Winco DGC-2020 - Manual (Page 335)

9400200990 Rev K DGC-2020 Tuning PID Settings C-5
KD - Derivative Gain
If the performance with KP and KI alone is satisfactory, you may stop here. Otherwise, KD, the derivative
controller gain, can be used in conjunction with TD, the noise filter constant, to reduce overshoot obtained
with the PI control. Setting KD and TD is an iterative process. Start with small values of KD such as 0.1 or
half the KI value, whichever is less.
TD is the constant of the low pass filter which filters the controller input if high frequency interference
presents a problem when derivative control is employed. TD ranges from 0 to 1. TD=0 is no filtering,
TD=1 is heaviest filtering.
Tuning of KD can be achieved through the following steps. Raise load control KD and check for stability.
Each time KD is raised, parallel another generator with the unit being tuned, and check for stable load
sharing. Then drop the second generator and check that the unit being tuned is still stable. Raise KD until
the system is unstable, and then lower it to half the value where instability is first attained. If high
frequency noise seems to be entering the system, raise TD and tune KD again. TD ranges from 0 to 1.
TD=0 is no filtering, TD=1 is heaviest filtering.
If high frequency noise seems to be entering the system, set TD to 0.001 and see if the noise induced
behavior is reduced. Raise TD until desired reduction of noise behavior is achieved. Once TD has been
set, tune KD again. TD ranges from 0 to 1 with an increment of 0.001. TD=0 is no filtering, TD=1 is
heaviest filtering. If noise again appears to be a problem, adjust TD until desired behavior is achieved,
then retune KD.
GENERIC GAINS FOR MULTIPLE MACHINE TYPES
The following method is suggested for determining generic gains for multiple machine types.
1) Decide what levels of reverse power and reverse var (loss of excitation) protection you need to
use.
2) Once the criteria of step 1 have been established, tune a unit so that you can parallel to another
unit at no load and not cause any trips.
3) Parallel two machines onto a load, and verify that acceptable load sharing occurs.
4) Add and drop loads with machines paralleled to verify acceptable load sharing occurs, and no
trips occur.
5) Once the settings are deemed “good”, save them as initial settings for a given machine
configuration for all future jobs. You shouldn’t need to change them unless you get trips or need
to change load sharing characteristics.
6) Test the units paralleled under no load and verify that no trips occur.
7) Parallel two machines onto a load, and verify that acceptable load sharing occurs.
8) Add and drop loads with machines paralleled to verify acceptable load sharing occurs and no
trips occur.
9) If you need to modify settings for a particular machine type, keep those settings to be used as
initial settings for all future machines of that type.
10) Test every machine with steps 6, 7, and 8.
It is not expected that you will ever get one set of numbers that will work for all machines, but it is
probable that you may end up with 6 to 12 sets of settings that cover a wide range of machine sizes and
engine manufacturers. However, once a set of gains has been determined for a particular machine type,
the same gains should work in all identical machines.