications.
  2. For signal currents greater than 10 to 100 kHz, use a solid ground bond at each end of the cable shield. Ground loops just don't tend to occur above 10 to 100 kHz. 

NASA spec mention to:

  1. Ground one end (or use some form of isolation to break the loop) for low frequency ground loop fields.
  2. Ground both ends for shielding against external high frequency fields.

DUT with shielded enclosure using unshielded cable

  • Minimize the common mode (noise) current loop through either diversion (back to the noise source) d or blocking with some impedance. Break (or block) the loop with common-mode chokes at the I/O connector signal lines. Add transient protection devices to guard I/O connections against ESD and other pulse-type signals.
  • Insert a common-mode ferrite choke in the power and it's return lines. It's always good EMC practice to design in common-mode chokes in both the signal and power lines. 
  • Ensure each signal and signal return wire pair within the cable is twisted. This will achieve several dB of shielding effectiveness by itself.
  • If using a ribbon cable, make sure there are adjacent signal (and power) return wires for each corresponding signal (or power) wire.
  • If running a clock signal, make sure there are clock return wires on each side of the clock wire.
  • If all else fails, use a clamp-on ferrite choke around the cable, positioned right at the I/O connector.

DUT with plastic (unshielded) enclosure

There will inevitably be common-mode noise sources on the PC board. To keep these noise currents off our I/O and power cables:

  1. block the currents from getting to the cables with a ferrite choke or
  2. divert the noise currents back to their source.
  3. A combination of blocking and diversion is the best method. Higher-end handheld consumer products use a diversion plate under the PC board. It is a thin meallic plate or metalized film with one end bonded or clamped well to the I/O and power connector ground shells. This offers a low impedance path for the common-mode currents to flow back to the source through distributed capacitance. It also protects sensitive circuitry from external ESD currents injected at the I/O connectors. In addition, it serves as an image plane which helps reduce radiated emissions. The cable shield must be bonded in some way to the digital ground (if a signal or I/O cable) and power ground (if a power cable). Ideally, all I/O connectors and power connectors should be grouped together on one side of the board. If they are spread all around the perimeter, then any noise sources on the PC board are potentially driving the midpoint of a dipole antenna.

Low Voltage Differential Signaling (LVDS)

Switches about 1.2V at very fast edge speeds. Theoretically differential signals should never radiate, but ANY unbalances in line length or routing can cause common-mode currents to form.

Solutions:

  • use flat ferrite chokes
  • shielding the cable and connecting the shield back to digital return in several places at each end of the shield.

Troubleshooting:

  • use ferrite
  • install copper tape to one side of the cable to provide a path for any unbalanced common-mode currents to return to their source.

Shielded Enclosures and Gaskets

Both the compression of the shields and gaps/cracks in the gasket may may affect slot emissions. It’s really a factor of both the manufacturer’s recommended compression, plus how well the gasket installation is designed. Minimize the length of any gaps between any two pieces of metal enclosure. The leakage can be measured using a near field probe and sliding it along all the enclosure seams. Preferable to be done in ALSE chamber.

 

Earth Grounding Rod

In EMC testing is needed for establishing a voltage reference, discharge high transient voltages, static discharge, personnel safety.

 

Pigtail connectors 

Are an insulation displacement connector that are filled with a di-electric grease to prevent moisture from getting inside the connector. No need to strip the ends of  the wires, just insert them into the connector, then squeeze the blue cap down with a pair of pliers.

Connectors

When the source of the radiation is from common currents on external cables such as those that connect to peripherals, using a “better” cable often has no impact at all on the radiated emissions. That’s because the common currents are flowing on the shield of the cable.

It only takes 3 μA of common current flowing on the shield of a cable, 1 m long, to cause an FCC class B failure

The most important driving voltage for these common currents that causes EMC failures is ground bounce in the connector attaching the cable to the chassis.

Ground bounce is the voltage generated between two regions of the return path due to a changing current flowing through the total inductance of the return path.

The total inductance of the return path is related to the total number of field lines around the conductor per amp of current flowing through it. When the dI/dt of the return current flows through the total inductance of the connector, it generates a voltage, and this voltage between the chassis and the cable’s shield is what drives the common currents on the cable, which results in an EMC failure.

    

A coax cable will have no ground bounce because there no external magnetic field around it.

The signal current generates an external magnetic field composed of circular rings of field lines circulating in one is direction.

The return current, if symmetrical about the signal path, generates the identical rings of magnetic field around the cable, but circulating in the opposite direction. These two sets of magnetic field lines exactly cancel out and there is no external magnetic field.

But suppose at the connector, the return current is not perfectly symmetrical about the signal current. Maybe there is a pigtail, maybe the clam shell is not well metalized, or maybe the connector only makes contact at one or two points to the chassis.

Any asymmetry will mean the magnetic field lines from the signal current and return current will not perfectly cancel out. There will be some net magnetic field lines and this will result in some total inductance of the return path. 

In a 50 Ω coax cable, with a 1V signal, having a 1ns rise time, the signal and return current is about 1 V/50 Ω = 20 mA.

Even if the asymmetry is so light as to generate only 0.1 nH of total inductance around the return path of the connector, the ground bounce voltage generated would be 2 mV. 

If the impedance the common current sees returning through all those fringe field lines is about 200 Ω, this 2 mV of ground bounce voltage will drive I = 2 mV/200 Ω = 10 μA.

It only takes 3 μA of common current to fail an EMC certification test.

This ground bounce driven current in the cable shield will cause an EMC failure.

 

DUT configuration for CISPR 25 ALSE chamber ambient

28. October 2020 07:03 by Christian in EMC/EMI, EMC TEST PLAN, OEM Specs, Standards, Test Methods
The automotive OEM specs do not specify how to configure the DUT during CISPR 25 chamber ambient measurements.

The automotive OEM specs do not specify how to configure the DUT during CISPR 25 chamber ambient measurements. DUT must be unpowered, all other DUT support equipment must be powered and as much as possible functional to correctly evaluate RF emissions noise floor before start testing. This leaves at least three scenarios for how to configure the DUT.

 
1) Disconnect the DUT from test harness. 
  1.  Test harness connectors are removed from
  2.  DUT is unpowered.
  3.  The 1.7 m test harness is unterminated on DUT side, no potential ground loops with Load Simulator.
  4.  The 5uH LISN remains present.
  5.  The Load Simulator and all support equipment remains powered.
 
2) Disconnect DUT's B+ line from LISN output.
  1.  Test harness connectors are plugged into DUT.
  2.  DUT is unpowered by disconnecting B+ line LISN input from Battery.
  3.  The 1.7 m test harness terminated on both ends, therefore potential ground loops with Load Simulator are possible.
  4.  The 5uH LISN remains present.
  5.  The Load Simulator and all support equipment remains powered.
 
3) Diconnect DUT's B+ line from LISN output.
  1.  Test harness connectors are plugged into DUT.
  2.  DUT is unpowered by disconnecting DUT B+ line from LISN output.
  3.  The 1.7 m test harness terminated on both ends, therefore potential ground loops with Load Simulator are possible.
  4.  The 5uH LISN is not present anymore, and this somehow violates CISPR 25 requirement.
  5.  The Load Simulator and all support equipment remains powered.
 
Christian Rosu

Automotive EMC Load Simulators

28. September 2020 11:20 by Christian in EMC/EMI, EMC TEST PLAN, Load Simulator
During EMC compliance validations we monitor DUT (Device Under Test) errors visible for the occupant

The Load Simulator is defined in ISO 11452-1:2015 as: “physical device including real and/or simulated peripheral loads which are necessary to ensure DUT nominal and/or representative operation mode.”

 

During EMC compliance validations we monitor DUT (Device Under Test) errors visible for the occupant in vehicle in parallel with stored or not stored yet DTC (Diagnostic Trouble Codes). Disruptions in data bus or communication bus that do not set a DTC are not visible for the end user, since many of them are controlled safely by Vehicle Software & BCM.

 

 The driver can be distracted by vehicle cluster signs/indicators turning red, like an imminent hazard. 

  • If the incident self-recovers, it may not be a problem but it depends on DUT's Classification and Required Immunity Level.
  • If the DUT does not self-recover and require driver's intervention, then the LS support software must mimic the user response to resume operation (automation). Such anomaly is marked in a data log but should not be a reason to stop on-going testing.
  • If the DUT does not self-recover requiring Hard Reset (VBATT on-off-on), then it's really bad. This is like a stop show but make sure it is always driven by DUT, never the LS.
  • The pass/fail criteria mentioned in EMC test plan must guide your LS design effort, especially to decide on what type of FO monitoring equipment is needed.
  • Ideally is to use production intent DUT's I/O loading and Vehicle Software reducing the entire effort to monitoring the communication bus & FO equipment (e.g. FO Voltage Probes, FO Signal/Data Probe).
  • The moment you’re forced to use excessive HW/SW simulation, you practically spend more time validating the Load Simulator instead of focusing on DUT's EMC performance.
  • If possible, avoid using active electronic components for the LS placed inside ALSE chamber.
  • Use production intent loads, ideally EMC validated by OEM.
  • Use FO devices that are certified for 200 V/m, CISPR 25, 30KV ESD.
  • The support software should not stop the show if errors occur, only the DUT should be able to stop the show.
  • Pay attention how is the shielding of I/O lines terminated/grounded in vehicle and use if possible production intent cables and proper wire gauge.
  • For remote grounded module, make sure the only possible connection to battery negative pole is via supply return line. 
  • The LS metallic enclosure is bonded to GP (ground plane) being used as shield. 
  • The LS metallic enclosure is not being used as grounding point for DUT or LS electronics.
  • All signal return lines are closed to their source, never to GP.

 

Grounding Requirements

If DUT and LS grounding requirements are not defined by the automotive OEM EMC spec or Test Plan, then using automotive industry standards is acceptable (ISO 11452-2:2004-11-01, ISO 11452-4:2011-12-15, ISO 7637-2:2011-03-01, CISPR 25:2016-10-27):

"The DUT shall be placed on a non-conductive, low relative permittivity (dielectric-constant) material (εr ≤ 1,4), at (50 ± 5) mm above the ground plane. The case of the DUT shall not be grounded to the ground plane unless it is intended to simulate the actual vehicle configuration.”  

 

Preferably, the load simulator shall be placed directly on the ground plane. If the load simulator has a metallic case, this case shall be bonded to the ground plane. Alternatively, the load simulator may be located adjacent to the ground plane (with the case of the load simulator bonded to the ground plane) or outside of the test chamber, provided the test harness from the DUT passes through an RF boundary bonded to the ground plane.” 

 

“Bonded – grounded connection providing the lowest possible impedance (resistance and inductance) connection between two metallic parts with a d.c. resistance which shall not exceed 2,5 mΩ. Note 1 to entry: A low current (≤100 mA) 4-wire milliohm metre is recommended for this measurement" . This resistance needs to be verified with a milliohm meter. (ISO 11452-1:2015-06-01, MIL-STD-461G:2015-12-11).

 

 

Grounding Solutions:

  • Copper Tape (colored) with conductive adhesive.
  • Silver Tape with pressure sensitive adhesive (better contact), and tin-plating allowing soldering the tape directly to the ground plane, overall better resistance to corrosion.
  • Bonding Strap made from a semi-rigid flat metallic braid/weave that is copper tinned/untinned. Bonding straps are better than wires since their length to width ratio has lower inductance per unit length. The EMC test plan ahould specify that any ground straps used maintain a “5:1 length to width ratio or less” per MIL-STD-464C:2010-12-01. The impedance of ground straps at high frequencies varies with their width, length and addition of connectors (e.g. banana plugs). Since the ends of the braid may fray, ideally is to solder the ends of the braid. If adding a hole for a fastener (e.g. screw), the edges of the hole should be soldered to prevent fraying. The best grounding solution is to solder the braid to the ground plane.

Before using any of the above grounding solutions, the ground plane should be cleaned from oxidazation to achive better conductivity.

 

Grounding Point:

The EMC Test Plan should specify the DUT's case grounding point to ensure repeatsble results. The same for Load Simulator. 

 

Christian Rosu, Sep 28, 2020.

 

 

CISPR-25 RE per CS.00054:2018

15. October 2019 10:00 by Christian in EMC/EMI, OEM Specs, Test Equipment, Test Methods
CISPR-25 Generic Test Setup for compliance to CS.00054:2018.

CISPR-25 Generic Test Setup for compliance to CS.00054:2018

CS.00054 Radiated Emissions Block Diagram
 
The vertical monopole element is centered at 1m from the center of the 1.7m test harness. Note that 1.5m of the harness is running at 10 cm parallel with ground plane edge. The antenna counterpoise is placed +10/-20 mm vs GP. 
 
CISPR-25 Generic DUT Setup. The DUT is placed @ 20 cm from the edge of GP. The 1.7 m Test Harness is routed 90 degrees towards DUT.
 
The ground plane is connected to chamber's floor to a dedicated Earth Grounding Rod.
 
LISN (700 V DC / 500 A) & Load Simulator side of the test setup. 
DUT's B+ & GND lines are connected to LISN's outputs.
 
THE BICONICAL ANTENNA IN VERTICAL POLARIZATION. 
The antenna is centered on the 1.5m harness running at 10 cm parallel with GP edge.
 
THE BICONICAL ANTENNA IN HORIZONTAL POLARIZATION. 
The antenna is centered on the 1.5m harness running at 10 cm parallel with GP edge.
 
THE LOG PERIODIC ANTENNA IN VERTICAL POLARIZATION. 
The tip of antenna is 1 m away from the center of the test harness.
 
THE LOG PERIODIC ANTENNA IN HORIZONTAL POLARIZATION. 
The tip of antenna is 1 m away from the center of the test harness.
 
Octave Antenna Vertical Polarization with its aperture centered on DUT at 1 m distance from test harness.
 
Octave Antenna Horizontal Polarization with its aperture centered on DUT at 1 m distance from test harness.
 
Horn Antenna Horizontal Polarization with its aperture centered on DUT at 1 m distance from test harness.
 
Horn Antenna Vertical Polarization with its aperture centered on DUT at 1 m distance from test harness.

 

 
3-METER ALSE CHAMBER & Equipment Control Shielded Room.
 

ALSE CHAMBER EARTH GROUNDING ROD.

CISPR-25 RF emissions ambient test pitfalls

25. April 2017 15:48 by Christian in EMC/EMI, EMC TEST PLAN, Standards
CISPR-25 is not very specific in regards to chamber ambient test setup configuration in regards to d

CISPR-25 is not very specific about device under test and support equipment configuration during chamber ambient test. The automotive OEM require the ambient for RE, CE-V, CE-I with support equipment energized. The test laboratories will typically disconnect VBATT line from LISN output. The GND line remains connected to LISN. By doing so is assumed that DUT is not energized. The support equipment remains connected to the input of the LISNs being turned on (energized). The CAN bus is powered but w/o traffic. It is unclear if the load simulator energized it means powered but inactive (standby). By activating PWM pulses as inputs for DUT it may yield unwanted CE-I and RE ambient noise. All these aspects must be clarified in the EMC test plan.

In the sample presented the CE-V ambient noise is well below the 6 dB requirement. However, this type of noise is being captured while DUT's integrated buttons are being pressed and released via a pneumatic system with no electrical connection to DUT or test ground plane. Specifying that DUT must be unpowered may not be enough, the DUT's buttons should not be mechanically activated, nor its inputs subjected to electrical signals.

 
Christian Rosu