This paper presents a fundamental external force detection framework for construction manipulator. Such industrial application primarily requires practicality that satisfies detection requirements such as accuracy and robustness while ensuring (i) low cost, (ii) wide applicability, and (iii) simple detection algorithm. To satisfy (i) and (ii), our framework adopts hydraulic sensors as force sensors. Hydraulic sensor inherently detects error force components that generate depending on joint kinetic state and differ in identification-difficulty due to hydromechanical uncertainty and nonlinearity. To satisfy (ii) and (iii), theoretical, experimental, and conditional identification methods without complicated model analysis are applied in static, uniform motion, and accelerated motion states for identifying self-weight, driving, inertial forces defined as dominant error components. Experiments were conducted using our instrumented hydraulic arm system. Result of no-load task indicates that our framework lowers a threshold to define the on-off state of external force applied to hydraulic cylinder, independent of joint kinetic conditions. Result of on-load task confirms that our framework identifies off-states in which external force is not applied to hydraulic cylinder.