Merge pull request #7 from JuliaControl/fp2

convert signals to correct type before computations

Project.toml

@@ -1,7 +1,7 @@
 name = "DiscretePIDs"
 uuid = "c1363496-6848-4723-8758-079b737f6baf"
 authors = ["Fredrik Bagge Carlson"]
-version = "0.1.2"
+version = "0.1.3"
 
 [compat]
 julia = "1.7"

README.md

@@ -35,6 +35,8 @@ u = calculate_control!(pid, r, y, uff)
 
 The parameters $K, T_i, T_d$ may be updated using the functions, `set_K!, set_Ti!, set_Td!`.
 
+The numeric type used by the controller (the `T` in `DiscretePID{T}`) is determined by the types of the parameters. To use a custom number type, e.g., a fixed-point number type, simply pass the parameters as that type, see example below. The controller will automatically convert measurements and references to this type before performing the control calculations.
+
 ## Example using ControlSystems:
 The following example simulates the PID controller using ControlSystems.jl. We will simulate a load disturbance $d(t) = 1$ entering on the process input, while the reference is $r(t) = 0$.
 
@@ -125,7 +127,7 @@ pid = DiscretePID(; K = T(K), Ts = T(Ts), Ti = T(Ti), Td = T(Td))
 res_fp = lsim(P, ctrl, Tf)
 plot([res, res_fp], plotu=true, lab=["Float64" "" string(T) ""]); ylabel!("u + d", sp=2)
 ```
-![Fixed-point simulation result](https://user-images.githubusercontent.com/3797491/249722782-2157d625-7eb0-4f77-b630-69199237f164.png)
+![Fixed-point simulation result](https://user-images.githubusercontent.com/3797491/249732319-0a3890d5-cb9c-45c2-93c7-20d3c7db0cf2.png)
 
 The fixed-point controller behaves roughly the same in this case, but artifacts are clearly visible. If the number of bits used for the fractional part is decreased, the controller will start to misbehave.
 

src/DiscretePIDs.jl

@@ -136,8 +136,12 @@ end
     (pid)(r, y, uff=0) # Alternative syntax
 
 Calculate the control output from the PID controller when `r` is the reference (set point), `y` is the latest measurement and `uff` is the feed-forward contribution.
+If the type of the input arguments differ from the numeric type used by the PID controller, they will be converted before computations are performed.
 """
-function calculate_control!(pid::DiscretePID, r, y, uff=0)
+function calculate_control!(pid::DiscretePID{T}, r0, y0, uff0=0) where T
+    r = T(r0)
+    y = T(y0)
+    uff = T(uff0)
     P = pid.K * (pid.b * r - y)
     pid.D = pid.ad * pid.D - pid.bd * (y - pid.yold)
     v = P + pid.I + pid.D + uff

test/runtests.jl

@@ -69,7 +69,7 @@ res2 = lsim(P, ctrl, Tf)
 
 ## Test with FixedPointNumbers
 using FixedPointNumbers
-T = Fixed{Int16, 10} # 16-bit signed fixed-point with 11 bits for the fractional part
+T = Fixed{Int16, 10} # 16-bit signed fixed-point with 10 bits for the fractional part
 pid = DiscretePID(; K = T(K), Ts = T(Ts), Ti = T(Ti), Td = T(Td))
 @test pid isa DiscretePID{T}