Some simple hints . for identifying and fixing EMI troubles. This second part of our EMI Troubleshooting guidelines is devoted to investigation and cure of EMI susceptibility problems. Since, according to our former article (Ref.1) conducted tests are faster and easier to perform, we will concentrate on these and avoid more complex and expensive radiated susceptibility test set-ups. The first half of this article will address the situation where the equipment at stake is still in the plant or lab, at the end of its development phase or in an early stage of production. The second half will cover the more difficult cases where the faultly equipment is already in service, installed at some customer site.
1. Susceptibility Problems on Prototype or Pre-Qualification item
Only a limited set of tests can be carried using essentially conducted excitation set-ups, without (or before) resorting to an actual Radiated Susceptibility check in an EMC lab. Preferrably, as explained in Ref. 1, susceptibility checks should be done after emission test have been carriedout, and eventually their non-compliance been investigated and solved. Notice that so-called ”Susceptibility” tests are in fact made to demonstrate EUT Immunity. Therefore, we should always try to reach, within a reasonable span, the actual ”Fail” threshold to get a measure of our Compatibility margin as compared to the Immunity ”No-Fail” requirements.
Like for emissions, we try staying away from formal Radiated Susceptibility (R.S) tests, and use substitute conducted tests instead. These conduction (or injection) tests are deemed to recreate on the EUT what its cables (power and signal) and internal circuits would actually receive from a severe, conducted or radiated threat. So, by order of simplicity and efficiency we will limit our tests to:
- Electrical Fast Transients (EFT Bursts)
- ESD ( Direct and Indirect)
- Bulk Current Injection (Continuous Wave )
2. Requirements for informal Suceptibility Test site and EUT set-up
The site requirements for an informal EMC Immunity test are the same as for emissions (Ref.1), except that the quiet RF ambient requirement is exactly the opposite: a quiet ambient is not necessary. Instead, since susceptibility tests will force strong HF signals or fast pulses, it is prudent to forbid, or temporary discontinue the use of sensitive electronic equipments ( computers, office equipment and instruments other than intentional EMI generators) in close proximity – say less than 5m from the test set-up. For the same mirror reasons as above, it is recommended to filter and if possible isolate the AC power branch circuit that feeds the EUT and set-up. This can be done by the same LISN and isolation transformer that were used for emission testing. During the susceptibility test, we will limit the Auxiliary Equipments (AE) to a minimum, using everywhere possible passive loads at the end of I/O cables. A good example is a Tx/Rx link that can be set in self-looping mode. As for emissions tests, the AEs must have an Immunity level equal (or greater) to the one we are testing for. If this is not feasible, an added, temporary filtering on their interfaces, or Coupling/Decoupling Network (CDNs) can be used to isolate the AEs from the EUT.
EUT Preparation for the test
Bear in mind that you will be ”teasing” the product to see if and how it fails. To easy-up the test, the EUT must be set in an automatic mode such as it performs its normal functions continuously. This can be done using an auto-test mode, or a specific emulation software which :
- Constantly activates a real, or close-to-real application w/o operator intervention
- Indicates clearly a disability or loss of performance by, like:
- displaying errors
- printing a report
- activating a visual, or audio alarm ( LED, buzzer etc ..)
- reset or lock-up in a frozzenstate
All this without the need of external instrumentation/ hook-up probes, unless they are fiber optic links. This is paramount to a meaningful test.
3. Testing for Immunity to Electrical Fast Transients
Of all suceptibility tests, the EFT (or ” Bursts”) is the simplest to install and run, and yet one of the most meaningfull. Thanks to its wideband spectrum covering kHz to more than 60MHz at each single pulse, it will track down most of the EMC weaknesses of digital as well as analog circuits.
Logic functions can:
- mistake the envelope of the bursts for a valid bits train of bits
- have some edge-sensitive gate inputs triggered by individual short glitches
- suffer clock stretching if the device fails to receive ACKs from the other device
- loose signal integrity due to high noise on DC supply and Gnd to which communication I/Os are referenced, violating the protocol specification
With analog circuits, the envelope of the burst can be detected in a same manner as pulse modulated radio signals.
Because it simulates the direct or indirect coupling of switching transients on the AC distribution wiring ( relays, repetitive arcing in switches, turn-off of heavy-duty motors, variable frequency converters etc…) caused by nearby power devices, the test is applied both on the power input and signal cables, per IEC 61000-4-4.
3.1. Instrumentation and set-up
EFT generator and EUT should be installed on the test ground plane (floor level or workbench with metal foil. Cables are laid at approximately 5cm above with insulating spacers (Fig.2). The generator should not be too close from the EUT, since 4kV pulses with 5ns riste times are hefty transients that radiates a strong E-field. Close exposure of the EUT could cause a direct box-to-box failure that would mislead the test issue. Same remark applies to the AE particiapting to the test.
– For superimposing HF pulses on power cord, the generator has its own internal Coupling/ Decoupling network, such as the fast bursts are injected on the EUT only and do not sneak into the lab power mains. Normal injection is CM ( Line 1+ Line 2 together vs chassis), but it is possible to select any combination of L1, L2 and earth ( Green/Yellow).
– EFT injection on I/O cables is a contact-less coupling (no stripping of cable sinsulation), applied with a special capacitive clamp, 1m long that can be folded over the tested cable (Fig 2). Beware that this folding armature is the coupling ”hot plate”, reaching up to 4kV and must not be touched or approached during the test. Since its size and weight can be impractical for some non-standard set-ups, it can be replaced by a tight wrap of aluminium foil, 30 to 50cm wide, around the tested cable, that will provide the necessary 60-100pF of distributed capacitance to the tested wires. The HV injection cable from the generator should be ≤ 1m long.
3.2. EFT Test routine
For all injections, test at least 30s on each polarity. Do not switch abruptly from (+) to (-) pulse polarity, always step through an ”Off” mode in-between. Consider this as a no-damage // no error criteria: the low amount of energy in each pulse is such that a damage level is seldom reached without a preliminary warning by a malfunction or error. DO NOT insist for reaching a true damage level! Keep your prototype safe and investigate the reasons of the temporary fault – or lock-up status.
a) Begin with power cable first. If nothing else specified, start with the lower level of IEC 61000-4-4, then increase progressively up to at least a No-Fail level #3 (2kV) which is genrally satisfactory. However, for a development test, it is advisable to try reaching one level higher to know your margin…
b) Increase the level progressively, preferably by octave steps, addressing combinations of L1, L2, L1+L2 and so on, with (+),(-) polarity. When (if) a FAIL level is reached, note the settings for all the last RUN levels. If the generator has progressive voltage setting,reduce the amplitude until you find the actual RUN/FAIL edge, such as you know your margin.
c) Inject onto each signal cable, keeping the capacitive coupling sleeve 1m away from the EUT. If external cables are shielded, the test is simply applied over the shield insulation jacket. If an AE was needed for the test monitoring, make sure that this AE has its input protected by proper filtering or snap-on ferrites toroids, located near the AE side.
3.3 EFT Test diagnosis and Fixes
If the test fails, concentrate on the failing port ( Power or Signal ) and check for:
- Incorrect mounting, or insufficient CM attenuation of the powerline filter
- Missing or poor cable shield connections
- Ungrounded metallic connectors
- Lack of, or improperly mounted CM filtering capacitors (long leads)
- I/O wiring running too far inside the EUT before they are filtered
Improve immunity by correcting the above, eventually adding filtered connector/adapters, or split-ferrites (2 or 3 turns) to the faultly I/O cable. Re-do the test, starting from the latest ”NO FAIL” level as a reference, to grade the improvements.
Warning: if a fix does not seem to bring a significant improvement, do not remove it. Add another one: EMI is often a multipath scenario.
4. Electro-Static Discharge (ESD) Test
Next to EFT, the ESD test is a also powerful teller of potential EUT weaknesses. It covers in a single flash an even wider frequency spectrum, up to 350MHz considering its 0,7-1ns risetime, but at difference with the EFT conducted bursts, it generates also a strong and fast rising H-field on the EUT box itself. You are directly zapping in a few ns a broad frequency spectrum that would take hours to explore during a real RF immunity test.
4.1. Instrumentation and test set-up
- The most common instrument is an ESD generator conforming to IEC 61000-4-2, with 330ΩInternal resistance and 150pF capacitance (Fig.3). Other simulators can be used according to the familly of EUT being tested.
- Horizontal and Vertical Coupling plates ( HCP, VCP) if the EUT must be tested for Indirect discharge (IESD). EUT preparation and set-up precautions are the same as for other immunity tests, with a strong emphasis on proper decoupling or safe distance of the associated AE, if any. The location of the ESD current ground return lead is critical: it should be attached to the test ground (floor) plane on the same EUT side as the face being tested.
4.2. ESD Test routine
At difference with the EFT continuous train of high voltage bursts, ESD replicates an isolated, rarely-occuring event. Therefore, exceptions accounted, there is no point in applying an ”absolutely-NO error” criteria. If the EUT has efficient error detection and recovery mode, this can be considered as an acceptable response, provided it does not result in a lock-up status, requiring a manual reset or creating a safety issue.
a) Establish the Direct or Indirect (IESD) procedure you will be using (ref. IEC 61000-4-2). Briefly, if the EUT is metallic on all its faces, only direct ESD will be performed. Otherwise direct discharge will be attempted on all touchable or arc-reachable metal parts, the rest of the EUT being tested by the IESD method on the HCP and VCP. Insist on cable entry areas, keyboard and display edges, touch sensitive panels. Mark-up the pre-determined discharge points directly on the EUT faces or on 3-D sketch.
b) Start with a low level (e.g 2kV) in a fast repetitive mode, like 5 or 20 pulses/sec for a coarse survey of all the discharge zones.
c) Increase the level progressively by 2kV steps until you reach the desired criterion plus a 1-2kV margin.
d) If failures are noted, abandon the fast repetitive mode and return on the failing area with a single shot, or ≤ 1 pulse/sec. mode, such as allowing for the EUT self-recovery process, if any. Run a minimum of 50 discharges per point. ESD is a super-short duration, random event: applying only a few discharges (the 10 discharges recommended in IEC norm are definitely not enough) would turn into a risky hit-or-miss process.
e) Record the Run/Fail levels, documenting the type of error/malfunction: Recoverable? Non-recoverable? Loss of data?
4.3 ESD Test diagnosis and Fixes
If the test fails, try understanding what are the sensitive circuitry that has been exposed to the close coupling of the discharge. Be on the look for long slots or seams leakages on the EUT enclosure, that act as re-radiating antennas to the nearby wiring or PCB traces inside. When trying fixes such as :
- Shielding (or shielding improvement) of I/O cables
- Shielded connectors, ferrites, decoupling capacitors
- Better bonding of panels, doors etc … to the main chassis (for metal cabinets or racks)
- Reduction of slots and seams leakage by gasketting or adding more screws
- Increasing the air gap crossing distance for zones where air discharges have occured
Always grade your progress via new Run/Fail mappings. Like for EFT, do not remove a fix that gave disappointing results. Accumulate them until success, then and only then, remove fixes sequentially to find the ones that could be useless.
4.4 Indirect ESD on a PCB prototype or pre-production item
If your analysis of the ESD test has indicated which functional circuit is disturbed, it could be worth focusing on the specific PCB (or set of) that is hosting this function. For this, the EUT housing should be removed, keeping only the metallic base plate, or if it is a plastic case, the topless EUT will be installed on a large aluminium or steel plate. Having spotted the circuits that are susceptible to ESD, identify some connector pins or printed traces that can ”witness” the functional status of the essential circuits (for inst. WatchDog, Reset, Power ON, IRQ etc…). On these traces or pins, solder a LED whose state will indicate a Good/Bad status for the corresponding function. Disconnect all I/O cables that are not absolutely necessary. The PCB can eventually made fully a stand-alone by attaching a battery pack to the board, equipped with clip-on teminals for easy turn-off.
Test Routine for IESD on a PCB
Attach the ESD gun ground post to the test ground plane by a short grounding strap (fig.5). The discharge is applied to the EUT bottom plate (or substitute coupling plate), following a contour of ≈ 5-10 cm from the board edge. Start with low level like 0,5kV and discharge the tip of the ESD gun all around the marked-up designated contour, trying to reach progressively 4-6kV. Although it gives no quantitative assessment, this test is a simple way to pin-point weak spots and find out the layout flaws or specific components responsive of the poorest immunity, thus inviting improvements. Then, try temporary fixes and grade your improvements. DO NOT try direct contact discharges on PCB traces or connector pins.
5. Bulk Injection of HF current on external cables (BCI test)
Next to EFT and ESD, this test is a powerful indicator of EUT susceptibility to strong RF ambients, from intentional radio transmittrers. It consist in injecting via a clamp-on current transformer the same HF signal as the cables would see if they were exposed to a real field. The difference being that you do not need to generate this field. This test is required by Mil Std 461/CS114, with values of injected HF current ranging from 15 to 300mA, and to some extent, is also part of IEC immunity test 61000-4-6. Unfortunately, running this test makes your EUT cables radiate a significant RF field, which could make you liable for an illegal interference to public radio and TV as far as km away. We do not recommend trying it elsewhere than in a shielded room, or at least an underground place.
5.1 Quick, simplified calibration of the injection set-up
The key element for making this test more than a qualitative investigation is a proper calibration of the injection tool (Fig.6). This calibration will allow a quantitative assessment of the radiated field coupling that you mimic by this substitution method.
A simple rule is as follows:
When cable length exceeds ≈ . wavelenght, the current Ii induced by a given Field in V/m no longer depends on frequency and settle to a maximum value around 1mA / (V/m). Only the cable height above the local ground plane has a small influence on the amplitude of Ii. More detailed values of cable Common Mode current is given on Table 1.
For a vehicle or aircraft equipment that can be exposed to a 100V/m field, the calibration current in the test jig must be set to: 1,5 mA x 100V/m = 150mA. For a civilian, table-top equipment that can be exposed to a 10V/m field, calibration current must be set to: 6,5 mA x 10V/m = 65mA
5.2 Bulk cable injection test routine
The EUT is installed like for the Conducted Emissions (Ref 1) or EFT test. The tested cable is laid 5cm above the ref. metal ground plane. The other end can be terminated on a passive load. The AE representing the peripheral unit is isolated, HF-wise, from the test by a Coupling Decoupling Network, or at least a set of CM ferrites, close to the AE ( Fig.6). The injection probe is fed by a signal generator, teamed with a power amplifier capable of feeding a sufficent power to the probe, taking into account the probe transfer ratio, some cable loss and a 4 times margin (6dB, in power).
Starting with low levels, increase the generator level up the its pre-calibration setting. The input signal must be amplitude-modulated by a low AM , with at least 50% modulation. For sake of checking that a loop current actually exist, a monitoring current probe is added, near the injection probe. But when monitoring the injected current, do not try re-adjusting the current for maintaining the calibration value. Let the bundle current establish by itself, depending on the actual impedance of the cable-to-ground loop. What counts is that the induced open-voltage keeps the value that was set during calibration.
6. EMC hardening for equipment installed at customer site
EMI problems occuring in the field are a difficult challenge. This is a case where there is very little possibility, if at all, for modifying the equipment. All solutions are relying on cable layout, shielding, additional I/O and power filtering, power conditioning and eventually room or site treatment. Even if, as of today, most electrical/electronic equipment put on market has to meet rather severe EMC requirements, suceptibility and emissions incompatibility problems still happen. Reasons for this are multiple:
- a recent, large system could include older equipments that do not comply with current EMC norms
- poor cabling and grounding practices by unaware contractors that degrade an otherwisecorrect immunity
- some sensitive equipments that are impossible to harden, by their very nature (magnetic sensors, E-Beam microscopes) or generate strong disturbances (elevators, arc welders, RF medical imaging, some indusrial process)
The following basic guidelines will save you from frustrating experiences at customer’s site where people telling you that they have already tried everything, rely on your magic wand for making miracles.
6.1 Before going to the site
In many cases, we have found that rushing to a problem site because the user is pressing you may induce two mistakes:
- you do not collect enough data for preparing a sound investigation scheme
- you either bring too many or too few instruments or measuring devices, leaving behind the one precious accessory that cannot be found on site.
Instead, there are things you must check in advance (”Forewarned is forearmed”):
a) Interview the users
- Is the problem continuous or intermittent ? in the latter case, is it predictable, like being re created on request ? This may dictate the equipment you will bring
- Does the failure correlate with a specific time of day ? Or with certain loads being on/off the power distribution ? When people use portable transmitters ?
b) Any instrumentation available on site ?
- Oscilloscopes: which Bandwidth ? Battery-operated ?
- Any recorders for intermittent problems ?
Thus you will bring whatever is necessary to complement the instrumentation available on site. In any case, don’t forget bringing an EMC clamp-on current probe.
c) Sometimes, based on the above, a fast, rough estimate can be done on the likelihood of the problem, if the data you have collected suggest what is the source, its frequency range, what and where is the victim device and what are the possible coupling path.
d) Plan some advance strategy for diagnosis and fix. Don’t count that the first attempt will work, anticipate alternate options.
6.2 Upon Arriving at the site
Perform a first Inspection of the EUT and its physical Installation:
a) On the faultly equipment: Does it have power line filters, with CM and DM elements? Are they bonded to a chassis or a grounded metal plate, and correctly mounted (no input and output wires entangled together)?
b) General Power grounding/Earthing scheme: Ask the customer’s staff or facility maintenance people about it. Don’t take anything for granted: bonding points that seem to make good contacts don’t; Grounding straps that people swear have been installed are missing or loose; Asked about the neutral earthing scheme, six different people will give you six different versions; Once checked, you’ll find that none of them is true.
c) Power and signal interconnecting cables layout. Examine all of them. Are there some power and signal bundles running thightly close to each other, or near to other power cables carrying large currents? Are they shielded? If so, where and how are the shields grounded? Are they in metal raceways, with continuous metal-to-metal bonding all along?
6.3 Continuous or quasi-continuous EMI problems
By this, we mean that the problem is always present or repeats within a reasonable observation period, such as it can be probed and quantified: S/N degradation, bad Bit/Error rate, false alarms…etc. If it has been checked that the equipment is in otherwise good condition, a continuously occuring problem is a blessing of sorts, since it eases-up the search for the source and the subsequent fixes. A procedure is recommended on Fig. 7.
6.4 Measuring CM currents as ”witnesses” of the EMI paths
Once (if) the source has been identified, the likely coupling path(s) can be identified by a quick measurement of the Common Mode HF currents at the following locations:
- power input wiring (CM)
- signal and various interconnect cables (CM)
- earth wires
Make sure that the culprit source is functionning when you make your measurement. An oscilloscope with sufficient bandwidth (≥100MHz) is recommended, using eventually its memory function if the culprit source is operating with various sequences. Make sure to use a 50Ωinput for the scope. Using the current probe transfer impedance Zt, convert your voltage readings into current, using:
I (Amp) = V / Zt(Ω)
Be careful that the probe response (Zt) is not always flat with frequency, so you need at least a rough idea of the frequency of the probe voltage signal. (Fig. 7.)
Upon the results of this CM current survey on all cables, a further investigation can lead to identify the cable(s) that is the likely coupling path for the problem. Fig.8 give guidance for treating such paths.
6.5 Intermittent Problems
These represent a respectable % of field calls, with a long search to find a correlation between a rarely-occuring problem and some external cause. If the source is found within the customer premises, it may be worthwile seeing if the noise can be suppressed at its source, by adding filters, or box / cable shielding. Sometimes after a patient search, the external cause is found but you have no control on it (an elevator nearby, an arc-welder in a machine shop one block away etc..). Ultimately, no correlation has been found. These are cases where we will use the forced-crash approach, byrecreating artificially an event that we cannot afford waiting for.
6.6 Failures that cannot be correlated: ”Forced Crash” Approach
If answer to Box ”B” routine in Fig.7 is NO, no correlation has been found between the failure and some intermittent event on the site. Such events can be artificial or natural, occuring at irregular times in this site like severe electrical fault triggering overcurrent protections or ground fault detectors (GFD), or eventually storms. In such case, some searching in the history of local lighning strokes, or power disruption in some part of the building can provide clues if they correlate with the moment of the failures. Although more difficult to track, a second random cause, may be the normal ON/OFF switching of low power electromechanical switches, relays, thermostats etc, in close vicinity (less than 10m) from the victim.
Finally a third threat, ESD, has generally a seasonal relationship, like increasing during cold and dry weather, aggravated by high personnel activity and type of floor covering. But it may be also machine-originated ESD, with some production tools, robots etc, causing ElectroStatic charging/ discharging phenomena with peak currents exceeding 50A, that may go un-noticed.
In all these cases above the ”Forced Crash” is an ultimate strategy where you decide that you cannot wait for some hard-to-catch problem to showup, so you deliberately inject onto the equipment a transient pulse (or train of pulses) that is ”shaking”, EMI-wise, the victim equipment. The best simulators for this kind of artificial threat are the EFT and ESD. Starting with low levels, the applied pulses will be increased until the equipment starts exhibiting malfunctions. If the ”FAIL” threshold is significantly lower than the immunity level that should be expected for this equipment and environment, you will harden one or the other until this installed equipment meets or exceeds the immunity for its category. Accordingly, over some probation period of a few days or weeks, the user should report a reduction, or disappearance of the problem.
The driving assumption is: If it withstands the standard immunity level on site, this equipment is vaccinated against any kind of short, fast-rising transient, even if the true cause is never found.
6.7 Common Precautions to EFT and ESD on-site ”forced crash” method
The principle of the set-up and routine are basically the same as for development tests, with some noticeable differences in the actual application, considering that you miss the commodities of a design/development lab. Instead, you are on a customer’s site, with probably other equipments that are still running, people doing their regular work etc… This implies some precautions, like working off-hours or temporary isolating some section of the plant or building.
a) There will probably be no ground plane, so you will use the next, closest metal framework (building girders for instance) as a reference for connecting a temporary aluminium foil. The EFT generator will be grounded to this foil. The EUT will NOT be grounded to the foil, but will keep its normal (presumably safety earth wire) connection to the power panel earth bar.
b) You are testing an equipment that is actually in use. As the test is progressing, watch for the symptoms that have been noticed by the users. Make sure that your crash evaluation does not lead to material damage or eventually safety hazard. Ask for putting in standby mode any peripheral that could create such a risk.
c) For EFT, if the EUT power cord does not fit into the generator outlet, use the aluminium foil wrapping on the power cord, like for the signal cable (See Sect. 3.1).
6.8 Forced Crash using EFT simulator
This method is, by far, the preferred test to start with, since it requires less time and preparation, being performed on power cables and signal cables only, that are typical weak spots of a system. Test ALL cables, starting from a low level. For those AE which must functionnaly operate with the EUT, use ferrite decoupling, as described in Sect 3. The EUT must resist on site, to levels that are at least equal to the IEC 61000-4-4 Table (Fig 2). If theselevels are not met, use the fix hints in section 3.3. Some of these fixes maybe applicable on-site, without too much disruption for the user. Other fixes could make-up for design weaknesses that you cannot improve on an installed equipment. Those you will apply on the installation itself:
6.9 Forced Crash using ESD simulator
This method is for the cases where there is a suspicion that the problem could be ESD (answer YES in Fig. 7 box D. It precisely reveals weak spots on the equipment box and I/O ports area, that an EFT test may have missed. The test procedure is similar to that of lab test, Sect.4. For an installed equipment, the NO-Fail levels of Fig.3 are recommended.
This supposed to be a No-Damage test. Errors are normally tolerated if they are self-recovered, without serious prejudice to the user. Hard errors, especially if they resemble the ones that the EUT user was complaining about are not acceptable. Do not discharge directly on EUT parts that are directly wired to sensitive circuits inside (like connector pins of Digital Bus interfaces, video or RF inputs etc ..), unless the normal factory test of the EUT did include this requirement. Due to the high voltages (up to 15kV) on the tip of the ESD gun, this method can be dangerous for an equipment whose susceptibility is unknown (never tested by the manufacturer) or even for personnel. Make sure your temporay test area is bound with visible warning tape.
7. Power Line Monitoring
There is a domain that an on-site EFT test will not reveal: long-duration power mains disturbances, typically longer than a few tens μs. These are difficult to simulate on-site, because it would require injecting significant disturbances that could affect part or all the AC distribution in this customer’s facility. So, if occasional power line disturbances are suspected, it is recommended to install a power-line disturbance monitor. This ”spy” instrument will survey constantly the short overvoltages and undervoltages, indicating their amplitude, duration, and precise time of occurance (day, hour, minute). If the sensors are correctly installed, they will show if the disturbance was DM (line-to-line) or CM (Line to Gnd). Choose (they are often available for rental) an equipment with battery back-up, and select a sensing threshold equal to what could exceed the normal (proven) immunity of the EUT to power line.
If after some observation period, the EUT had suffered the failures the user was reporting, you will find if they correlate with some power mains disturbances, and eventually install surge protection devices or a power conditioning (UPS).
Michel Mardiguian EMC Consultant, France
- EMI troubleshooting, Part I, September 2017 issue of EE magazine
- M.Mardiguian, EMI troubleshooting Techniques (Mc Graw Hill)
- M.Mardiguian, Electrostatic Discharge ( Wiley)
- H. Ott , Workbench EMC measurements, H.Ott 2012 www.hottconsultants.com
- K.Wyatt, Troubleshooting Radiated Immunity for Medical Products, Interference Technology Oct 9, 2017