Position sensors are key components in controlling a surgical robot’s motion. Select the right position sensor and the robot arm will move safely, smoothly, accurately and reliably. Get it wrong and things can go badly – with potential impacts on safety, reliability and your company’s reputation.
This short paper gives you 5 tips from an engineer who spent most of his 35 year career in the position sensor business:-
1. Be sure you understand the safety requirements
The impact of sensor failure on safety will be the biggest factor in how the overall position sensing or motion control system is designed. It is helpful to think of two types of sensor failure – the first is loss of operation (no signal output) and the second is incorrect operation (incorrect signal output). The first is usually less problematic than the second and is typically characterised by the position sensor outputting no signal or an error signal. The second is more difficult as the sensor appears to be operating correctly but is outputting an incorrect signal. For each type of failure it’s also helpful to think of the potential impact in 3 ways – not safety related, safety related or safety critical. In the first instance failure has no impact on safety; in the second, the failure may contribute to an unsafe condition and in the third, sensor failure will almost certainly contribute to an unsafe condition. In some safety related applications an acceptable level of independence can be achieved by some form of built-in test – for example, checking that an actuator’s position sensor output is within sensible limits from other system data. In some safety related and all safety critical arrangements, there will need to be some element of redundancy – in other words, at least two independent sensor outputs, whereby robot motion will only proceed if sensor outputs tally within defined limits. Naturally, fitting two sensors in to your design will have a big impact so be sure to know whether you’ll need a redundant approach from the start.
Figure 1 – example of an electrically duplex position sensor.
2. Be clear about what measurement performance is really needed
A common mistake is to specify a position sensor with too high a measurement performance or, conversely, with too low a performance. In the first instance the cost will be excessively high and in the second the robot’s performance will never meet requirements. Understanding what is really required is key and it is here that you’ll need to understand measurement terminology with regards to resolution, repeatability and accuracy. Resolution is the smallest change in position sensor that a sensor can detect; repeatability (or precision) is the degree to which a sensor’s reading repeats at the same true position and accuracy is the veracity of the sensor’s output to actual position. In other words, accuracy – sometime referred to as linearity – is the greatest error between a sensor’s output and true position. A common fallacy is that the greater the sensor’s resolution, the more accurate it is. A sensor with 16-bit measurement resolution can be more accurate than a lower, 14-bit resolution. Unlike many robot applications, robotic surgery typically requires high resolution and high repeatability (for precise, smooth motion over small distances) with only modest accuracy.
A useful tool in understanding measurement performance is an error budget. This is a list of all contributory factors and their effects on measurement performance – of which sensor measurement performance is only one. Other sources might include the sensor’s temperature coefficient (accuracy is usually stated at 25C); sensor installation tolerances; differential thermal expansion of the host components; backlash or lost motion in the host structure etc. Backlash is often a dominant factor and often leads to a requirement to measure directly the component of interest rather than calculating or inferring its position from say a gearbox or coupling.
3. Don’t forget cables & connectors
Cables and connectors are the number 1 source of sensor reliability issues and this can derive from EMC problems, excessive cable flexing or vibration, accidental damage during assembly, maintenance or operation. Design drawings will often show a nicely located position sensor but the associated cables and connectors have been left out for later consideration. This can be a mistake as cables and connectors need to have sufficient space to allow easy and efficient assembly; should be secured so as not to unnecessarily flex or move and should be routed so as to avoid obvious EMC sources such as motors and transformers.
4. Design for minimum cost
Shaving a few bucks off the cost of a position sensor is great but there are much bigger numbers to consider. Firstly, be aware of a positions sensor’s installation tolerances. These are the mechanical limits that the sensor needs for it to perform as the manufacturer claims and watch out – these tolerances are often only shown in the datasheet’s small print. It may be the case that a low cost sensor with tight installation tolerances will, overall, be more costly than a higher cost sensor with generous installation tolerances. Secondly, and more importantly, is the cost of field failure. One of my favourite design rules is the Rule Of Ten which states that at each stage a problem moves undiscovered during development, production or actual service – the higher (typically 10x) the costs for eliminating it. Systematic or frequent filed failure of a position sensor can rapidly cause an expensive, nightmare scenario. The few bucks shaved off a position sensor will appear miniscule to the bill that accrues from repairing and modifying equipment at customer premises – not to mention loss of reputation. It pays not to skimp on selecting reliable, good quality position sensors for your design.
5. Match sensor technology to the application
Position sensors have been around for a century or so and in that time man has pretty much used every physical phenomenon to measure position. Each of these approaches has their own strengths and weaknesses. The most common position sensor technologies in surgical robotics are optical, magnetic and inductive sensors. Optical are typically the most accurate but in unpackaged formats can require tight installation tolerances to meet the manufacturer’s spec and they are usually unsuited to wet or dirty environments. Magnetic and inductive sensors are more robust than optical but lack the measurement performance.