User Manual
115 of 562
V 1.0
2021-08-25
XDPP1100 technical reference manual
Digital power controller
Compensator
An example of a special case for the FF calculation is when the ACF converter (Loop 0) is followed by post-buck
topology (Loop 1). Thus, the Loop 1 input 1 is the output of Loop 0, and the following settings should be
applied:
•
Input source for Loop 1 should be selected through
pid1_ff_vrect_sel
to PRISEN (TS V
IN
)
•
Input voltage telemetry source should be set to 2 (Loop 0 V
OUT
)
through register
tlm_vin_src_sel
•
Filtering by Loop 0 output voltage telemetry LPF (register
tlm0_kfp_vout
) as well as Loop 1 input voltage
telemetry LPF (register
tlm1_kfp_vin
)
6.2.3
Override and adjustment options for FF
For the FF term there is a possibility to override the HW-computed FF with FW and/or perform other
adjustments depending on the application needs. These features are discussed below in more detail, and their
implementation is shown in
Figure 52
FF override and adjustment diagram
The option of overriding the computed FF term can be set via:
•
Register
pid_ff_override_sel
(select the override option)
•
Register
pid_ff_override
(set a suitable FF value)
This option is suitable for boost or buck-boost derived topologies, and it can be also used to disable the FF
functionality by setting the
pid_ff_override
to 0.
Further FF adjustments include the following:
•
Scalable FF gain via register
pid_ff_gain_scale
decreases or increases the FF effect depending on the gain
value. Typically, the gain is set to 1.0.
•
Adjustable offset for the FF term compensates for the dead time error through register
pid_ff_dt_adj
. As an
example, if the dead time is 1 percent of the period, this register needs to be set to 1 percent. If the value is
set to 0, the PID integrator will compensate.
In addition to these adjustments, optional low-pass filtering can be applied to the computed FF term before
being used by the PID compensator for duty calculation. The BW of this filter can be set via register
pid_kp_ff_lpf
. High BW results in fastest FF response, whereas low BW might smooth the noise at the sensed
signal but will slow the FF response. The
pid_kp_ff_lpf
uses exponential format, where the upper three bits
represent the exponent and the lower two bits represent the mantissa, as shown in Equations (6.29) to (6.32).
𝐾
𝐹𝑃
_𝑒𝑥𝑝 = 𝑝𝑖𝑑_𝑘𝑝_𝑓𝑓_𝑙𝑝𝑓[4: 2]
(6.29)
𝐾
𝐹𝑃
_𝑚𝑎𝑛 = 4 + 𝑝𝑖𝑑_𝑘𝑝_𝑓𝑓_𝑘𝑝𝑓[1: 0]
(6.30)
0
1
pid_ff_override
U0.10
pid_ff_override_sel
U0.10
pid_ff_gain_scale
U1.4
S
+
+
U1.10
pid_ff_dt_adj
U-2.10
LPF
U0.10
pid_kp_ff_lpf
3E2M
feed_forward
U0.10
computed_feed_forward
U0.10
