1. Analog Basics Workshop RFI/EMI Rejection
Rev 0.1

2. EMI or RFI? Both are sources of radio frequency (RF) disturbance
EMI – electromagnetic interference
Often a broadband RF source
RFI – radio frequency interference
Often a narrowband RF source
Terms are often used interchangeably
Electromagnetic Interference (EMI) and Radio Frequency Interference (RFI) are both electromagnetic interference. EMI is a more general term applicable to any electrical system interference but tends to be used to describe RF disturbances that are broadband in characteristic. While RFI is often associated with narrowband, radio transmission or reception based interference.
EMI is an undesirable byproduct of electrical systems and if it can be confined to the source or generator – all the better. But because EMI can produce a wide range of frequency spectra, it can affect otherwise properly operating circuits if they are sensitive to the particular frequencies generated by the source.
Even though an EMI source may be present in the vicinity of analog circuits their bandwidth or circuit design may be that they have little or no response to it. However, some analog circuits having surprisingly low bandwidth will respond to EMI in an undesirable manner. That is because individual components or subcircuits within them can have much higher bandwidths that see and respond to the EMI frequencies.

3. The necessary elements for EMI
Coupling medium
Receptor of
Electromagnetic
Energy 1
Three elements must be present for an EMI response from an analog circuit. First, there must be a source of EMI. Second, there must be a medium present that couples the electromagnetic energy to the circuit. And third, a sensitive receptor of the energy must be present. That may be the analog circuit or any other circuit.
All 3 elements must be simultaneously present for EMI to have an affect. The source must generate sufficient RF power at frequencies that the receptor is sensitive to. A coupling source or sources must be present to couple the RF to the receptor. And the receptor has to be sufficiently sensitive to that particular RF condition such that it produces an unintended response.
Even though EMI may start its life as radiated energy it is eventually converted to conducted EMI where it may be freely conducted into other circuits.
As often as EMI occurs in practice it doesn’t seem like it is all too difficult to create the right conditions. + _
Source of electromagnetic energy

4. Sources of electromagnetic energy
RF generating sources
Intentional radiators
cell phones
transmitters & transceivers
wireless routers, peripherals
Unintentional radiators
System clocks & oscillators
Processors & logic circuits
Switching power supplies
Switching amplifiers
Electromechanical devices
Electrical power line services
The source of electromagnetic energy may be a local, intended RF generator, but more likely it is a device in close proximity operating normally that produces RF energy as a byproduct.
Some sources may be more intuitive like clock circuits, oscillators and switching power supplies. Less obvious may be a normal operating, high-speed, digital IC. Or it may simply be an electromechanical device switching current and only produces EMI during make or break operations.
High-voltage electrical power system components can degrade and break down with time and provide a current path to ground. Arching connections often lead to corona discharge that interferes with sensitive radio receiving equipment.

5. Taming the EMI environment
Reduce receptor circuit’s susceptibility to EMI (Filtering)
Reduce the coupling medium’s effectiveness (Shielding)
Minimize EMI radiation from the source (Keep sensitive analog away from digital, soften digital edges)
It was mentioned that three elements are necessary for EMI to occur; an EMI source, a coupling medium and a sensitive receptor. Minimizing or limiting EMI from all sources is a good start. But once within compliance limits further reductions may become excessively difficult or costly to achieve. Designing a circuit and employing techniques that minimize EMI right from the start can prove to be the lower cost option. Having to revisit a non-compliant EMI source and applying remedies can prove difficult and expensive.
Reducing the effectiveness of the coupling medium is often addressed by careful layout, shielding and decoupling techniques. Depending on whether the EMI is common-mode or differential-mode the solution for reducing the coupling medium’s effectiveness will be quite different. This is a because of the difference in frequencies involved in the EMI.
Since sensitive analog circuits can be placed in almost any environment there is usually little way of knowing what EMI sources the circuit may encounter. A circuit might respond to a particular EMI source in one application and not to a different source in another. Therefore, one of the safest ways to avoid responding to EMI is to build effective EMI immunity into the circuit. That may be more easily said than done. But again, taking steps to reduce a circuits response susceptibility during the design phase is often the most prudent approach.

6. Analog receptors of electromagnetic energy
Op-amps
Low-speed: offset shift, RF noise
High-speed: linear and non-linear
amplification
Converters
EMI aliased into passband
corrupted output levels or codes
offset shifts
Receptors of EMI signals may be the sensors or transducers, the connecting wires or the analog circuits or the analog integrated circuits themselves. It depends on the their location relative to the EMI source and their susceptibility.
The most common response of analog circuits to EMI is signal rectification. Often the overall bandwidth of the analog circuit may be far less than that of the EMI source, but individual stages within the devices can respond due to their much higher bandwidth.
Op-amps usually exhibit a different slewing rate for positive and negative going signal transitions. When the EMI is rectified a voltage offset results at the output that is a function of the difference between the positive and negative slewing rates. This response is most often exhibited by low-speed op-amps.
High-speed op-amps may have sufficient bandwidth to amplify the frequency components of the EMI and sum them with the intended signal. If the EMI signal level is excessive non-linear amplification can result adding noise and distortion to the intended signal.
Depending on the converter architecture and speed, the response to EMI can range from an output offset shift, to the EMI being aliased into the signal pass-band. When aliasing occurs these unintended signals result in distortion products and increased noise.
Regulators, both linear and switching, may produce an EMI-related voltage offset resulting in an incorrect regulator voltage.
Regulators
Offset - output voltage error

7. Operational amplifier voltage-offset shift resulting from conducted RF EMI
The OPA376 is a precision low-noise, low power, CMOS operational-amplifier having a unity gain bandwidth of 5.5MHz. For this test it was powered with a +/-2.5V supply and configured as a unity gain buffer. The input was driven with a precision RF generator from 10MHz to 6GHz and the dc input offset captured and recorded from the amplifier output as the frequency range was swept. The frequency range was selected to exceed the unity gain bandwidth of the amplifier and cross through regions where RF EMI is most likely to generated by common RF sources. The input amplitude was stepped in 10dB increments from -10dBm to +10dBm.
The offset-voltage change is the result of rectification by junctions internal to the operational amplifier as the RF makes its way through the amplifier. As one might expect the voltage-offset shift is most pronounced at the highest RF input power levels. These curves are representative of the OPA376 performance, but it is important to keep in mind that every different operational amplifier and linear analog integrated circuit product will have its own unique dc offset vs. EMI frequency signature.
in a 50Ω system -10dBm = 100mVpk 0dBm = 318mVpk +10dBm = 1.0Vpk

8. Radiated EMI and its affect on an ECG EVM
(Vin ≈ 1mVp-p G = 2500V/V)
Transmitter
470MHz
Pout 0.5W
d ≈1.5 ft (46cm)
Significant DC Offset
when RF present
+4.0V offset RF present
RF noise
On ECG
Single Supply
CMOS
INA326
OPA335(s)
1.5V
Due to RFI
Transmitter
keyed 6 sec.
+2.5V offset normal
This oscilloscope diagram depicts the affect of a strong EMI source in close proximity to a sensitive analog circuit. A 0.5W output, UHF transceiver is keyed with the antenna approximately 1.5 feet (45.7cm) from the ECG simulator. An ECG signal is very small, about 1mVp-p in amplitude, and requires a very high signal chain gain to bring the signal to usable observation levels.
When the transceiver is keyed, 2 effects are observed. The DC offset increases about 1.5mV and RF-related noise is evident on the ECG waveform. The basic ECG repetition rate stays the same, but the offset shift would be evident to both the human eye and any digital processing.
Fly wire Proto board
ECG Full Scale 1Vp-p 0.5V/div
EMI slide
Information
by John Brown

9. Input RC filtering as applied to an instrumentation amplifier
An EMI/noise filter is commonly applied at the input of an instrumentation amplifier. Like the line filter in the previous example this filter provides attenuation of both differential (normal) and common-mode EMI signals.
This is a passive RC filter exhibiting a first-order roll-off (-20dB/dec). However, the -3dB cutoff frequency is different for the two modes. The differential cutoff frequency must be set such that the desired signals are not attenuated. For example if the maximum differential signal frequency is less than 100Hz, a -3dB cutoff of about 350Hz will be required to avoid excessive attenuation at 100Hz. The resistor values should be kept under 10kΩ, and preferably less, to keep the resistor noise contribution low. The smaller the resistor value is made the larger the filter capacitor values become for a given cutoff frequency. So a compromise is called for. In this example RA and RB are set 4.7kΩ, resulting in a CA capacitance value of 47nF.
The usual practice is to set each common-mode capacitor, CB and CC, to 1/10th that of the differential-mode capacitor, or 4.7nF in this example. The reason for doing so is to reduce the filter’s ability to convert the common-mode signal to a differential signal due to CB and CC mismatch. Using the 10:1 differential to common-mode capacitor ratio provides a 20x reduction in CMR errors to due common-mode capacitor mismatch.
The common-mode filter cutoff frequency is set by the selected CB or CC capacitance value and RA or RB. Here, the common-mode cutoff frequency is about 21 times higher than that of the differential-mode filter. Closer match between the common-mode and differential mode cutoff frequencies is often desired and requires better matched filter components.
Differential Mode
f-3dB = [2π(RA+ RB)(CA+ CB/2)]-1
let RB = RA and CC = CB
f-3dB = 343Hz
Common Mode
f-3dB = [2π∙RA∙ CB)]-1
let RB = RA and CC = CB
f-3dB = 7.2kHz

10. Newer op-amps have built-in EMI filtering
Reducing an operational-amplifier’s EMI susceptibility now may include an integrated input filter. A low-pass filter is created by using the input differential pair junction capacitances, and a small series resistances in the input paths. Since the differential pair exhibits both differential and common-mode capacitances the filter is effective for filtering both types of EMI.
An equivalent filter is shown having total differential and common-mode capacitances of 2.5pF and 5.0pF, respectively, and 1kΩ series input resistors. The filter response plot reveals a first-order, low-pass response, and a -3dB cutoff frequency of 32MHz. Both differential and common-mode cutoff frequencies are the same for the capacitance and resistance values used.
The actual unity-gain bandwidth of the operational-amplifier is commonly be a few Megahertz, so the low-pass filter cutoff frequency is well above it. That prevents the EMI filter from becoming a factor in the operational-amplifier’s normal AC response. However, the cutoff frequency is low enough to be effective in filtering of 100MHz and higher.
The series resistances are selected to set the filter’s cutoff frequency, yet must be keep low in value so as not to degrade the amplifier’s low noise performance.

11. EMIRR- a measure quantifying an operational amplifier’s ability to reject EMI
EMIRR - electromagnetic interference rejection ratio
Defined in National Semiconductor’s application note AN-1698
Measured as a dB voltage ratio of output offset voltage change in response to an injected RF voltage having a defined level
Provides a definitive measure of EMI rejection across frequency allowing for a direct comparison of the EMI susceptibility of different operational amplifiers + -
ΔVOS (DC)
VRF
Electro-Magnetic Interference Rejection Ratio (EMIRR), is simply a voltage ratio that compares an ac input to a dc output. The ac input represents an EMI or RFI signal that is coupled to any of the op amp terminals. The dc output voltage is the change in the output offset.
The change in amplifier’s dc offset voltage results from EMI/RFI signals being rectified by internal diodes formed by silicon pn junctions.
All radiated EMI entering a circuit becomes conducted eventually. If an operational amplifier or other linear analog circuit is in the EMI’s path it may be enter the device via the PC board traces. PCB traces will be tend to be much longer then the device dimensions and may act as effective EMI antennas at the frequencies of concern in the MHz to GHz range.

12. The EMIRR IN+ test set-up See TI Application Report SBOA128 for details
Simple schematic
for EMIRR IN+ test
Practical
implementation
Measurement of EMIRR IN+ is straightforward in concept, but quite involved in reality. In its most basic form a precision, wideband RF generator is programmed for a specific output level and to sweep over a wide frequency range. The frequency range has been selected to start at 10MHz which above the unity gain-bandwidth frequency of many precision operational amplifiers. It then is swept to 6GHz passing through many frequencies utilized by different wireless and communications devices. The dc voltage offset of the operational amplifier is measured with a precision DMM at set frequency points throughout the sweep range. A low-pass filter is used at the amplifier output to prevent any coupled RF from entering the DMM. Many DMMs become inaccurate if RF accompanies the dc voltage level.
Much of the difficulty attaining accurate results stems from the ever changing complex input impedance environment encountered from 10MHz to 6GHz. The voltage level at the operational amplifier non-inverting input must be held as constant as possible and be accurately known at all frequencies. The operational amplifier input impedance (Zin) changes greatly with frequency and that variability must be determined. One might be tempted to simply terminate the amplifier with a 50Ω resistor to ground, but the complex input impedance shunts the resistance resulting in a different complex impedance value.
Although the RF generator is programmed for a specific output level the reflections on the input line result in a different input level at each frequency. The solution is to not terminate the amplifier input and instead use a network analyzer to characterize the entire input path impedance in terms of their scattering parameters across frequency. The input impedance information obtained at each frequency is then used to correct and adjust the generator output such that a consistent volatge appears at the operational amplifier input.
The slide images show the practical implementation of the measurement. A bias-T is included such that the operational amplifier may be biased to a specific dc input level. The lower illustration depicts the complex input path and the various transmission line segments along the way to the amplifier input Zin. See TI Application Report SBOA128 for more information.
Zin of
Op-amp
The complex RF input
environment

13. EMIRR IN+ equation solved for |∆VOS|
Use this equation to solve for |∆VOS| of a unity gain amplifier if VRF_PEAK and EMIRR IN+ are known such as when a plot is provided
EMIRR IN+ is frequency dependant
Doubling VRF_PEAK Quadruples |∆VOS|!
For example: Consider a 100mVP RF signal at 1.8GHz applied to a device with an EMIRR IN+ of 60 dB. The associated voltage offset shift would be 100uV
The |∆VOS| is solved in terms of VRF_PEAK and the EMIRR_IN+. The quadratic relationship is becomes evident in this form of the equation.
This equation may be used to determined the offset shift required a RF signal of known amplitude applied to the input of an op amp and the established EMIRR_IN+ as provided. Note that because of the quadratic relationship, doubling VRF_PEAK quadruples |∆VOS|.

14. Higher EMIRR IN+ means lower amplifier EMI sensitivity
EMIRR IN+ equation
VRF_PEAK = peak amplitude of the applied RF op-amp input
ΔVOS = resulting “input-referred” DC offset voltage op-amp output
100mVP = standard EMIRR input level (-10 dBm)
Higher EMIRR IN+ means lower amplifier EMI sensitivity
Here is the equation for EMIRR IN+. This dB calculation is made based on the dc voltage offset shift (ΔVos) in response to the level of ac RF amplitude (VRF PEAK) applied to the non-inverting input.
The EMIRR IN+ equation has two terms on the right side of the equation. The first term is the standard linear ratio expression. As |∆VOS| gets smaller the EMIRR IN+ increases, corresponding to improved EMI “rejection”. The second term is the quadratic term. This term takes into account the quadratic relationship between VRF_PEAK and |∆VOS|.

15. EMIRR IN+ measurement results for TI CMOS rail-to-rail operational amplifiers
Larger EMIRR is better
This is an EMIRR IN+ plot for several Texas Instruments CMOS, rail-to-rail operational amplifiers. The difference between the highest and lowest EMIRR IN+ performance is dramatic, but some explanation will help explain why the results vary to this extent.
First, the OPA348 which has the least EMIRR IN+ is an older operational amplifier and was designed before the input filters were developed. Any rejection it and other operational amplifiers not having an intentional input filter provide is the result of inherent filtering and/or non-response to the RF. The fact that the EMIRR IN+ increases beyond 1GHz infers input filtering. Likely, the input differential input pair junction capacitances are acting in conjunction with the inherent trace and junction resistances to create the filter pole.
The OPA376 provides increasingly higher rejection beyond 100MHz. It has a built-in EMI filter, but because of its 5.5MHz gain-bandwidth the filter pole frequency has to be set high in frequency such that it doesn’t interfere with the amplifier’s bandwidth.
Similarly, the OPA378 and OPA333 show even higher rejection at lower frequencies. They have lower gain-bandwidth respectively allowing the filter pole to be set lower in frequency. The earlier the filter begins to attenuate the input EMI signal the greater the rejection that can be had.
Model GBW Filter Model GBW Filter
OPA333/ kHz Yes OPA376/ MHz Yes
OPA kHz Yes OPA348/2348 1MHz No

16. EMIRR testing applied to instrumentation amplifiers
IA under test
Differential mode EMIRR
Common-mode EMIRR
Test Configuration
Bipolar supplies (+/-V), reference pin grounded, RF level -10dBm
EMIRR testing has been applied to instrumentation amplifiers providing a means for assessing both differential and common-mode EMI responses. The two circuit schematics show typical electrical arrangements for the instrumentation amplifiers during EMIRR testing. The RF environment and concerns described for the operational amplifier apply here as well.
An instrumentation amplifiers gain (V/V) set is most often designed to be accomplished using a single external resistor. Any gain within the amplifiers specified operating may be set. A common range is 1 to 1000V/V. Both the differential and common-mode EMIRR can be evaluated at the selected gain. Note that the gain will affect the dc voltage offset level measured at the output. The unpredictability of the amplifier’s behavior in the presence of the strong RF input assures that the input offset value derived will not correspond directly to the gain setting.
Differential measurement
RF signal applied to non-inverting input
Inverting input grounded
Common-mode Measurement
RF signal applied to both inputs

17. EMIRR testing applied to instrumentation amplifiers INA118 – INA333 differential mode comparison
The differential EMIRR is compared for two, 3 op-amp instrumentation amplifiers. The INA118 is a low-power, bipolar device designed long before EMI avoidance was being incorporated in precision linear integrated circuits. Its EMIRR performance is plotted against a more modern CMOS, auto-zero amplifier, the INA333, which incorporates built-in EMI filters at each input. The improvement in EMIRR is readily evident from the plots.
INA118
3 op-amp current feedback design
Av range 1 to 10kV/V
70kHz BW, G = 10V/V
Iq 350uA
circa 1994
no internal EMI filter
INA333
3 op-amp CMOS auto-zero design
Av range 1 to 1kV/V
35kHz BW, G = 10V/V
Iq 50uA
2008 introduction
internal EMI filter

18. EMIRR testing applied to instrumentation amplifiers INA118 – INA333 common-mode comparison
This plot compares the common-mode EMIRR for the INA118 and INA333. Again, the effectiveness of the INA333 input filtering is revealed.
INA118
3 op-amp current feedback design
Av range 1 to 10kV/V
70kHz BW, G = 10V/V
Iq 350uA
1994 introduction
no internal EMI filter
INA333
3 op-amp CMOS auto-zero design
Av range 1 to 1kV/V
35kHz BW, G = 10V/V
Iq 50uA
2008 introduction
internal EMI filter

19. EMI/RFI Lab
Simulation
Calculation
Measurement

20. Ex 6.1: Hand Calculations ∆vos211 / ∆vos333 OPA211 EMIRR OPA188 EMIRR
1. The figures below illustrate the EMIRR for two different op-amps. Assume the same magnitude and frequency (476MHz) of RF signal is applied to the circuit below. H
OPA211 EMIRR
OPA188 EMIRR
∆vos211 / ∆vos333

21. Ex 6.1: EMIRR (Noise) Schematic
Two copies of the same two stage amplifier is on the board. Each two stage amplifier has four jumpers to configure the circuit.

22. Ex 6.1: Amplifier I/O PCB Setup
U0 = OPA2211
U1 = OPA2188
Connect antenna to JMP5 & JMP6.
Jumper
Position
JMP7, JMP8
DC — Right position for DC coupling
JMP1, JMP3
FLT — Top position for filtering.
JMP2, JMP4
GND — Right position for GND connection to input.
JMP5, JMP6
GND — Connect “antenna” to top position. This antenna is from the amplifiers noninverting input to GND.

23. Ex 6.1: Instrument Setup
The instrument setup above will configure the signal source and scope for the circuit below so that we can see the bandwidth limitations. Use the curser to determine the bandwidth (-3dB).

24. Ex 6.1: Expected Results Transceiver Keyed OPA211 output offset 2V/div
20mV/div
1. Does the relative change in offset match the theoretical EMIRR plots from the hand calculations?
Answer