emc basics
TRANSCRIPT
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMCDaryl Gerke, Kimmel Gerke Associates
2/24/2011 1:09 PM EST
(Editor's note: we are pleased to begin a new series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Here is his introduction to you, followed immediately below it by first
entry in the series, which looks at printed circuit board EMC, starting with the clock circuit.)
Hi! I'm Daryl Gerke, and I am delighted to again be working with the good folks at EDN, now part
of UBM Electronics. Way back in 1994, my business partner (Bill Kimmel) and I wrote the
original EDN Designer's Guide to EMC. At that time, we wanted to share our collective half
century of industry experience with our design colleagues.
As full-time EMI/EMC consultants since 1987, we had seen a multitude of EMI problems.
Although spread across different industries, the underlying causes were often similar. So when
former EDN Editor Steve Leibson (still blogging for EDN) put out a call for design tutorials, we
responded with what eventually became the Designer's Guide.
Fast-forward 17 years and the EMI problems are still with us. At this stage in our careers, we've
become "old warriors" with over 80 years of collective experience. Our goal now will be to help
you sharpen your "EMI spears". I hope you enjoy our efforts.
We will focus on design and troubleshooting, not test and regulations. As we are fond of saying,
"An ounce of EMI prevention is often worth a pound of EMI shielding." This is best accomplished
at the design stage, when most EMI fixes are cheap or even free. Thus, we felt EDN was the
perfect place to share our insights.
For personal information, please visit http://www.emiguru.com. You'll find lots of additional
resources there, too. Most are free, and a few are available for a nominal charge. This blog also
originates there. Finally, if you are curious about consulting, check out my other blog at
http://www.jumptoconsulting.com. (Be sure to visit the special welcome for geeks.)
And now, let's get started on our first entry in the EMC series:
Welcome to the first post on EMC issues here at Planet Analog and EE Times. Since this is a
design-oriented site, we'll begin with some EMC design issues. Specifically, we'll address what
should you look for when doing an EMC design review on your printed circuit boards (PCBs).
Not doing these kinds of reviews? Well, you should. An hour or two at the beginning the project
can save thousands of dollars and a lot of grief at the end of the project. One extra trip to the
EMC lab can easily cost $10K or more when you include your engineering time. And who know
how much it costs by being late to market?
Convinced yet? I hope so. A quick EMC design review is pretty simple. We do these reviews for
clients all the time, and you can do them for yourself. For the next half dozen posts, we'll give
you a quick overview.
To start, we look at the following five critical circuits: clocks, resets, power regulators, analog, and
I/O. These five circuits probably account for 90% of the EMI problems at the PCB level. Today,
we'll look at clocks.
As the most periodic of signals, clocks are the richest in harmonics that often result in radiated
emissions problems. As a minimum, we worry about the first 20 harmonics. Thus, for a 50 MHz
clock, we are concerned all the way up to 1 GHz. But that is just a starting point, as we've seen
higher harmonics cause problems, particularly if they excite a resonance.
We also check for clock-like or clock-derived circuits, such as memory enables or busses. These
may operated at a sub-harmonic of system clocks, but can still cause emissions problems. Of
course, many systems have multiple clocks, so all the clocks be addressed.
Prior to testing, we recommend making a chart showing clock harmonics all the way up to the
maximum frequency of concern for radiated emissions. Then repeat the chart for all clocks
divided by two, and again for all clocks divided by four. (Other division ratios may be needed if the
system uses other clock derivations.) All this can be done easily on a spread sheet.
This data is very helpful during testing. If a radiated emission failure occurs, you can quickly
check your spread sheet to determine which clock is a culprit. If a subdivision, this may also point
to additional clock-like circuits.
Incidentally, if your failure frequency is NOT on your charts, the test data may be pointing you to a
parasitic oscillation. We'll talk about these in a future posting.
Finally, typical solutions for clock problems include power decoupling, series termination (or even
filtering) on outputs, and attention to clock trace routing. Also, keep clock circuits away from I/O
ports to prevent unwanted radiated coupling to the I/O. In extreme cases, selective shielding may
also be needed on the PCB.
We'll revisit these issues in more detail in future entries. Next up: resets.
About the author
Daryl Gerke, an EMI/EMC consultant since 1987, along with business
partner Bill Kimmel, focuses on design and troubleshooting (not test and regulations). He and Kimmel
have been chasing EMI problems for over 80 years (combined, of course.)He is a published author and columnist,
and theirEDN Designer's Guide to EMC (1994) is still in relevant and in demand. He can be
reached viahttp://www.emiguru.com or his other blog athttp://www.jumptoconsulting.com
EMC Basics #2: Resets as critical circuitsDaryl Gerke, Kimmel Gerke Associates
3/7/2011 6:12 PM EST
[Editor's note: we are pleased to continue our new series on the vital and sometimes
unappreciated topic of electromagnetic compatibility (EMC), presented by well-known expert
Daryl Gerke of Kimmel Gerke Associates. There is a link to Entry #1 at the end of this item.]
After clocks, we like to focus on reset circuits when doing an EMC board
review. The reset circuits are often upset by transients such as ESD (electrostatic discharge) or
EFT (electrical fast transient). A secondary threat is RFI (radio frequency interference.) The latter
is not very common, but we have seen it happen at high RF levels. Fortunately, these problems
are easy to prevent.
False reset effects can range from simple nuisances to a complete system lockup. The actual
response is often dictated by the system software. As such, it can be easy to fix a reset problem
at the software level. Alternately, one can often push the "reset" button or even power down to
restart the system. In critical systems, however, this may not be an option so hardware fixes may
still be needed.
The first thing we check is adequate power decoupling. This is particularly important when using
a "voltage monitor/power-on reset" IC. Since these often use sensitive internal comparators, even
a short disturbance on the Vcc can initiate an unwanted reset.
Incidentally, this was a serious problem with early reset ICs. Since then, the IC vendors have
incorporated small internal delays (such as Schmidt triggering) with good success. Nevertheless,
we still pay attention to decoupling to assure an extra margin of safety.
Next, we check the inputs. On simple devices, there may be none, as the IC relies solely on the
Vcc rail. More sophisticated devices, however, may include a separate sense input or an external
reset input. The latter is often connected to a button, with the input pulled high or low to initiate
the reset. Both input types may need light filtering—a 1000 pF capacitor can work wonders.
If the external reset goes off the board, additional filtering may be needed. Consider a ferrite in
series with the button, followed by a 1000 pF capacitor at the IC input. Yes, this will slow down
the system response to a reset, but if that is a problem, just push the button faster! After all, the
typical delays are less than a microsecond.
The last thing to check are the outputs. If they run more than an inch or two, consider a RC filter
at the IC output, plus 1000 pF capacitors at the loads. A better choice, if available, is to place the
reset controller IC close to the device(s) it is controlling.
To recap, the typical hardware solutions for reset problems include decoupling of the Vcc, filtering
of inputs (particularly if an external reset control is used), and filtering of outputs if the trace
lengths are over an inch or two. Finally, do not overlook software fixes, a very cheap and very
effective solution to reset problems.
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #3: Voltage regulators as critical circuitsDaryl Gerke, Kimmel Gerke Associates
3/24/2011 2:16 PM EDT
[Editor's note: we are pleased to continue our new series on the vital and sometimes
unappreciated topic of electromagnetic compatibility (EMC), presented by well-known expert
Daryl Gerke of Kimmel Gerke Associates. There is a link to previous entries at the end of this
item.]
After clocks and resets, we like to focus on voltage regulator circuits when doing an EMC board
review. Voltage regulators can cause both radiated emissions and susceptibility problems.
Emissions are the result of parasitic oscillations, and susceptibility is the result of RFI (radio
frequency interference.)
The first problem, parasitic oscillations, are due to high-frequency
feedback from the output to the input. The criteria for an oscillator are (a) 180°phase shift from
output to input, and (b) gain greater than one at the feedback frequency. Or, as the old saying
goes, "Oscillators won't, but amplifiers will."
With today's devices, oscillations typically occur in the 100 to 500-MHz range. We've seen these
problems occur with both linear and switching regulators. Both types have feedback, and both
have gain. The resulting levels are often high enough to cause radiated emissions failures during
EMI testing.
Note that these are free-running oscillations. As such, they will NOT be exact harmonics of any
oscillators. Also, the frequencies may vary from system to system. For example, it is not unusual
to see an oscillation at 222 MHz in one system, and to see an oscillation at 231 MHz in another.
Same circuits, same layout, different parasitics. These are useful clues when troubleshooting
radiated emissions.
The second problem, RFI, is due to rectification. Even a small amount of demodulated AC or DC
voltage at a critical feedback node can drive the regulator out of range. In digital systems, this can
result in unpredictable and unrepeatable behavior, such as system lockups. In extreme cases, the
out of specification voltages can even change the state of programmable components.
Like parasitic oscillations, the RF susceptibility problems typically occur at 100 MHz and above.
At those frequencies, the dimensions of the circuit boards and traces become efficient antennas.
This is particularly problematic for circuit boards that are not in a shielded enclosure.
Fortunately, both problems are easy to prevent. Small high-frequency capacitors placed directly
across the component inputs and outputs will "short out" both adverse effects. We typically
recommend 1000 pF capacitors at these locations. Keep the leads short!
Note that many regulator circuits have electrolytic capacitors across their inputs or outputs. These
capacitors alone are NOT adequate at high frequencies. Most electrolytics are not a good "short"
at frequencies above 10 MHz. In those cases, you must add the small capacitors in parallel,
located at the device. Just like stereo speakers, think "woofer-tweeter" capacitors.
To recap, add small high frequency capacitors at regulator device inputs and outputs. These
provide cheap insurance against unwanted parasitic oscillations, and at the same time protect
against RF threats. You will never know when they are working, but if you need them and they
are not there, you WILL experience EMI problems.
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #4: Analog devices as critical circuitsDaryl Gerke, Kimmel Gerke Associates
4/18/2011 8:29 AM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. There is a link to previous entries at the end of this item.]
If a circuit board includes analog circuits, we also like to focus on those for an EMC board review.
Similar to voltage regulators (discussed in the previous posting), analog circuits can also cause
radiated emissions and susceptibility problems. Once again, the emissions are the result of
parasitic oscillations, and the susceptibility the result of RFI (radio frequency interference.)
Analog circuits differ, however, from voltage regulators. In
analog circuits, the RF susceptibility problems predominate. Parasitic oscillations (resulting in
unwanted emissions) can occur, but are less likely. Since voltage regulators are already
feedback-based devices, unwanted oscillations are more likely than with non-feedback based
devices.
The RF susceptibility problems are typically due to rectification. A CW (continuous wave) RF
threat can result in a DC offset, which can often be blocked with a capacitor. With a modulated
RF threat, however, a demodulated signal is the result. If the demodulation is within the expected
signal passband, there is no way to filter it.
That is why modulation is used during most RFI tests. A 1000-Hz modulation is typical, but that
may be further constrained depending on the equipment under test. For example, most medical
devices require modulation within the passband of the physiological function to be measured (a
few Hertz or less is typical.) The goal is to uncover unwanted effects in the presence of RFI.
Fortunately, the problems are easy to prevent at the circuit level. Start with high-frequency
filtering on the analog circuit inputs. As a minimum, place small capacitors (100 to 1000 pF
typical) placed directly across the component inputs to "short out" the RF energy. For additional
protection, these capacitors can be augmented with series resistors or ferrites. The goal is to
prevent the RF from reaching critical input circuits in the first place.
Due to circuit bandwidth, the allowable input filtering may be limited. Fortunately, most analog
circuits operate at audio frequencies (or below), so the small amount of capacitance needed for
RF protection usually does not affect normal operation. The key is to use enough filtering, but not
too much. In some cases, even 10 pF across inputs is enough to protect against RF threats.
Added protection may be needed. Power decoupling should include high-frequency ceramic
capacitors (1000 pF typical) in addition to larger low-frequency capacitors. These should be
installed adjacent to the analog devices. In extreme cases, output filtering and even local
shielding may be necessary for very sensitive devices (or very large threats.)
To recap, consider high-frequency protection for your analog circuits. Don't assume that just
because they operate at low frequencies they won't be affected by high frequency (RF) threats.
With today's modern devices, sooner or later RFI problems will occur unless protection is
included. Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #3: Voltage regulators as critical circuits
EMC Basics #5: I/O as critical circuitsDaryl Gerke, Kimmel Gerke Associates
5/2/2011 10:36 AM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. There is a link to previous entries at the end of this item.]
Last, but not least, we like to look at the I/O (Input/Output) circuits during
EMI circuit board reviews. As a port of entry for external currents, I/O circuits are particularly
vulnerable to threats like ESD (electrostatic discharge) and RFI (radio frequency interference).
Radiated emissions are also a concern, as the I/O circuits are often the last chance to keep
unwanted currents from leaving the board.
There are several parameters of concern -- internal/external, digital, analog, relays, and contact
closures. They each present different problems, and may require different solutions. One size
does not fit all.
Internal/external -- I/O circuits connected to external traces/cables are the primary concern. I/O
circuits are connected to wiring or traces that leave the board, and that can act as "hidden
antennas." If the system is shielded, internal I/O is less of a concern. If not shielded, however,
both internal and external I/O deserve EMI attention. Regardless, we like to check out both types
of I/O.
Digital inputs/outputs -- The key concern for digital interfaces is ESD. A high level discharge
may cause damage, while lower levels may simply cause upsets. The solutions include transient
protection (must be fast enough), filtering, or even software (ACK/NACK protocols, etc.)
A secondary concern is radiated emissions, with small currents sneaking out the I/O port. The
resulting voltages are usually so small you can't see them with an oscilloscope. Radiated
susceptibility is rare with digital I/O, although we have seen problems at very high RF levels. The
solutions for both radiated problems include filtering at the interface and/or or shielding of external
cables.
Analog inputs/outputs -- The key concern for analog interfaces is RF. High RF levels can cause
rectification in the I/O circuits, resulting in either a DC offset (no modulation) or a low frequency
AC signal (with modulation.) If the modulation is in the signal passband, you can no longer filter it.
Thus, the best strategy is to prevent rectification from occurring in the first place.
Typical solutions include high frequency filters and/or shielding of the external cables. If diodes
are included for ESD protection, the filtering must precede the diodes to prevent rectification at
the protection diodes.
Relay outputs -- Since relay drivers are usually digital, the regular digital concerns apply. In
addition, inductive transients from the relay coils may pose a self-compatibility problem. Snubber
circuits may be needed at either the relay (best) or at the driving circuit on the boards.
Contact inputs -- Since the receiving circuits are usually digital, the regular digital concerns
apply. Do NOT assume that because contacts are relatively slow that they are immune to EMI
problems. Remember, high frequency EMI currents love to exploit unprotected low frequency
ports.
To recap, when designing or reviewing circuit boards for EMI, ALL of the I/O circuits deserve
EMI attention! Even one unprotected I/O port can wreak EMI havoc.
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #3: Voltage regulators as critical circuits
EMC Basics #4: Analog devices as critical circuits
EMC Basics #6: Looking at circuit board "stackup"Daryl Gerke, Kimmel Gerke Associates
6/5/2011 5:09 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries at the end of this item.]
After the critical circuits, we like to examine the printed-
circuit board (PCB) stackup. Like the schematic, board stackup decisions are often made prior to
routing and placement, so this is a good time to look at the proposed stackup. When routing and
placement are complete, a second EMC review of the board is prudent. Stackup matters for
EMC!
The first stackup question is the number of layers. If it is one or two layers (no power or ground
planes), this raises serious EMI serious concerns about both the layout and the technology used.
With today's devices, one- or two-layer designs are usually OK for embedded controllers with
clocks under 10 MHz. But even here, EMI precautions must be taken.
We'll address EMC design recommendations for one/two layer guidelines in a future post. For
now, we'll assume multi-layers. As a minimum, you can start with four layers: a ground plane, a
power plane, and two signal layers. The multi-layer concepts can be extended to as many layers
as you want or need (or can build.)
Multi-layer boards are preferred for high-frequency designs. This generally means RF (radio
frequency) circuits or digital circuits with clocks over 10 MHz. But even low-frequency analog
circuits (audio or instrumentation) can benefit from multi-layer designs when subjected to RF
susceptibility requirements. Remember, low-frequency circuits and be affected by high-frequency
threats.
Our experience has shown that multi-layer boards provide at least 10× reduction in radiated
emissions, and 10× improvement in immunity (both RF and ESD.) We have repeatedly seen that
when replacing simple two-layer boards with four-layer boards. The power and ground traces are
now solid planes -- everything else is the same. But as the old saying goes, your mileage may
vary.
The first EMC miracle occurs due to proximity of the planes to signal traces. The image-plane
effect provides a return path for high-frequency currents that greatly reduces loop size. Every
trace is now a transmission line, instead of an unwanted loop antenna. This works as long as the
adjacent plane is continuous all along the trace.
The second EMC miracle occurs due to reducing power/ground loops and impedances. These
loops form additional hidden antennas for emissions and immunity. We refer to these loops as the
"back door" for EMC. No, the clock Vcc CURRENT is NOT CONSTANT; rather, the clock Vcc
current pulses at the clock frequency as the internal loads change.
So what do we look for in the board stackup? Here are four simple features to examine:
(1) Are all trace layers adjacent to a solid plane? Either power or ground planes are fine, since a
well-decoupled power plane is just another high-frequency ground (return) plane.
(2) Are associated power and ground plane adjacent? For example, are the analog-voltage
planes next to analog ground, and digital-voltage planes next to digital ground? Are there
overlaps?
(3) If a plane is split (common for voltage planes), will traces run across these cuts? If so, you are
just begging for EMI problems. It is best to define design rules prior to routing, and then examine
the actual routing results. Additional split-plane issues will be addressed in a future post.
(4) Are the trace and solid planes symmetrical about the center of the board? While not an EMC
issue, this can affect board construction and is worth a quick look.
For simple circuit boards, the entire process of examining critical circuits and the board stackup
for EMC should not take more than a couple of hours. This time is well spent early in the design,
to prevent EMI disasters later!
[Additional Editor's Note: for a somewhat "whimsical" look at printed circuit boards, click here.]
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #3: Voltage regulators as critical circuits
EMC Basics #4: Analog devices as critical circuits
EMC Basics #5: I/O as critical circuits
EMC Basics #7: An introduction to troubleshooting EMI problemsDaryl Gerke, Kimmel Gerke Associates
6/27/2011 9:36 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries at the end of this item.]
Time to switch gears for a while, and look at troubleshooting
EMI problems. We'll examine four key EMI problems -- emissions, ESD (electrostatic discharge),
RFI (radio frequency interference), and power disturbances. We'll look at these problems in two
contexts -- the EMI test lab, and the engineering lab. We'll also discuss specific troubleshooting
techniques.
Troubleshooting consists of trying to isolate a problem and the underlying causes, and then
applying appropriate fixes. Often times, we are acting like a medical doctor to diagnose an EMI
illness.
Diagnosis is important -- don't just start throwing solutions at the problem.The medical profession
has a saying for this --"Prescription without diagnosis is malpractice." I think this applies to EMI
problems, too.
To continue with the medical analogy, doctors use a methodology known as differential diagnosis.
That means ruling things in, and ruling things out. In simple terms, diagnosis is often a process of
elimination. Or at least, a process of playing the odds.
Diagnosis involves several stages: looking at clues, examining the equipment, and perhaps
gathering additional information (usually through tests.)
The first step is to look at the clues. For example:
• What are the symptoms? Resets? Lockup? Bizarre readings?
• How bad is the problem? Small outage? Damage? Catching on fire?
• Is there an obvious cause and effect? In the test lab, this may be very obvious. In the field,
this may be unknown, so you may have to speculate.
• What are some key parameters? Frequencies? Amplitudes? Dimensions? Impedances?
The next step is to examine the equipment. For example:
• How does the electrical design look? Multilayer or two layer boards? Layout? Etc.?
• How does the mechanical design look? Metal enclosure or all plastic? Seams? Penetrations?
• What about cables and connectors? Shielded? Filters?
• And what about the power interface? Filters? Transient protection?
At this point, one should make a preliminary diagnosis. If the data is still fuzzy, you may need
additional testing. The tests can either be monitors,or failure forcers. Both can provide critical
information.
By the way, it is OK to change your diagnosis as you proceed -- doctors do this all the time. More
important, don't fall in love with your initial diagnosis, but keep an open mind as new data
becomes available.
Once comfortable with a diagnosis, you are finally ready to try fixes (prescriptions.) Install, test,
and observe. If nothing happens, try another fix. And so on. Keep notes as you go along so you
can backtrack.
By the way -- don't try only one fix at a time, but rather stack them up. To change analogies, EMI
problems are often like a leaky boat. If you have five holes in the boat, but you only apply one
patch one at a time, you'll never get dry.
A final admonition -- at this stage, don't worry about the practicality of your fixes. The initial goal is
to find a fix - any fix. Once you find that first fix, you can always try for a better one.
Over this next series, we'll examine various EMI problems. We'll look at the symptoms, and we'll
discuss various troubleshooting tests. We'll also include recommended fixes. Stay tuned...
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #3: Voltage regulators as critical circuits
EMC Basics #4: Analog devices as critical circuits
EMC Basics #5: I/O as critical circuits
EMC Basics #6: Looking at circuit board "stackup"
Also relevant to this topic:
Debugging: The 9 Indispensible Rules for Finding Even the Most Elusive Software and
Hardware Problems (Chapter 5, Part 3 of 3)(and see its preceding sections, which are linked
within)
EMC Basics #8: An introduction to troubleshooting EMI problems (con't)Daryl Gerke, Kimmel Gerke Associates
7/11/2011 2:40 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries at the end of this item.]
Unintended electromagnetic emissions can cause interference to nearby communications
receivers—radio, TV, GPS, WiFi, and more. Many years ago, EMI problems were called RFI,
or radio frequency interference.Marconi suffered EMI (RFI) problems over 100 years ago.
Mandatory emissions tests are required for most electronics devices.These include both radiated
emissions (RE) and conducted emissions(CE). The RE tests look at electromagnetic fields from
the entire system, while the CE tests look at voltages/currents on the input-power mains.
As an aside, although the CE tests are not radiated tests, they still aim to prevent direct radiation
from the power lines. The CE tests also aim to prevent interference from directly coupling through
the power system to other equipment.
There are two broad categories for emissions tests, with different goals:
•Commercial limits aim to protect a nearby TV receiver
•Military/avionic/vehicular limits aim to protect a nearby radio receiver
Since radio receivers are much more sensitive than television receivers, the latter limits are much
lower. For example, at 100 MHz the military/avionic limits are often 40 dB or more lower than
corresponding commercial limits. The exact differences vary depending on the actual
environments.
Even if you pass all the emissions test at the EMC lab, you can still have problems in the field. If
that happens, don't assume you can hide behind your FCC/CE certifications. In the US, the FCC
can invoke the "noninterference clause". You cause a problem, you clean it up.
There are two major components to emissions problems: hidden transmitters and hidden
antennas. Even though they are not really hidden, most designers don't see them as transmitters
and antennas (RF designers being an exception, of course.) But the electrons don't care—if it
looks like a transmitter and an antenna, it's time to party.
The primary hidden transmitters in most systems are digital circuits. Highly repetitive signals such
as clocks and clock-like signals (busses, repetitive control lines, etc.) generate strong harmonics.
But even switching power electronics (power supplies and motor drives) can get into the act. As a
helpful hint, we usually assume the first twenty harmonics of any repetitive signal are potential
transmitters.
But even lower frequency circuits can be hidden transmitters, thanks to parasitic oscillations.
Long a problem with vacuum tubes, we've seen a significant increase these unwanted oscillations
in solid state voltage regulators, op amps, and other more. As a helpful hint, these usually occur
above 100 MHz.
The hidden antennas are highly dependent on physical dimensions and frequency. The higher
the frequency and the longer the antenna, the more the radiation. We usually assume anything
over 1/20 wavelength is an efficient antenna. That means six inches at 100 MHz, and about 3/4
inch at 1 GHz.
As such, cables are very likely hidden antennas, followed by traces on circuit boards and even
the components themselves in the GHz range. Openings in shielded enclosures can act as "slot
antennas", and follow the same guidelines. A two-inch slot at 300 MHz leaks like a sieve.
So how do we troubleshoot these emissions problems? Here are five quick suggestions:
•Turn off clocks or change their frequencies to see if the emissions more or disappear. This can
isolate the hidden transmitters.
•Use sniffer probes to identify hidden transmitter circuits. These are small hand held magnetic
probes you connect to a spectrum analyzer. Since they are quite localized, you can quickly sniff
around a circuit board for hot spots for emissions.
•Remove cables to see if the emissions change. This can isolate the hidden antennas.
•Use current probes on the cables. These are high frequency probes you clamp on a cable and
connect to a spectrum analyzer. Currents in excess of a few microamps are suspect.
•Shield the entire enclosure with aluminum foil. This is very useful in identifying any hidden
antennas in the mechanical enclosure.
Next time, we'll share some troubleshooting techniques for ESD (electrostatic discharge.)
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #3: Voltage regulators as critical circuits
EMC Basics #4: Analog devices as critical circuits
EMC Basics #5: I/O as critical circuits
EMC Basics #6: Looking at circuit board "stackup"
EMC Basics #7: An introduction to troubleshooting EMI problems
EMC Basics #9: Troubleshooting RFI EMC problemsDaryl Gerke, Kimmel Gerke Associates
8/7/2011 3:51 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries at the end of this
article.]
RFI (radio frequency interference) is a rapidly increasing EMC problem, thanks
to the proliferation of wireless devices. These range from low power Wi-Fi and cell phone
transmitters to high power radio/television broadcast transmitter to extremely high powered radar
systems.
[As an aside, commercial testing has resulted in more robust products. Prior to 1996, when CE
testing became mandatory, we often saw RFI problems in the field. These were due to both
broadcast transmitters (radio and television) as well as mobile transmitters (cell phones, hand
held VHF/UHF transceivers, vehicular transmitters, and similar. Today, however, most RFI
problems get caught prior to market release. Some, however, still make it into the field.]
RFI as an EMI source
Transmitter power is not the sole issue, but rather a combination of power and proximity. The
lowly cell phone a few inches away may cause more problems than the broadcast transmitter a
mile away. The key parameter is the magnitude of the electromagnetic field at the victim.
In the EMI world, we focus on the electric field magnitude. The field levels can be easily
measured with suitable equipment. But here is a simple approximation that will get you in the right
ball park for an initial assessment:
E = [5.5 × √(PA)] / d,
where
E = Electric field in Volts/meter
P = Transmitter power in Watts
A = Antenna gain relative to isotropic
d = Distance from transmitter antenna to victim in meters
This formula assumes a point source and a "far field", both valid for most RFI situations.
For example, a 1W radio at 1 meter with a relative gain of 1 (good assumptions for a hand-held
radio or cell phone) produces a field level of 5.5 V/m. In fact, it is this hand-held transmitter model
that results in the 3 V/m and 10 V/m limits for commercial equipment. Higher field levels reflect
higher transmitter levels, such as the 200 V/m limit common for many military/automotive
environments.
RFI coupling paths
Since radio transmitters work by electromagnetic radiation, the primary path is radiated. As
"hidden antennas" are involved, physical dimensions are critical. For frequencies <300 MHz,
cables are the most likely factor. For frequencies >300 MHz, everybody gets in the act: circuit
board traces, enclosure openings (slot antennas), and even components themselves.
A good rule of thumb is to assume any conductor greater than 1/20 wavelength long is a potential
antenna. We've seen smaller antennas, but this criterion is widely used in the EMI community.
This means six inches (15 cm) at 100 MHz, two inches (5 cm) at 300 MHz, and ¾ inch (2 cm) at 1
GHz.
Direct conduction is also possible, but less likely. Nevertheless, when troubleshooting an RFI
problem, don't overlook this possibility. We've seen it happen.
RFI victims
The primary RFI failure mode is rectification, but failure levels vary with circuit types. The more
sensitive the circuit, the lower the thresholds will be. Here are some typical levels:
Analog circuits: 0.1 to 1 V/m
Power circuits: 1-10 V/m
Digital circuits: 10-100 V/m
These are simple guides. As the saying goes, "Your mileage may vary..."
Different circuits exhibit different symptoms, which may be helpful in troubleshooting:
When analog circuits are upset, you may get errors in sensor information, but everything else
works fine.
When power circuits are upset, the system often locks up or exhibits other strange behavior.
When digital circuits are upset, repeatable upsets are typical: resets, unwanted interrupts, or
memory upsets. The digital upsets are similar to those seen with ESD and other transients that
“flip” critical bits. .
Troubleshooting RFI problems
If you fail an RFI test at the lab, you already have details of frequency, amplitude, and failure
mode. In you fail in the field, the picture is less clear. You may need to make some
measurements, or to use approximations as described above.
Here are five quick RFI troubleshooting suggestions:
1. Remove cables to see if RF susceptibility changes. As an alternate, apply clamp on
ferrites for frequencies above 100 MHz.
2. Wrap the unit under test in aluminum foil. This will show if shielding is adequate (or will
help if Unit Under Test is unshielded.)
3. Harden critical circuits -- ferrites and 1000 pF capacitors are very helpful above 100 MHz.
Apply to inputs, power, and even outputs.
4. At the systems/box level, troubleshooting is best done in a shielded enclosure with
suitable equipment. In a pinch, a hand-held VHF/UHF radio can be useful. If testing in the
field, keep transmissions short (1-2 seconds) on unused frequencies.
5. At the PCB level, a signal generator connected to a sniffer probe can also be helpful in
"injecting" a signal at the component/trace level. Not new, this technique was used 50
years or more ago by those who repaired radios and televisions. It still works today.
Next time, we'll look at troubleshooting power disturbances.
Previous entries in the series
EMC Basics #1: Welcome!; and Clocks: critical circuits for EMC
EMC Basics #2: Resets as Critical Circuits
EMC Basics #3: Voltage regulators as critical circuits
EMC Basics #4: Analog devices as critical circuits
EMC Basics #5: I/O as critical circuits
EMC Basics #6: Looking at circuit board "stackup"
EMC Basics #7: An introduction to troubleshooting EMI problems
EMC Basics #8: An introduction to troubleshooting EMI problems (con't)
EMC Basics #10: EMC troubleshooting and power disturbancesDaryl Gerke, Kimmel Gerke Associates
9/8/2011 1:16 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries here.]
This is the final installment in the mini-series on
EMI/EMC troubleshooting. Often seen as an EMC stepchild, power disturbances are becoming
increasingly important.
At the design level, we’ve seen a significant increase in recent years with power disturbances.
These problems are driven by improved power components, which allow faster switching rates,
higher power levels, and unfortunately, increased EMI problems.
We’ve seen an increase in power disturbance problems at the systems level, too. The recent
2011 IEEE EMC symposium even held a full-day special session on EMC issues and the “Smart
Grid”. The session was well attended and promoted a lot of discussion. As one wag observed,
“megawatts are finally meeting gigahertz.”
But back to the design issues. Due to these problems, mandatory power-disturbance testing is
now required for EMC qualification on a wide range of products. The specific tests vary with
industry, platform, and even location. One size does not fit all when it comes to power
disturbances.
Power disturbance specifications
There are power-disturbance specifications for electronics used on AC mains (often varies with
country), military platforms, vehicles, telecommunications facilities, commercial aircraft, and more.
Most are unique to the environment, and are typically based on empirical data.
Some specifications are part of the general EMC requirements, while others may reside in
separate documents. In the latter case, the popular term is “Power Quality,” or PQ. While the
EMC requirements focus on transients, the PQ requirements usually address longer term power
perturbations like sags, over/under voltages, outages, and others.
EFT and surge
Two very popular commercial EMC requirements are the EFT (Electrical Fast Transient) and the
lightning surge. These are applied to the AC inputs. In the real world, these are two of the more
common causes of equipment malfunctions and damage. Other industries, such as military,
vehicular, and telecommunications, have similar requirements for both AC and DC inputs.
The EFT tests simulate arcing at contacts, which results in short bursts of very fast transients.
The EFT is described in ANSI/IEEE C62.41; the corresponding CE test requirement is EN61000-
4-4. The individual transients uses a 5 nsec rise time, which is pretty close to the nominal 1 nsec
for ESD. As such, upsets such as resets or other “bit-flipping” is common.
The surge tests simulate a lightning hit on the power mains. The surge is also described in
ANSI/IEEE C62.41; the corresponding CE test requirement is EN61000-4-5. These transients
(both voltage and current) are much slower but with much more energy than the EFT. As a result,
both upsets and damage are common.
Power quality
An excellent resource for PQ on the AC mains in North America is IEEE Std-1000, also known as
the “Emerald Book.” This guide is put out by the IEEE Power Engineering Society, and focuses
on wiring practices for computer equipment. As such, it is an excellent place to start. We have
even used this as the basis for developing internal power specifications.
There are unique PQ specifications for other industries, too. Sometimes these are separate
documents, and sometimes they are separate chapters in detailed equipment specifications.
Several of the European Norms address PQ concerns for European power mains as well.
Troubleshooting power disturbances
There are two methods when troubleshooting power disturbances—failure forcers and monitors.
You may need a combination of both to isolate and fix a problem.
To force failures, both the EFT and Surge tests are good starting points. Start at low levels and
work your way up. And remember, the surge can cause damage, so don’t do surge testing on a
valuable one-of-a-kind prototype.
For monitoring power at the equipment, power disturbance analyzers (PDAs) are very helpful. If
you don’t own one, these can be rented and left in place for a period of time. These devices will
check numerous various parameters, such as over/under voltages, outages, transients, and
more. They date/time stamp the events, and even capture the waveforms for later evaluation.
One caveat with a PDA: due to the bandwidth, they may miss EFT events, so you may want to
augment one with a storage oscilloscope. The bandwidth should be 100 MHz or higher.
Here are five quick power disturbance suggestions:1. Add a modular power filter at the input. If in a metal enclosure, be sure to locate directly at the power entrance, and
provide a low-impedance ground connection between filter and enclosure.2. Add transient protection at the input. Install both differential mode and common mode devices. For the surge, MOVs
are adequate. For the EFT, you will usually need faster silicon devices, with short connections.3. If EFT upsets occur, try adding a multi-turn common-mode ferrite (3-4 turns through the core) to the input power line.
Note that single-turn ferrites may not be adequate.4. If resets occur due to EFT, try adding 0.01 μF capacitors and multi-turn ferrites right at the reset circuit. Inputs are
particularly vulnerable, but be sure any reset/voltage monitor devices are also well decoupled at the chip.5. If lightning damage occurs, try to assess the failure path. Consider isolation transformers and transient protectors
designed to for the full lightning-surge levels.
At various times, we’ve done all of the above. Next time, we’ll start looking at some EMC
shielding problems and solutions.
EMC Basics #11: An introduction to EMC shieldingDaryl Gerke, Kimmel Gerke Associates
10/7/2011 12:33 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries here.]
Time to shift gears, and look at the mechanical side of EMC — or more correctly, the
electromechanical side of EMC. In this next mini-series, we'll look at various aspects of shielding,
and how it works. It is not enough to just hang metal — you need to understand what you are
doing, and why.
The primary purpose of a shield is to block electromagnetic radiation. This includes both
radiated emissions and radiated susceptibility. In fact, most of the time a shield behaves in a
reciprocal manner, and what works for one direction works equally well for the other. The
exceptions are subtle, and will be ignored for now.
Shields can be applied at different levels. The most common for electronic equipment is at the
"box" level, but shielding can also be employed at the component, board, or even systems
level. (In the latter case, think of shielded rooms.) In fact, multiple levels of shielding are quite
common. You don't need to depend on just one shield for all your EMI protection.
Shielding performance is traditionally defined as "shielding effectiveness" (SE). This is the ratio
of the field level before the shield is in place, divided by the field level after the shield is in place.
It is customary to express this parameter in deciBels. The number should always be zero (no
shielding) or positive. (If negative, you must be creating energy. Quick — patent it!)
As with many EMC issues, there is a lot of duality with shielding:
•Two modes - reflection and absorption
•Two design issues - materials and mechanical
•Two field concerns - near field and far field
•Two frequency concerns - low frequency and high frequency
•Two impedance concerns - low (magnetic) and high (electric)
With all these variables, it is no wonder one shield design does not work for all cases.
All of the above leads to thinking of shielding in three regimes: magnetic, electric, and
electromagnetic. Figure 1 is a curve from a military design handbook showing the SE of copper.
Notice the two modes and three regimes. Don't panic: we'll look at this in more detail to help
you decode the mysteries of shielding.
Figure 1: Typical shielding curves for copper
As you can see, there are a number of things to consider when designing an EMC shield, both
electrical and mechanical. If you need shielding, it is not enough to just throw it over the wall to
the mechanical engineers. You need to be involved in the design decisions, too.
We'll explore all of these topics in more detail in future posts. We'll also augment these with
practical design guidelines along with our favorite shielding "rules of thumb."
EMC Basics #12: Shielding materials solve electromagnetic-compatibility issuesDaryl Gerke, Kimmel Gerke Associates
11/20/2011 8:29 AM EST
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries here.]
The first shielding decision is usually about materials. Do we need thick steel or mu-metal, or
will a thin shield or even conductive paint suffice? Well, it depends...
To better understand this issue, we'll delve into basic shielding theory. Don't panic — we
won't be deriving Maxwell's famous equations or bogging you down in electromagnetic field
theory. Rather, we'll use a simple theory developed in the 1930's by Dr. Sergei Shelkunoff that
still serves us very well almost a century later.
If you look at the shielding diagram in the previous post, you can see that prior to
Shelkunoff, it must have been very difficult to figure out what was going on. Did shielding
increase or decrease with frequency? Was it linear or exponential? What was the rate of
change? The answer, common in the EMI world, was "It depends..."
Shelkunoff proposed a simple transmission line model for shielding — specifically a lossy
transmission line. This resulted in two major mechanisms, reflection (R) and absorption (A). He
also added a fudge factor for reflections through a thin shield that he dubbed B.
The resulting top-level equation was rather elegant:
SE (dB) = R(dB) + A (dB) + B
The mechanisms are illustrated in Figure 1.
Figure 1: Basic shielding mechanisms modeled as transmission-line effects.
Incidentally, since B is relatively small for most EMI issues, most of us just ignore it. (It can
be important, however, for very thin shields or at optical frequencies.) As such, we'll focus on
the two remaining shielding mechanisms, R and A.
Reflection: This is a surface mechanism, and is the result of the mismatch (transmission line
theory) of the "barrier impedance" and the "wave impedance." The former is simply the surface
impedance, given in "ohms/square", while the latter is the ratio of the magnitudes of the
electric and magnetic fields.
In free space, the wave impedance is 377 Ω, but at very low frequencies (such as 50/60 Hz)
the wave impedance may be drastically altered by the circuit impedance.
Since the barrier impedance for metals and metallic coating is often measured in milliohms,
you can see we have a huge mismatch at higher frequencies. As a result, reflection is the
primary mechanism for shielding at radio frequencies (RF) above 10 kHz. Even thin coatings like
conductive paints can provide 60-80 dB or more of shielding.
Absorption: This is a volume mechanism, and is the result of loss through the shield (lossy
transmission line theory), and is the result of "skin depths." The loss is exponential. One skin
depth results in 8.68 dB (one neper) of loss, two skin depths in 17.4 dB of loss, etc. Since we
usually don't need absorption for RF frequencies, any added absorption is a bonus.
At power frequencies with high currents (low-impedance fields), the reflection is minimal so
you need absorption. Furthermore, skin depths are hard to come by at low frequency. At 60 Hz,
you need at least 3-4 inches of aluminum to start to have even a small effect. So what to do?
Well, you can boost the skin depth by permeability. The improvement is proportional to the
square of the relative permeability. Thus, 0.1 inch/2.5mm of steel ( μr= 1000) gives the same
absorption as about 3 inches/75mm of aluminum. This is why we use steel (or other permeable
materials) for shields around power supplies or around devices that are sensitive to power-line
magnetic fields.
The bottom line: the two mechanisms often drive our choice of shielding materials. At RF
frequencies (above 10 kHz) thin conductive coatings are usually fine. For power frequencies
(50/60/400 Hz) with high currents, thick permeable materials are often required.
In the next post, we'll start to look at the mechanical issues of shielding, such as how seams
and other discontinuities affect RF shielding. These are the common limitations of RF shielding.
We'll also share some of our favorite "rules of thumb."
EMC Basics #13: EMC shielding--destroying a shield
Daryl Gerke, Kimmel Gerke Associates
1/22/2012 5:43 PM EST
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known expert Daryl Gerke of
Kimmel Gerke Associates. Note that there are links to all previous entries here.]
There are two ways to destroy a high-frequency shield: seams and penetrations. As discussed in
the previous article (EMC Basics #12), even thin conductive materials work well for frequencies
above about 10 kHz. Thus, the weak points are mechanical rather than materials.
Building a high-frequency shield is like building a wooden water tank. Once the planks are thick
enough, the leaks occur at seams, joints, penetrations, and even knotholes. And even a small
hole can be a problem : drill a ¼ inch hole in the bottom of the tank, and eventually all the water
leaks out.
Incidentally, the “planks” don’t need to be thick for high-frequency EMC shielding, as long as
they are conductive. Obviously, wood doesn't work for EMC, but aluminum foil is very effective,
along with surface treatments like conductive paint or plating. Remember, however, you still
may need the “thick conductive planks” for low-frequency magnetic field shielding (60 Hz and
harmonics).
Before we go further, we need to look at some simple physics.Although the water-tank analogy
is useful, it falls apart in several ways. Here are a couple of key points:
•For seams, the longest dimension is critical -- NOT the area. Unlike water, a six-inch seam will
leak the same as a six-inch hole under worst-case conditions. The difference is that the seam will
be highly polarized, while the hole will not. When designing an EMC enclosure, we need to be
pessimistic. After all, Murphy and his law will make sure that the worst case will occur.
•For penetrations, the depth of penetration is critical, NOT the hole size. If a wire, cable, or even
a pipe extends beyond the EMC shield and is NOT shorted to the shield, the extensions act like
antennas connected by a coaxial cable. For example, carrying wires or cables through a hole to a
connector on the circuit board can completely destroy a shield at high frequencies.
A good rule of thumb for seams and penetrations is the “1/20 wavelength rule.” Antenna
designers often use this guideline as a practical limit when making small antennas. You don’t
need a half wavelength (or even a quarter wavelength) to support electromagnetic radiation –
1/20 of a wavelength will still do a credible job.
As EMC designers, we’re trying to do just the opposite – that is, NOT design antennas into our
shields. Yet, that is what happens. The seams look like slot antennas, and the penetrations look
like monopole antennas. Both can radiate (leak) a surprising amount of energy at high
frequencies.
Most EMC engineers use a 1/20 of a wavelength as a starting point, but even that only provides
about 20 dB (10×) reduction. If you need 40 dB (100×) this reduces to 1/200 wavelength, and 60
dB (1000×) reduces to 1/2000 wavelength.
You can quickly calculate physical dimensions using this formula,based on the speed of light in
free space:
Frequency (MHz) × Wavelength (meters) = 300
•For example, at 100 MHz, a wavelength is 3 meters, so a 15-cm seam or penetration (about six
inches) provides 20 dB of shielding. If you need 40 dB, that reduces to 1.5 cm, and if you need 60
dB, that further reduces to 1.5 mm.
•It’s even worse at 1 GHz, where a wavelength is 30 cm. A seam or penetration of 1.5 cm (less
than an inch) only provides 20 dB of shielding. If you need 40 dB, that reduces to 1.5 mm, and 60
dB it is only 0.15 mm. No wonder we need EMI gaskets (or even welded seams) and bulkhead
connectors at the higher frequencies!
We’ll look at how to plug those leaks in a future post of this series. In the meantime, the
1/20th wavelength rule is a good place to start. After all, if you can’t get a 10× reduction, why
even bother with shielding in the first place?
EMC Basics #14: Making plastic coatings work for EMIBill Kimmel, PE, Kimmel Gerke Associates
3/1/2012 9:56 PM EST
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known experts Daryl Gerke and
Bill Kimmel of Kimmel Gerke Associates. There are links to all previous entries in the EMC
Basics series here.]
Two factors are combining to create EMI shielding problems: increasing clock frequencies and
the shift to plastic housings. The trend to higher clock frequencies continues. Modern laptop and
desktop computers run in the GHz range, a thousand times faster than the early personal
computers, but even modest applications are running clocks in the 20- to 100-MHz range. CISPR
22 calls for emission testing up to six GHz, depending on maximum clock frequency.
For shielding effectiveness, you need to keep openings in your enclosure less than 1/20
wavelength of the highest-applicable frequency, typically to the tenth harmonic of the fastest
clock. If you have a 100-MHz clock, you will be testing to 1 GHz, so expect to limit openings to
about 2 cm, and even less if you have higher clock frequencies. It has now reached the point
where continuous closure is necessary; occasional contact is not sufficient.
The second aspect is the increasing use of plastic enclosures. As we know, plastic provides no
shielding unless provided with a conductive coating. The conductivity of the coating is not the
driving factor in high-frequency shielding effectiveness; it's the longest dimension of the
opening, which almost always occurs at the mating seams. So select your coating for criteria
other than EMI, such as availability, cost, durability, and ease of application.
Herein lies the problem: it's difficult to get conductive closure at the seams, and the problem is
with the design of the mold. Done right, the shield works very well. Unfortunately, most of the
molded plastics are poorly designed to contain EMI. Radiated-emissions failure is almost a
foregone conclusion, if the plastic enclosure is not properly designed.
How to make the shield work
You need to bring the conductive coating right up to the mating seams, then ensure the surfaces
conductively mate pretty-much continuously along the entire seam. This requires the plastic
enclosure be designed so as to facilitate proper coating.
A reliable method is to use tongue and groove, making sure the mating surfaces are stiff enough
to ensure continuous contact. Better, yet, the groove can provide a "nesting" place for
conductive EMI gasketing, Figure 1.
Figure 1: a) Conductive coating brought up to mating surfaces;
b) EMI gasketing placed in groove.
However, we encounter strong resistance when discussing this with the mechanical designers.
They want to mask off the coating back from the seam to ensure the coating doesn't show on
the outside. The fact is, if the coating doesn't get into the seams and close the gaps, the
shield will not work.
In fact, if you don't close the gaps, the measured emissions may well increase! Why? The shield
may well collimate the available RF energy, resulting in emission hot spots.
You also need to make provision to terminate cable shields and filters. In most cases, these need
to be grounded firmly to the shield, as casual contact will not do. The worst case is to run the
cable shield without terminating it to the enclosure shield: that's a guaranteed test failure.
Summary
Don't fool yourself. If you want the shield to work, design the enclosure to close the seams.
Tongue and groove works well, preferably with EMI gasketing. Also make sure your cable shields
and cable filters are well terminated to the coating.
EMC Basics #15: Use gaskets to seal and solve leaky RF seam problemsDaryl Gerke, Kimmel Gerke Associates
5/9/2012 5:41 PM EDT
[Editor's note: we are pleased to continue our series on the vital and sometimes unappreciated
topic of electromagnetic compatibility (EMC), presented by well-known experts Daryl Gerke and
Bill Kimmel of Kimmel Gerke Associates. There are links to all previous entries in the EMC
Basics series here.]
Continuing with our shielding theme, we'll look at EMI gaskets. As discussed previously, seams
and other openings can be a "weak link" in EMI shielding. Two key concerns are gasket choice
and gasket mounting. We'll look at the former now, and save gasket mounting for a future
article.
Years ago, EMI gaskets were widely used in military systems and radio communications
equipment, but rarely seen in commercial equipment. Thanks to increasing processor speeds
and increasing EMI threats, gaskets are now common in a wide range of electronic equipment.
As a rule of thumb, we recommend gaskets whenever shielding needs exceed 60 dB, although
gaskets can still help at lower levels.
We regularly encounter EMI problems due to poor shielding and gasketing. While you might
be tempted to leave EMI gasket choices solely to the mechanical engineers, don't do that. You
need to work with your mechanical colleagues. Like many EMI problems, both disciplines need
to be involved. Fortunately, the solutions are usually simple once you understand some basic
principles.
How gaskets work
EMI gaskets perform their magic by providing a conductive path across seams and other
discontinuities in an electronic enclosure. This “shorts out” any potential difference across the
shield surface while maintaining smooth current flow. As a mechanical analogy, gaskets simply
plug the leaks.
In a perfect EMI shield (a “Faraday cage”), the EMI currents induced on the shield remain
inside (or outside) the shield. In the real world, however, seams or other joints present a
discontinuity. Shield currents are diverted, and a voltage appears across the seam or joint. As a
result, time-varying voltages and currents can launch an electromagnetic wave, just like a wire
antenna.
In fact, seams in shields are often modeled as “slot antennas.” The only difference between a
wire antenna and a slot antenna is that a wire antenna is metal surrounded by space, and a slot
antenna is space surrounded by metal. That means even a thin slot (such as two metal surfaces
separated by paint) can radiate if it is long enough.
A critical parameter is length, not thickness. Most of us in the EMI business worry when slots
are longer than 1/20 wavelength (e.g. 5 cm at 300 MHz, 1.5 cm at 1 GHz.) The secret to success
is to minimize impedance across the joint with clean, continuous metal-to-metal contact.
An alternate would be to reduce the current flowing in the shield, but this usually isn’t
practical. However, don’t overlook this. We once had a case where hundreds of amps of high-
frequency power-return currents were flowing in the cabinet. That case gave us two options to
explore -- either improve the gaskets, or reduce the currents -- we ended up choosing the latter
with good success.
Types of Gaskets
There are several types of popular EMI gaskets. All will work well when properly installed, so
the choice is often usually based on overall mechanical issues. Here are some pros and cons on
different EMI gaskets:
•Beryllium-Copper: These gaskets provide very-high EMI performance. The material has high
conductivity and is very springy, which makes it ideal for doors and panels. The material can be
formed into many shapes, such as fingerstock, serrations, and spirals, and can be plated for
corrosion protection. The drawbacks are cost, mechanical vulnerability (such as snagging of
fingerstock), and the lack of an environmental seal.
•Conductive elastomers: These gaskets provide good performance. They often have metallic
particles or wires embedded in them, so they do require pressure to assure an EMI seal. A big
advantage is that they can also provide an environmental seal as well as an RF seal. The main
drawback is the compression force, but hollow elastomer gaskets often overcome this issue.
•Wire mesh: Like fingerstock, these gaskets can provide very high levels of EMI performance. A
drawback is that many mesh gaskets take a set, and thus can not be reused. Those are fine for
permanent seals, but are not suitable for doors or access panels. They also lack an
environmental seal.
•Conductive cloth over foam: These gaskets are very popular in commercial applications, and are
quite cost effective. Most use a silver-plated cloth over foam to create a soft gasket that can
take up a lot of mechanical slack. The major drawback is a lack of an environmental seal.
•Conductive epoxies: these are permanent gaskets, formed from a metal impregnated caulk.
Silver loading is very common, and the seal is usually also watertight. The major drawback is that
any repaired joint must be cleaned and recaulked to maintain a seal.
•Form-in-place gaskets: These gaskets often make sense for high volume applications that can
automate the creation of a gasket right on an enclosure. These are similar to conductive
elastomers. When cured, these gaskets are resilient and thus may be reusable. The major
drawback is that they are not practical for low volumes.
Corrosion
No discussion of gaskets would be complete without a few comments on corrosion. Even a
small amount of corrosion can render a good gasket ineffective. Equipment used in harsh
environments, such as medical, military, vehicular, or industrial are often subject to corrosion.
Fortunately, corrosion is not usually a concern for commercial products.)
To fight corrosion, plated gaskets can be used to minimize the effects of dissimilar metals.
Another option is to seal out moisture at the gasket interface. Hybrid gaskets are available that
incorporate both an environmental seal and an EMI seal. In those cases, be sure to install the
environmental seal to the outside to protect the EMI gasket from external moisture.
In conclusion, consider gaskets in your shield designs, particularly if you need high levels of
shielding. And work with your mechanical colleagues to assure the best choice.