Torque talk

Terry Allen of Industrial Measurements looks at the evolution of the modern torque transducer.

The early days of torque measurement were confined to test bed applications.

The loading machine, whether water brake or electric generator, was suspended in bearings and the resultant torque reaction was measured by a spring balance.

This device was torsionally soft and was fine for steady slate running but not capable of measuring fluctuations induced by, for example, an internal combustion engine.

The introduction of a strain-gauge load cell to replace the spring balance has allowed dynamic measurements, but the stiffness of water pipes or electric cables to the dynamometer can, without careful design, detract from torque measurement accuracy.

As the support bearing arrangement for larger machines can be expensive, the cost-effectiveness of inline torque measurement (where the swinging arm principle is not required) should be considered.

The general concept of placing a measuring device in a machinery drive line to measure torque sounds very simple but, unlike most transducers, they have to transmit what they are measuring and any mechanical failure in the transducer results in a loss of drive between the drive and driven machinery.

The transducer must also be insensitive to axial loads or bending loads, which may occur due to misalignment in the driveline.

These problems can be overcome by selection of the correct type of transducer and associated couplings from the range available.

The first "inline" torque transducer I am aware of was used on the liner Queen Mary in the 1930s.

It employed the phase displacement principle, which in those days, with crude electronics, needed a length of 4m to adequately measure the angle of twist with around +/-5% accuracy.

The modern phase displacement torque meter measures the twist between a pair of toothed flanges, which generate sinusoidal signals in magnetic pickups in the form of internally toothed rings and circumferential coils.

The resultant phase change of the two signals varies linearly with torque and is measured by digital electronics.

Accuracies of +/-0.1% can now be achieved.

(All accuracies quoted in this article relate to the full-scale measurement of the device).

Phase displacement transducers are only available from a few sources in the world, and are best suited to high speed and high temperature applications where cost may not be a major issue.

Another European-made transducer uses a differential transformer measuring system.

It consists of two concentric cylinders fixed to a shaft either side of a "torsion section" (smaller diameter to give a reasonable angle of twist) and two concentric coils attached to the housing.

Both cylinders have circumferential slots, which rotate inside the coils.

An alternating current flows through a primary coil and when the slots start to overlap due to shaft twist, a torque-proportional EMF is induced in a secondary coil.

The EMF is conditioned to give a 0+/-10V output signal.

Although its performance may not quite match that of the phase displacement transducer, its cost is generally lower.

In the 1950s, still before most of you were born, reliable strain gauges were developed and beliefs that they were not suitable for long-term measurement were soon dispelled.

There are tens of thousands of load cells and torque cells using strain gauges, which are used all day, every day and give long-term stability and high accuracy.

The first rotating strain gauge torque sensor employed a system of slip rings in order to make the electrical connections from the casing to the rotating shaft.

Because the slip rings are carrying only millivolt signals from the strain gauges, the materials for both the slip rings and the brushes have to be vary carefully selected.

The normal procedure is to use coin silver for the slip rings and silver graphite for the brush gear.

Slip ring torque transducers still remain very popular for slower speed and short-term test applications.

The rotary transformer (inductive) data transmission system, using modern high-performance, high-temperature electronic components became popular in the 1980s.

It became known as a telemetry system but it only uses kilohertz carrier frequencies, not the megahertz of radio systems.

(Radio telemetry is used for data transmission in some torque applications, typically temporary ones).

Several companies use the rotary transformer system with strain gauges, and with typical inaccuracies of +/-0.2%, and speeds up to 50,000rev/min it can be a very cost-effective torque transducer.

The basic data transmission system is often integrated into proprietary spacer couplings.

It obviates the need for further mechanical couplings, and can give a substantial cost saving especially for couplings above 2kNm.

It is available for retrofit (clamp-on) applications on shaft diameters from 20mm (vehicle shafts) to 1000mm (wind turbines) allowing torque measurement up to several meganewton-metres with fast transient response.

This strain gauge plus rotary transformer system is versatile enough for use in special transducers including those where space is limited.

A German company has just launched a new design of inline rotary-transformer torque transducer, which overcomes the problem of increased inaccuracy below 10% of full scale.

It effectively creates a dual range torque transducer, which offers a 10:1 turn-down ratio with a high scale accuracy of +/-0.1% and a lower scale accuracy of +/-0.2%, ie +/-0.02% of its maximum capacity.

There is still only one strain gauge bridge but new "smart" onboard electronics and new calibration methods make this dual range possible at relatively little extra cost.

A British company makes a transducer using surface acoustic wave (SAW) technology.

The SAW device is used as a frequency dependent strain gauge and measures the change in resonant frequency caused by the strain in the shaft.

An RF couple transmits the signal from the shaft to a fixed pickup.

It is a relatively low-cost device offering a +/-0.25% accuracy giving a total system accuracy of around +/-0.35%.

It competes very well on price.

A few years ago, a magnetoelastic device was developed in the USA, which uses a magnetoelastic sleeve fixed to a stainless steel shaft.

This permanent magnet generates a magnetic field proportional to torque.

The resultant magnetic field is less dense that the earth's field, and so internal shielding is needed for the Hall-effect probes.

It is not an expensive transducer and fine for applications where measurement accuracy is not critical.

Along the same lines is the magnetostrictive transducer.

It is mechanically simpler than the magnetoelastic unit and relies on a premagnetised shaft proportionally changing its magnetic field when torque is applied.

Zero drift with time and temperature and the effect of adjacent magnetic fields, eg e-motors, can be a problem.

However, it is a low-cost transducer.

These two devices can provide an accuracy of around +/-1%.

Around 1999, a very large German company designed a laser system.

In essence, it is like a phase displacement system with photodiode detectors receiving light from mirrors at each end of a rotating shaft.

The reflectors are staggered to produce an offset at no load so that the differences in time intervals can represent the twist in the shaft.

About a year later, a USA institution patented a laser system design.

It differs from the German design, in as much as it uses one laser and one detector.

It examines the shaft signal over a 30 degree angle and "displays the torque as a function of angular displacement of the laser beam".

As far as I am aware, neither system has been developed into a product.

9 years ago I said it would take something revolutionary to make a big impact on the torque transducer market.

Since then we have had several new designs but none of these have challenged existing products.

Some believe there is an opportunity for low-cost mass-production, but I doubt if the market is big enough for a reasonable return on investment, even if a low-cost unit (to suit all environments) could be produced.

Back to top