Preliminary Technical Data UG-1828
Rev. PrB | Page 189 of 277
• Properly tune the transmitter attenuation and/or the step attenuator to make sure that the ACPR performance at the device
transmitter output is satisfactory before passing to power amplifier. In addition, make sure that the transmit peak signal is around
P1dB compression region for optimal DPD performance.
The user could compare the ACPR performance through spectrum analyzer with and without using the integrated DPD. Significant
ACPR performance improvement with the integrated DPD should be observed even with internal LO sources. For TETRA waveforms,
the ACPR after the second iteration of DPD is between −70 dB and −60 dB at an amplifier compression of P1dB. For LTE waveforms, the
ACPR after the second iteration of DPD is between −55 dB and −50 dB at an amplifier compression of P1dB.
Tuning the Model Tap Order
DPD can be considered as an adaptive filter which is modelled according to the behavior of the power amplifier. As mentioned
previously, the ADRV9001 default model (Model 4) consists of four taps. Each tap consists of a series of polynomial terms to fit the
nonlinear behavior due to compression at higher output power. The order of polynomial terms is determined by intermodulation falling
closer to the carrier spectrum. In DPD, the orders of intermodulations that must be considered are usually third, fifth, and seventh
orders, with decreasing magnitude, respectively. An nth-order intermodulation expands the signal bandwidth by n times. By inspecting
the bandwidth expansion factor on a spectrum analyzer, the user can estimate how many orders of intermodulations that must be
included in the polynomial terms, in order to suppress the spectral regrowth down to the required ACPR. It is important not to include
higher order power terms than needed, which might cause the DPD unstable.
DPD Model 4 consists of four taps as shown in the example tap arrangement diagram in Figure 182. The four taps can be classified into
three categories:
The main tap – The main tap of the DPD adaptive filter that suppresses most of the spectral regrowth due to intermodulation; hence it
has the greatest number of polynomial terms. It is labelled as TAP
1
.
The side taps – There are two side taps on each side of the main taps. They are memory terms that compensate for frequency-dependent
distortion in the frequency domain, and time misalignment between the transmit and receive captured data. The side taps have the same
number of polynomial terms; and each side tap has about half of the number of polynomial terms of the main tap. They are labelled as
TAP
0
and TAP
2
.
The cross-term tap – The cross term is designed for further suppressing the residual spectral regrowth left over by the other three taps.
The number of polynomial terms is usually equal to or less than that of each side tap. It is labelled as TAP
3
.
x(
t
–2)x|(t–2)|
k–1
k = 1:3
x(
t
)|x(
t
)|
k
–1
k = 1:3
x(
t
–1)|x(
t
–1)|
k–1
k = 1:7
DECREASING
t
IN x(
t
)
DECREASING
t
IN |x(
t
)|
k
INCREASING
t
IN |x(
t
)|
k
INCREASING
t
IN x(
t
)
x(
t
–1)|x(
t
–2)|
k
–1
NONE or k
= 2:3
73 3
3
SIDE TERMS
MAIN TERMS
CROSS TERMS
TAP
1
TAP
3
TAP
2
TAP
0
24159-155
Figure 183. Example Polynomial Constellation Configuration for Model 4
Note that in Figure 182, k represents the order. To handle 7th harmonics, the main tap must include the power terms up to k = 7.
To find the best model tap order for a specific power amplifier design model, the user can take the following recommended procedures:
1. Set the amplifier output power to have a compression ratio of 1 dB or slightly less, as shown in Figure 1, i.e. the maximum peak of
the output signal is 1 dB below ideal linearity.
To Next Page
Loading...