You are on page 1of 7

IM products estimation via computer

program, Part 1
H. Ng - May 28, 2014

In the February and March issues of Via Satellite publication, R. Brooker published a two-part series
describing the basic principle of intermodulation products, the cause of intermodulation distortion
and the result of distortion and sideband regrowth in nonlinear amplifiers.1 R. Brooker further
described how a system engineer could predict, by computational algorithm, and how to minimize
the intermodulation distortion and sideband regrowth. In the high power amplifier (HPA)
manufacturing side, CPI pointed out the desire for linear amplifiers.2 CPI further indicated that HPA
operations could be optimized by linearizer and the enhanced performance could be estimated by
empirical formula to provide the optimum output-back-off (OBO) to achieve a desirable carrier-t-
-intermodulation ratio.

In this article, we are implementing a computer program based on the methodology published by J.
C. Fuenzalida,

O. Shimbo, and W. L. Cook, in the COMSAT Technical Review.3 Shimbo et al. provide a
computational method to estimate the undesired interference caused by intermodulation products of
angle-modulated (e.g., constant envelope) carriers in a nonlinear amplifier. The implemented
computer program provides a hands-on experience in:

(a) the computation of intermodulation products (IM) and the distribution of IM inside and outside of
the transponder band or the earth-station high power amplifier pass band;

(b) the computation of carrier-to-intermodulation ratio (C/IM) and the distribution of C/IM across the
transponder or the earth-station HPA;

(c) determining the advantage of using linearizer with high power amplifiers; and

(d) tabulating a simple power sum of C/Nup+C/Ndown+C/IM for the operator to determine the
optimum transponder operating point.

In a two-carrier environment, for example, the intermodulation products not only rob power from the
main carriers but also suppress the amplitude of the lower-power level carrier. In a multi-carrier
environment, the intermodulation products interfered with the main carriers to create the carrier-t-
-intermodulation ratio (C/IM) interference. This C/IM ratio degrades the satellite link performance.
Hence, it is essential to optimize the satellite links by managing and controlling the intermodulation
product levels.

Fuenzalida, Shimbo, and Cook have shown that the complex intermodulation product amplitude and
phase could be represented by

where

For a two-carrier environment, if we pick K1 = 1 , K2 = = Km = 0, we get a replica of one of the two


carriers. And if we pick K1 = 2 , K2 = -1 , K3 = = Km = 0, we get one of the third order IM products.
And if we pick K1 = -1 and K2 = 2, we get the second third order IM product. In particular, the output
power of the two carriers are given by

where, according to equation 1,

and A1 = (2P1) , A2 = (2P2) . Using the above Ki sequence, the third-order intermodulation
products are given by

and the fifth-order intermodulation products are given by

We could continue in a similar manner for larger number of carriers and other higher order
intermodulation products.
Complex coefficients, bsr and bsi

Complex coefficients, bsr and bsi

Fuenzalida, Shimbo, and Cook further indicated that equation (1) is a complex intermodulation
product amplitude and phase equation and is also based on a set of single-carrier transfer curves
relating to the output power, Ao, and phase, o, to the input power, Ai. Hence, equation (1) could be
rewritten as

They further state that the resulting optimum coefficients Real(bs) and Imag(bs) could be obtained by
minimizing the squared difference between the left-hand-side real component and the right-han-
-side real component and the imaginary components, in the same manner, of equation 2. They
suggested that the Fletcher-Powell optimization scheme is an efficient algorithm to determine the bsr
and bsi coefficients. For us, we decided to use the MathCad built-in functions to solve 10
simultaneous equations (i.e., L = 10 sets of coefficients and Shimbo et al. also indicated that 10 sets
should be sufficient for typical nonlinear characteristics) and to minimize the squared difference. As
an example of the application of this technique, a typical traveling wave tube (TWT) amplifier on one
of the early domestic satellites is considered. The known measured transfer curves for the TWT
amplifier are shown in Figure 1.

Figure 1 Amplitude and phase characteristics of the TWT amplifier

Applying the MathCad built-in functions (e.g., Given and Minerr(b1, b2, b3, )), the resulting
coefficients, bs , for S= 1, 2, , 10, are shown in Table 1 and the parameter, , is equal to 0.98.
This set of data has been formatted into a data file for the computer program, see below, and is
labeled as twtbrbi.txt.

Table 1 Coefficients ( bs) for the TWT amplifier

As suggested by CPI that linearizer and TWT amplifier (i.e., LTWTA) can be optimized to provide a
more efficient amplifier in terms of intermodulation products. Without showing the known measured
transfer curves for the LTWTA and applying the MathCad built-in functions the resulting
coefficients, bs , for S= 1,2, , 10, are shown in Table 2 and the parameter, , is equal to 1.38. This
set of data has been formatted into a data file for the computer program and is labeled as
ltwtbrbi.txt.
Table 2 Coefficients ( bs) for the LTWTA

2-carrier environment

2-carrier environment

A Fortran program has been compiled to implement the above equations for a 2-carrier environment.
The purpose of this program is to illustrate a couple of known behaviors of nonlinear amplifiers, e.g.,
small signal suppression and the intermodulation products rob power from the output amplifier. An
example-input file for the computer program is illustrated below showing two equal amplitude input
carriers. The bs coefficient data file is twtbrbi.txt.

Table 3 input example for two equal-power input carriers

In a simplified frequency domain, the 3rd-order intermodulation products could be represented by:
cos(21 - 2t) and cos(22 - 1t), and the 5th-order intermodulation products could be represented by:
cos(31 -22)t and cos(32 -2 1)t. For MCPC operation (e.g., no carrier-on-carrier operation and
having |1- 2| > 0), there would not be any spectrum overlap between the main carrier(s) and the
intermodulation products. Hence, the nonlinear satellite transponder output would only replicate the
input main carriers and without any intermodulation products. Consequently, the transponder could
operate at saturation, if desired. Figure 2 shows a single carrier reference (i.e., the transponder is
operating with one carrier only and it is used as the reference) and the sum of the 2-carrier output.
The figure also shows the output for one of the two carriers.

Figure-2 Two equal-amplitude carrier performance

The salient point of this figure is that the total output power (the sum of the two equal-output-power
carriers is represented by the total OBO dash-blue curve) tapers off towards 1.6 dB, on the
ordinate, as the transponder is operating towards and at saturation (i.e., ibo = 0.0 dB). The plateau
level of 1.6 dB is the maximum loss of power by the two carriers. The figure also shows that, for
each of the two output carriers, the individual maximum output power (i.e., obo) is about 4.6 dB
below saturation. In other words, -4.6 is resulted from the 3.0 dB for splitting the total output power
for each of the two carriers and the 1.6 dB for the output power loss resulted from the creation of
the intermodulation products in the nonlinear amplifier.

Table 4 is an example-input file for two unequal-power carriers. The lower-power carrier is 8.5 dB
lower than the higher-power carrier. The computer program output is shown in Figure 3.

Table 4 input file example for 2 unequal power input carriers


Figure 3 small signal suppression and the amount of suppression of the lower-level carrier

The salient points of Figure 3 suggest that:

a)-the power loss due to the generation of the intermodulation products is must less than 1.6
dB, as shown in the previous case;

b)-the higher power carrier output approaches the single carrier saturation;

c)-in the linear region of the amplification, e.g., IBO < -7 dB, the amplitude difference
between the high-power and the low-power carriers is 8.5 dB which is exactly the difference at the
input; and

d)-the amplitude of the lower-power carrier has been suppressed by as much as 4.5 dB as the
nonlinear amplifier is operating approach saturation and at saturation. See the right-hand-side
graph of Figure 3.

This small signal suppression phenomenon had been used by cable-head-ends to switch analog
FM/TV transmission to digital MCPC operation, in the early 2000 time frame. This small signal
suppression could also be used to investigate the suppression of the satellite transponder front-end
thermal noise by the nonlinear transponder, in a single saturated-carrier operation. The resulting
effect is to have a higher C/Nup-thermal at the transponder output than the C/Nup-thermal at the input to the
transponder.

References

1 R. Brooker, Make Sense of Nonlinear Distortion. Part 1: Viewing distortion as a modulation


effect, Via SatelliteR. Brooker, Make Sense of Nonlinear Distortion. Part 1: Viewing distortion as a
modulation effect, Via Satellite, February 2014.

2 CPI SuperLinear HPAs The Latest Advancement in High Power Amplifier Technology,
Communications & Power Industries.

3 J. C. Fuenzalida, O. Shimbo, and W. L. Cook, Time-domain analysis of intermodulation effects


caused by nonlinear amplifiers, Comsat Technical Review, Vol. 3 No. 1, Spring 1973.

Editors note: Part 2 will cover Multi-carrier environment and a computer program

You might also like