Intermodulation Fundamentals White Paper

Innovative Communication Solutions
whitepaper
INTERMODULATION FUNDAMENTALS
1.0 Definition
Intermodulation (IM) or intermodulation distortion (IMD) is a frequency conversion process that occurs
when two or more signals pass through a non-linear system or device(s)/component(s) within a system.
The essential result of the process is that energy contained in the input signal of a non-linear system is
transformed at its output into a set of frequency components at the original frequencies plus additional
components at new frequencies that were not contained in the input. The IM phenomenon is often
referred to as mixing.
For example, consider a signal composed of two fundamental tones f1 and f2 that could represent two
transmitter signals co-located at a communications site. If this composite signal is passed through
a non-linear device (of third-order), the most general form of the output signal will contain frequency
components at dc, f1, f2, second-order products and harmonics as well as the third-order products
at 2f1 - f2, 2f2 - f1. These last products are often troublesome because they fall closest to the original
tones at f1 and f2. It is possible that the newly generated third-order products could fall close to or
within the receive band of a communication system located at the same site, which could degrade the
performance of the receiver.
As another example, consider the same two tones at f1 and f2 passing through a stronger non-linear
device of fifth-order. The set of most potentially troublesome IM products that can be produced by a
fifth-order non-linear system would fall at the frequencies
2f1 – f2, 2f2 – f1, third-order products,
3f1 – 2f2, 3f2 – 2f1, fifth-order products.
Note that the order of the non-linearity is determined by the sum of the coefficients. If the non-linearity
were stronger still (such as a seventh-order), it would have an output containing the following most
potential interfering carriers
2f1 - f2, 2f2 - f1, third-order products,
3f1 - 2f2, 3f2 - 2f1, fifth-order products,
4f1 - 3f2, 4f2 - 3f1, seventh-order products.
With respect to the original tones at f1 and f2, the third-order components are closest, the fifth-order are
the next closest and the seventh-order are furthest removed but still ‘ close’ to f1 and f2. This pattern
continues for devices of increasing non-linear severity.
When more than two tones of sufficient strength are present at a site, the generated IM products will
consist of the set of tones occurring at all linear combinations of the original tones (up to the order of
the non-linearity). Some of these IM tones will be potentially threatening to system performance, with
the exact threat being dependent upon the particular frequencies and bandwidths of the receivers
present at the site.
A more detailed mathematical analysis showing the spectrum widening effects of IM appears in Section
3.0 of this white paper.
United States Canada Asia Pacific EMEA Latin America © Copyright 2017
1 800 236 3275 1 800 236 3275 +1 905 726 7687 +44 1487 842819 +1 905 726 7676 Sinclair Technologies
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
whitepaper
2.0 Practical IM Discussion
In the context of transmit and receive components of wireless communication systems, there are
basically two classes of IMD that are commonly discussed. The first is Active Intermodulation and the
second class is Passive Intermodulation (PIM).
2.1 Active IM
Active Intermodulation is associated with active devices in a system. This can occur, for example, in
active devices typically contained in the output stage of transmitters, any portion of an inline amplifier or
the input stage of receivers.
In the case of a transmitter, active IM can occur if a signal from another transmitter enters the final
stage of the transmitter under consideration. Since the finals are frequently biased as class-C, they are
inherently non-linear and are prone to IMD if the interfering tone(s) are strong enough. A rule of thumb
is to assume an IM conversion coefficient of about 10 dB for solid state transmitters.
If there is a choice at the initial stages of system design, it is wise to choose a set of transmit and
receive frequencies that minimize the potential of generating IM products that fall in or close to
the receive bands of interest. When choosing a frequency plan, a software tool in the form of an
intermodulation analysis program should be run. After entering a set of potential transmit and receive
frequencies, the code will determine the potential generated IM frequencies and given a representative
receiver bandwidth can determine if there are any IM ‘hits’ on any of the chosen receive pass bands. It
is important to consider the effects of IM products at least up to the seventh order. In some cases, the
frequency set can be juggled to minimize potential IM effects.
In the design of systems where a number of transmitters are co-located, isolation between transmitters
can be achieved by placing each on its own dedicated antenna. Isolation is achieved by virtue of
physical separation between antennas - separation in the vertical direction is much more effective than
horizontal separation. Since the expense in terms of antennas, feed cables and especially tower space
is at a premium, it is typically impractical to place each transmitter on its own antenna, so transmitter
combining systems are employed to enable a number of transmitters to share a common antenna(s).
In transmitter combining systems where the transmitters are literally ‘wired together”, it is imperative to
provide sufficient isolation between the transmitter output stages using in-line devices such as isolators,
cavity filters, attenuators, etc. A common practice (and requirement) at frequency congested sites is to
have a minimum of a one or two-stage isolator and one or more cavity filters for each individual transmit
line.
Isolators provide signal isolation in one direction and serve to block other energy from entering the
transmitter through its output line. They also furnish the transmitter a constant VSWR even if the match
conditions at the antenna change. Cavity filters provide isolation to energy at frequencies outside of its
pass band - this inherently suppresses the second harmonic of the connected transmitter.
2
whitepaper
It is also possible for transmitter energy to be radiated from its associated antenna and be directly
picked up somewhere in another transmitter(s). If the undesired signal is strong enough, this can again
cause IM in the power amplifier and a portion of it will be radiated out of the antenna system of the
offended transmitter(s). The IM energy can be picked up at the same site or radiated out to areas far
removed from the original IM source(s).
In receivers, offending IM usually occurs when multiple signals of sufficient strength overload the first
r.f. amplifier stage or the first mixer stage. Typical receiver designs employ narrow band pre-selectors
to limit the band of signals entering the r.f. amplifier. The pre-selection (band pass function) provides
immunity to signals out of the desired receive band. Examples of such are transmit carrier signals,
out of receive band transmitter noise and generated IM signals removed from the receive band. If the
offending IM or other interfering signal energy falls close enough to the desired receive frequency, it
will pass through the pre-selector and enter the first stage of the r.f. amp.
To test if the receiver front end is the cause of IM, an attenuator can be placed in front of the receiver
and the IM product energy observed. If the IM products drop by the value of the attenuator, the
offending energy not caused by the receiver and other sources must be identified and dealt with. If
the IM products drop by three or more times the value of the attenuator, then the receiver front end is
generating IM and the system must be appropriately modified to drop the levels of the multiple tones
entering the receiver.
Sinclair Systems Engineers are very experienced in dealing with problems of transmitter and receiver
generated IM. The Sinclair Systems Engineering staff can run IM studies on a given frequency plan
and use the results to appropriately design a system to minimize the potential of IM.
2.2 Passive IM
The other form of IM that has received much attention in recent years is that of Passive IM, also known
as PIM. This is the production of new signal frequency components in a wide array of passive devices
commonly found on and around radio sites.
In transmitter and transmit/receive combining systems, all components in the lines that contain multiple
tones are potential sources of PIM. Assuming that the transmitters and receivers are not IM sources,
all devices that carry the combined signals of different frequencies must be scrutinized. A partial list of
common devices in combining systems that have the potential to produce PIM is:
• Cables
• Connectors
• Adapters
• Transmission line junctions such as tees and stars
• Filters
• Directional couplers
3
whitepaper
• Antennas
• Antenna feed lines
• Surge protectors.
These sources responsible for IM can often be deduced by systematically searching the combining
system assemblies and components until the problem is isolated.
Other less obvious potential sources of PIM at a typical site can be:
• Any mounting brackets for antennas, feed lines, etc.
• Rusty or loose fasteners such as bolts, nuts and washers on the tower.
• Tower member joints.
• Tower guy wires.
• Nearby fences, gates and metal signs.
• Hazard/warning lights atop the tower (these can also produce active IM).
These IM sources are much harder to trace. After the combining system is thoroughly probed,
sometimes IM sources can be found by ‘sniffing’ in various directions around the site using a highly
directional antenna and a spectrum analyzer.
There are two general principles by which PIM is generated in passive devices – magnetic effects and
joint/contact effects. These general mechanisms play a strong role in PIM generation where significant
multi-tone r.f. currents are present.
Magnetic effects cause IM by virtue of hysteresis in magnetic or paramagnetic materials at locations

carrying significant multi-tone, high frequency currents.
To combat magnetically generated PIM, avoid the use of ferrous materials in base materials and
platings. (In some cases, magnetic material could be used for a base material if good plating is used
and the plating is not damaged in any way during component or cable assembly.) Avoid the use of
nickel as either a base material or outer/under platings.
Joint/contact effects result when metal oxide layers develop in between poor metal-metal contacts.
These oxide junctions behave in a similar manner to a semi-conductor diode in terms of having a non-
linear transfer characteristic. They are important in locations of significant currents.
To minimize PIM caused by poor contacts and joints, the total number of joints should be minimized. It is
best for all metal joints to be welded, brazed, or soldered. If these are used they must be implemented
correctly using non-corrosive fluxes.
4
whitepaper
In cases where welding, brazing, or soldering is not possible, all contact points should be scrubbed
clean and high contact pressure must be evenly maintained. Maximum contact area should be
maintained between all mating surfaces that are not welded/brazed/soldered.
All cables in a combining system that are required to pass multiple tones at high signal levels should
meet the following conditions:
• Minimum use of braids for inner and outer conductor of transmission lines- solid conductors are best.
• If braids/strands are present, minimize cable flexing when in use.
• Soldered braids/strands are better than not soldered.
• Whenever possible avoid nickel anywhere in the cable material system. Sometimes nickel is ‘hidden’
as an underplating on braids and solid conductors.
• Minimum use of ferrous materials in outer/inner conductors - avoid if possible. If ferrous material is
present it must be properly plated and the plating must not be compromised. It is best to avoid ferrous
materials altogether.
Choice of connectors and adapters in a system is an important aspect of low-PIM construction. The
total number of connectors and adapters should be minimized. Also important is that the total part
count of connector sub-assemblies should be minimized.
Connectors will realize the best contact/lowest PIM generation when they are soldered in place. ln
cases when soldering isn’t possible, scrubbed contacts, high mating force and maximum contact area
should be maintained.
As in the case of cables and other components, no Nickel outer or under plating should be used.
Silver and Gold are good plating materials - Gold also does not form an oxide layer. All material
surfaces to be plated must be correctly prepared both mechanically and chemically. Plating must be
implemented properly with minimum contamination. Avoid plating materials with excessive ‘brighteners’
- these can significantly decrease conductivity.
It is important to minimize the number of times that connectors are assembled and disassembled.
Connectors that are designed to be accurately torqued are best for maintaining low PIM generation
over long term. At critical locations in a system, there has been a strong trend in recent years to
replace Type-N connectors with 7/16 DIN, as the design of the DIN type is inherently low PIM
generating.
Joints between cables and connectors should be of low PIM character. When possible, minimize the
use of dissimilar metals at contact points. Soldering is best - when one cannot solder, proper clamping
is next best. Clamping, when implemented properly, is much better than crimping. Crimped joints can
loosen over time, develop oxide layers and cause PIM.
5
whitepaper
The above outlined principles and techniques are used in the construction of all low PIM components in
transmitter/receiver combining systems as well as in the antenna related portions of the systems. It is
important to maintain low PIM generating conditions at all points where there are multiple carriers and
where significant multi-carrier r.f. currents are present.
Sinclair has many years of experience in designing components and systems that minimize PIM. To
test low PIM designs, Sinclair has a number of IM measuring systems in-house that are used to
characterize PIM behavior.
3.0 Non-linear model and simple amplitude analysis

We begin the discussion with the choice of a simple non- linear device model and then proceed to
illustrate the important effects of IM by carrying out a simple analysis of a two-tone signal passing
through a non-linear system.
In common electrical engineering terminology, the signals may be any of the related quantities of
voltage, current, fields, etc. For the purposes of our analysis, we will choose to use the voltage V = V(t).
To help shorten the expressions, we will suppress the parenthesis expressing the time dependence.
One of the simplest ways to understand the phenomenon is to consider two pure, single frequency
sinusoidal tones and pass the sum of the two through a non-linear device. After carrying out the
necessary manipulations, we will make observations of the output amplitudes in the frequency domain.
The first step is to define the type of non-linearity describing the system or device. There are an infinite
number of non-linear transfer functions that could be considered but it is common to use one of the
simplest models applicable to mild non-linear behavior.We will consider the following non-linear
transfer function:
Vout = K1 Vin + K2 (Vin)2 + K3 (Vin)3 + … (1)
In this power series representation, the strength of the non-linearity depends upon the values of
the complex coefficients K1, K2, K3, .... For mild non-linear behavior, we will assume that the series
representation is convergent for values of voltages Vin under consideration.
To illustrate the frequency domain distortion effects, we will start by considering only the first few terms
of the series. (This is a valid approximation if the series is convergent because the neglected terms will
introduce only a small error compared to the exact representation of Vout)
Neglecting the terms beyond the cubic, we get
Vout ~ K1 Vin + K2 (Vin)2 + K3 (Vin)3 . (2)
Eq. (2) allows easy calculation of Vout to show the intermodulation effects in the frequency domain.
Consider now the pure two-tone input voltage
Vin = A cos ω1t + B cosω2t, (3)
6
whitepaper
where ω1 ≠ ω2 and it is understood that ω = 2πf
Inserting (3) into (2), expanding and collecting terms, we obtain the resulting equation (4)
Vout ~ ½K2 (A2 + B2) DC term
+ (K1A + 3/2 K3AB2 + ¾ K3A3) cos ω1t Fundamental terms
+ (K1A + 3/2 K3AB2 + ¾ K3A3) cos ω2t
+ (1/2 K2A2) cos 2ω1t 2nd harmonic terms
+ (1/2 K2A2) cos 2ω2t
+ (K2AB) cos (ω1 – ω2)t 2nd -order products
+ (K2AB) cos (ω1 + ω2)t
+ (1/4 K3A3) cos 3ω1t 3rd harmonic terms
+ (1/4 K3B3) cos 3ω2t
+ (3/4 K3A2B) cos (2ω1 + ω2)t 3rd -order products
+ (3/4 K3AB2) cos (2ω2 + ω1)t
+ (3/4 K3A2B) cos (2ω1 - ω2)t
+ (3/4 K3AB2) cos (2ω2 - ω1)t (4)
In each case, the ‘order’ of the harmonic or product is determined by the sum of the nonnegative
coefficients (m + n) as they occur as arguments of each sinusoidal component in cos (±mω1 ± nω2)t.
From (4) we note a number of important facts. First, the amplitude of the two input sinusoids at
the fundamental frequencies ω1 and ω2 are distorted by the odd-order coefficients K1 and K3. Each
fundamental frequency component is affected differently by the corresponding initial component
amplitudes A and B.
Next, we find that the output consists of a number of additional signal frequency components that were
not present in the initial two-tone input signal – this is the wellknown intermodulation problem. The new
signal components appear in the form of DC, 2nd and 3rd harmonics as well as 2nd and 3rd-order ‘sum
and difference’ products.
7
whitepaper
We see that the second harmonics and second-order products are affected by the K2 power series
coefficient only. As with the fundamental terms, each component 2ω1 and 2ω2 is affected differently
depending upon bow the amplitudes A and B enter into the coefficients. In other words, the harmonic
at 2ω1 is determined only by the amplitude A of the input tone at ω1 and the harmonic at 2ω2 is
determined only by the amplitude B of the input tone at ω2.
In the case of the third-order harmonics and third-order ‘sum and difference’ products, the same
observations related to K3 hold as with the second-order case with K2. When viewed in the light of real
communication systems, there is an additional consideration that is unique, however, to a subset of the
third-order products in (4). If the two input tones comprising Vin are relatively close together (as would
occur in the case of two transmitters operating in the transmit band of a cellular system), the products
at the frequencies (2ω1 - ω1) and (2ω2 - ω1) will occur relatively close to the fundamentals ω1 and ω2. In
real systems operating in the same band, all of the other extraneously generated signal components
in (4) will either be out of the receiver pass bands or will be filtered out by various components of the
system. If the signal components at (2ω2 – ω2) and (2ω2 - ω1) enter a nearby receiver, they can cause
severe reception problems.
On some antenna sites where systems in different bands are co-located, it is possible for some of the
higher-order harmonics to affect systems operating in a different band. Consider the case where a
VHF system operating at 150 -160 MHz is operating on the same site as a UHF system operating near
450 - 480 MHz. Depending upon the frequency plan, it is possible for third harmonics (and possibly the
last set of third-order ‘sum and difference’ products) generated in the VHF system to enter the receive
bands of the UHF system.
In the case where the amplitudes of the input tones are equal, A = B in (3) and therefore (4) shows
that each subset of equal-order tones appearing in Vout will have equal amplitudes. For example, both
coefficients of output tones at ω1 and ω2 are equal when A = B. Considering the second-order signals,
the same is true of the set of tones at 2ω1 and 2ω2. In the case of the generated third-order tones, the
set of third harmonics have equal coefficients and the set of ‘third-order sum and difference products’
have identical amplitudes.
When A = B, it is common to compare the relative amplitudes of the fundamental output tones at ω1
and ω2, with the amplitudes of the closest third-order tones at (2ω1 - ω2) and (2ω2 - ω1). For a mild
non-linearity, | K1 | >> | K3 | and the output amplitude at ω1 and ω2 will be nearly directly proportional
to the input amplitude at ω1 and ω2. The thirdorder products at (2ω1 - ω2) and (2ω2 - ω1) have an
amplitude that is proportional to the cube of the input amplitude at ω1 and ω2. On a log – log scale with
input power on the abscissa and the output power on the ordinate, it is apparent that the third-order
products have a slope three times greater than the slope of the fundamental. This means that on a
decibel scale, the third-order products change three times faster than the fundamental tones at ω1
and ω2. In other words, a 6 dB change in input power will have a corresponding 6 dB change in output
power at ω1 and ω2 but a 18 dB change in power at (2ω1 - ω2) and (2ω2 - ω1).
Although this is accepted as a common rule of thumb when measuring and comparing intermod.
levels, it must be understood that there is the underlying assumption that | K1 | >> | K3 |. This is held
to be true for 1) a wide range of input power levels and 2) for each mechanism causing the non-linear
transfer characteristic. This assumption is not necessarily true if one compares the different
8
whitepaper
mechanisms of non-linearity due to magnetic hysteresis versus non-linearity caused by metal oxides
occurring between a crimped connector in partial contact with the outer conductor of a cable. In
systems which measure high levels of third and higher-order IM, the non-linearity is not necessarily
amplitude and mechanism invariant. Although beyond the present discussion, some IM producing
mechanisms have been found to be frequency dependent.
For the simple case of a weak Cubic non-linear system or component, we see from (4) the general
effect of taking the input spectrum energy at ω1 and ω2 and transforming the energy to a new spectrum
that includes the original. Depending upon the frequency spacings between transmitters and receivers
that are co-located or located near each other, some components of the new spectrum can cause
interference in the receiving systems. Higher-order harmonics can cause problems on bands far
removed from the original band that contains components that function as I.M. sources.
If we consider the case where the non-linearity can be expressed as
Vout ~ K1 Vin + K2 (Vin)2 + K3 (Vin)3 + K4 (Vin)4 + K5 (Vin)5, (5)
the resulting output for the same two tone input (3) is a very large expression and will include
a) harmonics up to the fifth-order
b) ‘sum and difference’ products at the second through fifth-order.
The coefficients of the fundamental frequencies ω1 and ω2 present in the output Vout are
(K1A + 3/2 K3AB2 + ¾ K3A3 + 15/4 K5 A3B2 + 15/8 K5 AB4 + 5/8 K5A5) cos ω1t
+ (K1B + 3/2 K3A2B + ¾ K3B3 + 15/4 K5 A2B3 + 15/8 K5 A4B + 5/8 K5B5) cos ω2t . (6)
1st -order
The IM products of primary importance are those that fall closest to the fundamental input frequencies.
These occur at the odd orders only and are
(3/4 K3A2B + 15/8 K5A2B3 + 5/4 K5A4B) cos (2ω1- ω2)t 3rd order
+ (3/4 K3B2A + 15/8 K5B2A3 + 5/4 K5B4A) cos (2ω2- ω1)t
+ (5/8 K5A3B2) cos (3ω1- 2ω2)t 5th -order
+ (5/8 K5B3A2) cos (3ω2- 2ω1)t (7)
If we write the relationship between the two input tones á2 and á1 as
ω2 - ω1 = Δ,
9
whitepaper
then the generated additional frequencies closest to á2 and á1 will occur at
ω1 - Δ,
ω2 + Δ,
for third-order non-linearity (2) above and
ω1 - Δ,
ω2 + Δ,
ω1 - 2Δ,
ω2 - 2Δ,
in the case of a fifth-order non-Iinearity in (5). This pattern continues for higher-order non-linearities.
In the fifth-order case, the ‘rule of thumb’ relating the relative change in product order w.r.t. input
level change will hold if | K1 | >> | K3 I >> | K5 | is true for a wide range of input amplitudes and IM
generating mechanisms. From (6) and (7) we see that on a loglog scale, the fifth-order products will
change at a rate five times faster than the fundamental for any given input change. For example, when
the inequalities between K1, K3 and K5 hold, a 6 dB change in input amplitude will correspond to a 18
dB change in third-order product levels and a 30 dB change in fifth-order product levels.
In the cases of 7th, 9th,11th -order non-linearities and beyond, the general pattern for the important
coefficients has been established in equations (4), (6) and (7). It is apparent that the odd-order
coefficients in the model of the system/component non-linear behavior determine the magnitude of the
IM generated additional frequencies that fall closest to the fundamental input tones.
The same frequency generating effects occur when more than two signals pass through a non-linear
device or system. In these cases, each input tone will ‘ beat’ with the other tones to produce all
combinations of the ‘sum and difference frequencies’ associated with the degree of non-linearity under
consideration. The analysis will exactly parallel that shown above and the associated expressions for
the coefficients making up the output voltage will become extremely large. When viewed as expressed
in (9) and (10), the various IM generated tones close to the input frequencies are apparent.
4.0 Conclusion
The subject of intermodulation is extremely important for modem day communication systems. An
understanding of the fundamentals of this subject is important for equipment designers, system
designers, network operators, or anyone else responsible for proper operation of a communication
system. Sinclair openly welcomes further discussion of this subject with any interested parties.
10

Intermodulation Fundamentals White Paper

Uploaded by

Copyright:

Available Formats

Intermodulation Fundamentals White Paper

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Intermodulation Fundamentals White Paper

Uploaded by

Copyright:

Available Formats

Innovative Communication Solutions

2f1 – f2, 2f2 – f1, third-order products,

3f1 – 2f2, 3f2 – 2f1, fifth-order products.

2f1 - f2, 2f2 - f1, third-order products,

3f1 - 2f2, 3f2 - 2f1, fifth-order products,

4f1 - 3f2, 4f2 - 3f1, seventh-order products.

• Transmission line junctions such as tees and stars

• Antenna feed lines

• Any mounting brackets for antennas, feed lines, etc.

• Tower member joints.

• Tower guy wires.

• Nearby fences, gates and metal signs.

Magnetic effects cause IM by virtue of hysteresis in magnetic or paramagnetic materials at locations

• If braids/strands are present, minimize cable flexing when in use.

• Soldered braids/strands are better than not soldered.

3.0 Non-linear model and simple amplitude analysis

Vout = K1 Vin + K2 (Vin)2 + K3 (Vin)3 + … (1)

Neglecting the terms beyond the cubic, we get

Vout ~ K1 Vin + K2 (Vin)2 + K3 (Vin)3 . (2)

Consider now the pure two-tone input voltage

Vin = A cos ω1t + B cosω2t, (3)

Vout ~ ½K2 (A2 + B2) DC term

+ (K1A + 3/2 K3AB2 + ¾ K3A3) cos ω1t Fundamental terms

+ (K1A + 3/2 K3AB2 + ¾ K3A3) cos ω2t

+ (1/2 K2A2) cos 2ω1t 2nd harmonic terms

+ (1/2 K2A2) cos 2ω2t

+ (K2AB) cos (ω1 – ω2)t 2nd -order products

+ (K2AB) cos (ω1 + ω2)t

+ (1/4 K3A3) cos 3ω1t 3rd harmonic terms

+ (1/4 K3B3) cos 3ω2t

+ (3/4 K3A2B) cos (2ω1 + ω2)t 3rd -order products

+ (3/4 K3AB2) cos (2ω2 + ω1)t

+ (3/4 K3A2B) cos (2ω1 - ω2)t

+ (3/4 K3AB2) cos (2ω2 - ω1)t (4)

If we consider the case where the non-linearity can be expressed as

Vout ~ K1 Vin + K2 (Vin)2 + K3 (Vin)3 + K4 (Vin)4 + K5 (Vin)5, (5)

a) harmonics up to the fifth-order

b) ‘sum and difference’ products at the second through fifth-order.

+ (3/4 K3B2A + 15/8 K5B2A3 + 5/4 K5B4A) cos (2ω2- ω1)t

+ (5/8 K5A3B2) cos (3ω1- 2ω2)t 5th -order

+ (5/8 K5B3A2) cos (3ω2- 2ω1)t (7)

If we write the relationship between the two input tones á2 and á1 as

for third-order non-linearity (2) above and

You might also like