diff --git a/src/pdiagmat.jl b/src/pdiagmat.jl
index 70ab2c5..0c7a62c 100644
--- a/src/pdiagmat.jl
+++ b/src/pdiagmat.jl
@@ -51,7 +51,7 @@ function pdadd!(r::Matrix, a::Matrix, b::PDiagMat, c)
     return r
 end
 
-*(a::PDiagMat, c::Real) = PDiagMat(a.diag * c)
+*(a::PDiagMat, c::Real) = Diagonal(a.diag) * c
 function *(a::PDiagMat, x::AbstractVector)
     @check_argdims a.dim == length(x)
     return a.diag .* x
diff --git a/src/pdmat.jl b/src/pdmat.jl
index 7f4fcf0..cd81661 100644
--- a/src/pdmat.jl
+++ b/src/pdmat.jl
@@ -64,7 +64,7 @@ function pdadd!(r::Matrix, a::Matrix, b::PDMat, c)
     _addscal!(r, a, b.mat, c)
 end
 
-*(a::PDMat, c::Real) = PDMat(a.mat * c)
+*(a::PDMat, c::Real) = a.mat * c
 *(a::PDMat, x::AbstractVector) = a.mat * x
 *(a::PDMat, x::AbstractMatrix) = a.mat * x
 \(a::PDMat, x::AbstractVecOrMat) = a.chol \ x
diff --git a/src/pdsparsemat.jl b/src/pdsparsemat.jl
index eb5f640..8fc76cc 100644
--- a/src/pdsparsemat.jl
+++ b/src/pdsparsemat.jl
@@ -65,7 +65,7 @@ function pdadd!(r::Matrix, a::Matrix, b::PDSparseMat, c)
     _addscal!(r, a, b.mat, c)
 end
 
-*(a::PDSparseMat, c::Real) = PDSparseMat(a.mat * c)
+*(a::PDSparseMat, c::Real) = a.mat * c
 *(a::PDSparseMat, x::AbstractMatrix) = a.mat * x  # defining these seperately to avoid
 *(a::PDSparseMat, x::AbstractVector) = a.mat * x  # ambiguity errors
 \(a::PDSparseMat{T}, x::AbstractVecOrMat{T}) where {T<:Real} = convert(Array{T},a.chol \ convert(Array{Float64},x)) #to avoid limitations in sparse factorization library CHOLMOD, see e.g., julia issue #14076
diff --git a/src/scalmat.jl b/src/scalmat.jl
index feabe8a..b5975fc 100644
--- a/src/scalmat.jl
+++ b/src/scalmat.jl
@@ -38,8 +38,8 @@ function pdadd!(r::Matrix, a::Matrix, b::ScalMat, c)
     return r
 end
 
-*(a::ScalMat, c::Real) = ScalMat(a.dim, a.value * c)
-/(a::ScalMat, c::Real) = ScalMat(a.dim, a.value / c)
+*(a::ScalMat, c::Real) = Diagonal(fill(a.value * c, a.dim))
+/(a::ScalMat, c::Real) = Diagonal(fill(a.value / c, a.dim))
 function *(a::ScalMat, x::AbstractVector)
     @check_argdims a.dim == length(x)
     return a.value * x