Chris Jones of Micro-Epsilon explains the design factors and signal processing characteristics involved in the use of laser displacement sensors.
Most non-contact laser triangulation displacement sensors in the marketplace today use the same measuring principle ie the laser triangulation technique, to convert a target distance into an output signal. However, the individual design of the sensor mechanics, the optics, the mechanical stability and signal processing algorithms are likely to vary widely from supplier to supplier. This means the ‘real world’ accuracy of a laser displacement sensor will differ significantly with each sensor supplier.
There are many factors that each play an important role in determining actual laser sensor accuracy. It is therefore vital to not only understand these factors and to what extent they can affect sensor accuracy, but also to be aware of the recent innovations that have been developed by sensor suppliers to help overcome these potential errors.
Dynamic changing of the target colour
The change in reflected light received from the target surface into the laser from changing surfaces can now be compensated for in real time. The exposure time or the amount of light produced by the laser can be optimally matched to the reflection characteristics of the target surface for each measurement value and without needing to apply any averaging filters. This results in light exposure perfectly matched to the surface conditions being measured, resulting in a stable and correct measurement value. However, not all sensor suppliers can provide this automatic real time surface compensation feature resulting in an error in the output measurement value.
Standard, commercially available laser triangulation displacement sensors normally operate with a time-shift control, which operates on intensity values received from previous measurement cycles. In this case, the amount of reflection from previous measurements, typically 3 to 5, is averaged and then used to anticipate the amount of laser intensity required for the next measurement. With changing or textured surfaces, this method of surface compensation produces measurement values that deviate significantly from the actual measurement value.
Optimising for surface roughness
When measuring on shiny metals or machined surfaces that have small micro-machining marks on their surface, using a small laser spot is not recommended. This is because the micro-machined surface features distort the reflected light from a small spot, resulting in a noisy or unstable measurement signal. One method to overcome this issue is to produce laser displacement sensors with a different spot geometry. Simply moving to a larger laser spot may overcome some of the issues described above, but this will have the adverse effect of reducing the sensor resolution. Also typically, laser sensors with a large spot have a large measuring range, which results in a larger linearity error.
Therefore some laser sensor manufacturers have created high resolution laser sensors with a short ‘laser line’ spot of just a few mm in width. Combined with specific software algorithms, this combination filters out extremely well any interference caused by surface roughness, defects, indentations or holes down in the sub-micrometre range, especially on polished metals.
In addition, these types of sensors are ideal for measuring on structured surfaces, where the distance to the surface and not the structure itself need to be measured, ie the distance measurement should not be influenced by the structure of the surface, but instead should provide a constant, reliable value of the distance from the target.
To achieve the highest performance in dynamic measurements, some suppliers offer pre-defined functions with their sensors for specific measurement conditions. These functions may include viewing and outputting of the video signal display, selection of individual signal peaks and region of interest (ROI). These functions allow interference signals from outside the ‘measurement area’ to be filtered out. Advanced filters such as advanced real time surface compensation, median, recursive, mean and ‘spike correction’ ensure that errors due to changes in surface colour, texture or geometry are reduced to a minimum.
Standard laser sensors normally operate with a simple mean or recursive averaging filter. While this may reduce the noise level of the output signal, it will also reduce the responsiveness of the sensor to changes in target distance or geometry. This type of filtering should be avoided in most dynamic applications, except in measurement of continuous products, eg strip products, where a long term trend is preferred over short term dynamic response.
Most suppliers can only offer sensors with a fixed measurement rate, which is selected by the manufacturer as a ‘best case fits all’. This means that for most surfaces, other than matt white, the measurement light exposure from the surface into the laser is not ideal, which results in measurement values that are generally noisier. The user then has to rely on averaging the measurement data to reduce the noise level or error on the output. This method will only result in poor data being averaged to a lower value; it does not give more accurate data. However, some suppliers use software algorithms to enable the laser sensor measurement rate to be adjusted for different target surfaces. For example, slowing the measurement rate down will result in much lower noise level on the measurement output. Then no or very little filtering is needed, resulting in a more accurate measurement, whilst maintaining a dynamic response from the laser sensor.
Faster measuring rates are not always best as the laser sensor accuracy can suffer on difficult-to-measure surfaces when using high speed measurement rates. A slower measurement rate allows more exposure time for low reflectivity surfaces such as matt black or shiny, mirrored objects.
Check out the technical datasheet and you may find that most suppliers of low cost laser sensors do not state the ‘temperature stability’ of their sensors. Quite often, the error in measurement due to a change in temperature is an order of magnitude greater than the sensor linearity over its full operating temperature range. So how do you know the actual measurement error or how to correct your results to account for this? A good rule of thumb is that if it is not stated in the technical literature, it will be quite high. Typically, measurement errors can be as high as 400-500ppm/K for laser sensors.
On the other hand, a supplier of high performance laser sensors is much more likely to state the temperature stability of a sensor on the datasheet. In addition, active temperature compensation algorithms may also be incorporated into the sensor, reducing temperature error to as low as 100ppm/K. A factor 5 smaller error will result in a measurement performance that is up to factor 5 greater.
Measurement output types
It is also important to consider the different output types available with laser displacement sensors. All suppliers will offer analogue, serial or digital output options and some can offer both from the same sensor. Analogue output sensors tend to have limited cable lengths (typically up to 10-20m) and can suffer from inherent noise levels due to electrical connections in the wiring circuit, which reduce the resolution of the output signal. However, serial output signals tend not to suffer as much degradation in resolution and can operate over longer cable lengths (up to 100m). Some sensor suppliers also offer digital outputs such as Ethernet and/or EtherCAT, which result in no reduction in resolution over virtually unlimited cable lengths. These laser sensors also operate with a dedicated IP address, enabling remote connection and configuration.
Synchronisation and triggering
For many applications, it is necessary to measure or acquire data simultaneously using multiple sensors. For example, a true synchronous measurement is required to precisely measure thickness or differential measurements of moving or oscillating objects. In this case, one laser displacement sensor acts as the ‘master’, which provides the corresponding cycle pulse for the second sensor (slave).
Triggering is also important, particularly in production and process automation environments where a permanent measurement output is often not the best approach. Initiating a measurement or a controlled output of measurement data will reduce the system load on any downstream monitoring and control units. Instead, the sensor should wait for a signal from an external source (eg encoder) that specifies the time for a measurement to take place, thus initiating the output of data.
Measurement of vibration
When it comes to measuring vibration or amplitude of vibration (which has a sinusoidal output), it is important to look for a sensor supplier that offers a ‘true analogue’ laser displacement sensor. This means a sensor that uses a PSD (photo sensitive detector) in a high speed analogue circuit, rather than a CCD- or CMOS-based sensor, which relies on a digital microprocessor to convert and output a measurement signal at a fixed measurement rate.
Almost all laser sensors today have been developed based on CCD or CMOS digital processors, which are not designed for measuring vibration. A true analogue laser sensor will give a much truer representation. On the technical datasheet, a key indicator in recognising a true analogue, PSD-based laser sensor is if the datasheet includes a “frequency response” statement, typically to a -3dB level rather than a “measurement speed/rate”.
For displacement measurement tasks where available space to mount the sensors is limited or restricted, the size of the sensor is critical. Non-contact laser displacement sensors are now available with very compact dimensions despite having a fully integrated controller in the laser sensor head.