Logo of the Physikalisch-Technische Bundesanstalt

Creep measurements on precision torque transducers in comparison


Due to internal shifting processes, torque transducers react only slowly to a changing of loads. After a change has been effected, it takes several minutes until the output signal goes asymptotically into a stable state (Figure 1). If one wants to assess the suitability of a torque transducer for dynamic or high-precision tasks, this behaviour - which is called "creep" - must be known. For that reason, DIN 51309 prescribes that during the calibration of torque transducers, corresponding creep measurements have to be carried out.

In such measurements, the load must be kept exactly constant after a fast load change has taken place while the response of the transducer to the load step is recorded. In the case of calibration facilities with a direct deadweight effect (deadweight torque machines) which generate the torque by means of a lever arm and load masses, this is ensured by the design. In comparison-calibration facilities whose torque generation is based on motor drive this is, however, difficult to achieve. This is why the method for creep measurements in calibration facilities according to the comparison principle, which is recommended in the DIN standard, cannot be applied. In an investigation it has been determined whether alternative measurement methods lead to results which are equivalent to those achieved in deadweight torque calibration facilities [1].

Creep curve. Signal progression of a torque transducer after a fast load change from 0% to 100% of the nominal load. During the measurement, the load remains constant at 100%.

Figure 1: Creep curve. Signal progression of a torque transducer after a fast load change from 0% to 100% of the nominal load. During the measurement, the load remains constant at 100%.

As a matter of principle, the creep measurement procedures investigated are affected much more strongly by interfering signals than creep measurements in deadweight torque machines. For that reason, the data for the evaluation of the creep measurements in comparison-calibration facilities were adapted to an exponential model function. The model parameters then lead to the creep characteristics searched for, but they also show what must be taken into account if the results of the different procedures are to be comparable.
The development of creep processes depends not only on the amount and on the speed of the load change, but also on the load history of the transducer. Apparently, the stress states in the transducer, which are connected with the loads, are - depending on the duration of loading - subject to a more or less strong accumulation which leads to a typical saturation behaviour. For that reason, creep measurements are comparable only if the results have been obtained from equivalent loading processes. Here, the loading process comprises not only the period of time in which the actual creep process is recorded – i.e. after the load change – but also a period of time prior to the load change. Due to the asymptotic character of the creep, these periods of time comprise 20 to 60 minutes each time.
In addition, it must be ensured - for the comparability of creep measurements - that the time windows in which the creep processes are recorded agree in both the length as well as in the position within the creep development.
If all these requirements are complied with, the relative deviations between creep measurements in comparison facilities and creep measurements in deadweight torque machines can be limited to approx. 0.1 · 10-4. The deviations thus lie in the order of magnitude of the relative measurement uncertainties by which creep measurements are affected. In this sense, the creep measurements performed are, thus, equivalent - independent of the procedure by which the results have been obtained.


[1] Brüge, A.: Creep measurement in reference torque calibration machines, IMEKO 2010: TC3, TC5 and TC22 Conferences, Pattaya, 21-25, November, 2010, Download:

Contact person:

Andreas Brüge, Dept. 1.2, WG 1.22, e-mail: andreas.bruege@ptb.de