Analog Devices ADRV9005 Reference Manual page 238

Reference Manual
DIGITAL PREDISTORTION (DPD)
Figure 218 shows that d(t) is the raw complex transmit signal before predistortion. Its amplitude is the basis used by the DPD actuator to
predistort the d(t) through its LUT. The LUT consists of four taps calculated with precomputed DPD coefficients α, as shown here:
TAP = a + a |d(t)| + a
0 0,0,0 0,0,1 0,0,2
TAP = a + a |d(t − 1)| + a
1 1,1,0 1,1,1
TAP = a + a |d(t − 2)| + a
2 2,2,0 2,2,1
Note: The TAPx equations represent the default power term setting for each tap in the model 4, from which, derive b
TAP = a |d(t − 2)| + a |d(t − 2)|
3 3,2,1 3,2,2
depending if a power term is included or excluded:
Tap 0: b = 1, b = 1, b
0,0,0 0,0,1 0,0,2
Tap 1: b = 1, b = 1, b
1,1,0 1,1,1 1,1,2
Tap 2: b = 1, b = 1, b
2,2,0 2,2,1 2,2,2
Tap 3: b = 0, b = 1, b
3,2,0 3,2,1 3,2,2
If using an array B for 4 taps, and for each tap using a byte to represent the above setting (the least significant bit represents the 0 power term),
it is clear that the default setting is equivalent to B[0] = 0x1F, B[1] = 0x7F, B[2] = 0x1F and B[3] = 0x1E.
Combine the connections from the four outputs to produce the final output, x(t), as the following:
x(t) = TAP [|d(t)|] × d(t) + {TAP
0
changeModelTapOrders
This flag selects the default model tap orders or chooses customized model tap orders. If this flag is "TRUE," use the next field in the data
structure, "modelOrdersForEachTap, to set the model tap orders for the specified channel. If it is "FALSE", then "modelOrdersForEachTap" is
ignored and it uses the default tap orders as discussed (B[0] = 0x1F, B[1] = 0x7F, B[2] = 0x1F and B[3] = 0x1E).
This is an array of bitmaps b
modelOrdersForEachTap
t, l t , i
It provides option to customize the order so that a power term can be included or excluded in the polynomial to better model the PA .
shows recommendations to set this field. Try these suggestions and find the best model through tests. The
discusses the method to select the best model tap orders as a part of DPD tuning recommendations.
Table 98. Suggested Model Orders for Narrowband Waveforms
Taps Model Orders for Each Tap
Tap 1 B[1] = 0x1F, 0x3F, 0x7F, 0xFF
Tap 0 and Tap 2 B[0] = B[2] = 0x03, 0x07, 0x0F, 0x1F, (Tap 0 and Tap 2 should be the same.)
Tap 3 B[3] = 0x0, 0x02, 0x06, 0x0E, 0x1E, 0x3E
analog.com
Figure 218. ADRV9001 DPD Model 4 LUT Configuration
2 3
|d(t)| + a |d(t)| + a |d(t)|
0,0,3 0,0,4
2
|d(t − 1)| + a |d(t − 1)|
1,1,2 1,1,3
2
|d(t − 2)| + a |d(t − 2)|
2,2,2 2,2,3
2 3
+ a |d(t − 2)| + a
3,2,3 3,2,4
= 1, b = 1, b = 1, b
0,0,3 0,0,4 0,0,5
= 1, b = 1, b = 1, b
1,1,3 1,1,4 1,1,5
= 1, b = 1, b = 1, b
2,2,3 2,2,4 2,2,5
= 1, b = 1, b = 1, b
3,2,3 3,2,4 3,2,5
[|d(t − 1)|] + TAP [|d(t − 2)|]} × d(t – 1) + TAP
1 3
7 for each tap t (t=0 to 3), formulated in the same way as discussed for the default setting.
i = 0 to
4
3 4
+ a |d(t − 1)| + a |d(t – 1)|
1,1,4 1,1,5
3 4
+ a |d(t − 2)|
2,2,4
4
|d(t − 2)|
= 0, b = 0, b = 0
0,0,6 0,0,7
= 1, b = 1, b = 0
1,1,6 1,1,7
= 0, b = 0, b = 0
2,2,6 2,2,7
= 0, b = 0, b = 0
3,2,6 3,2,7
[|d(t − 2)|] × d(t – 2)
2
ADRV9001
5 6
+ a |d(t – 1)|
1,1,6
t, l t , i
as the following
DPD Tuning and Testing
Rev. A | 238 of 377
Table 98
section