Protostar Vision Box – Part 2: Sensor suite

Welcome back 👋


This is the second part of our series of blog posts on Protostar Vision Box. Last week, we showcased the hardware and lightning design. Today, we are going to discuss the sensor suite, how to choose the right sensors for the job and how we implement them in Protostar Vision Box.


Let’s dive right in!



Choosing the right sensor for the job is often a difficult task and the market is flooded with different types of sensors, sizes, protection classes etc. but if chosen right a good sensor can reduce development costs, increase efficiency and reduce rate of errors that occur during production which in turn saves money for the buyer.


This post will mostly be focusing on optical sensors since they are the ones most often used in production lines along with level and proximity sensors which we will cover in future. Apart from the sensors mentioned above we also have temperature sensors, flow sensors, pressure sensors.


Often while searching for sensors apart from the type of sensor you will see terms PNP and NPN in sensors product page. These terms refer to the type of switching transistor used in the output of the sensor. The choice between PNP (Positive-Negative-Positive) and NPN (Negative-Positive-Negative) is determined by the nature of the system where the sensor is installed. See Figure 1 for a wiring example.



PNP Sensors – In a PNP sensor, when the sensor is not triggered, the output transistor is ‘OFF’ and the signal line is effectively disconnected from the supply. When the sensor is triggered, the transistor switches ‘ON’ and connects the signal line to the supply. This is often referred to as a “source” because current is sourced from the sensor to the load.


NPN Sensors – Conversely, in an NPN sensor, when the sensor is not triggered, the output transistor is ‘OFF’ and the signal line is effectively disconnected from the ground. When the sensor is triggered, the transistor switches ‘ON’ and connects the signal line to the ground. This is often referred to as a “sink” because current is sunk from the load into the sensor.





Figure 1. Example of wiring for PNP and NPN sensors


Choice between the NPN and PNP depends on the hardware you are using to receive the signal from the sensors.


In addition to the switching type sensors can have different operating modes. Most sensors have the option between Light-On mode and Dark-On mode. In the Light-On mode sensor output turns on when the receiver receives the beam, while in Dark-On mode the output turns on when the receiver is not receiving the beam.


Now that we covered the basics let’s start with the photoelectric sensors. Photoelectric sensors work by emitting a light beam from transmitter to receiver. Transmitter is usually a LED or laser diode that produces a specific light wavelength, receiver is typically a photodiode or phototransistor that identifies changes in light. Electronic circuit typically consists of an amplifier to boost receiver signal, a comparator to contrast the signal to a predetermined threshold and an output generating stage for generating a signal that can be utilized by different devices like PLCs. Some sensors also include extra features like background suppression, time-delayed response and adjustable sensitivity/output type. There are 3 categories of photoelectric sensors that define the way they work. We have Through-beam Sensors, Retro-reflective Sensors and Diffuse-reflective sensors.


Through-beam Sensors come in two parts, there is a transmitter and there is a receiver which is placed opposite to the transmitter. Transmitter projects a laser into the receiver and when the laser is interrupted we get a signal. Main advantage of using Through-beam sensors is their long operating distance no matter the object type. Drawback is the increased price because you need 2 different packages and you have 1 additional wire coming to your controller/plc. See Figure 3 for block diagram.



Figure 2.  Overview of different sensors



Figure 3. Block diagram of through-beam sensor


Retro-reflective Sensors combine transmitter and emitter units into a singular package which makes them easier to install and manipulate. Insides are similar to the through-beam sensors but on the other side now there’s a reflector that bounces emitted light back towards the sensor. When an object is present, the amount of reflected light is reduced, triggering the sensor. Main advantage over the through-beam sensor is a singular package with just a slightly reduced range, their price is also lower even when accounting for the reflector. Most of the issues are caused by incorrect positioning of the sensor in relation to the reflector and environmental conditions like vibrations. In a try to mitigate some of those issues most of the sensors come with features such as automatic gain control or polarizing filters to improve accuracy and reliability. See Figure 4 for block diagram.




Figure 4. Block diagram of Retro-reflective sensors


For the last of the photoelectric sensors there are diffuse sensors. For diffuse sensors light is scattered and returned to the receiver by the object being detected rather than by a reflector. Their main advantage is simplicity and flexibility because they don’t require any external reflector or receiver component. They can also be mounted vertically if needed and their detection range can often be adjusted which allows an additional degree of freedom. However their range is significantly lower than the previously mentioned sensors and they are more sensitive to ambient light and object surface. Same as for retro-reflective sensors some of the issues can be reduced by methods such as background suppression,polarization filters and multi-frequency light emission.


In Figure 5 we see a setup containing multiple sensors of multiple types. Specifically we see Leuze branded retro-reflective sensor which in this case is used for bottle cap detection, Leuze branded through-beam sensors for liquid level detection + 2 more retro reflective sensors on the left side at various heights of the bottle. Using an array of multiple sensors allows us flexibility while tracking the bottle and significantly reduces error rates.




Figure 5. We can see a small stepper rail with attached through-beam sensor and 3 retro-reflective sensors.


We are also using diffuse sensors in situations where bottle heights vary and we are not able to automatically adjust the sensor position using motors. Examples of which can be found in Figure 6 and Figure 7 this is a cost-effective and efficient method to track products of different sizes without the need for mechanical adjustments.




Figure 6. Diffuses sensor at the start of the reject mechanism.




Figure 7. Diffuse sensors tracking a bottle



To summarize the photoelectric sensors part, generally it’s recommended to use Through-beam sensors for long distance applications where we still need precision and additional wiring is not a problem. Retro-reflective sensors offer similar accuracy and are a great choice for detection of both large and small objects in cases where budget is limited and/or there exists a wiring constraint. Diffuse sensors are a great choice in applications where flexibility of sensor placement is limited,requirement for range is reduced and we aren’t doing small object detection. Some real world applications for through-beam and retro-reflective sensors would be bottle cap detection on bottles and diffuse sensors could be used for counting the bottles.


That’s it for this week, next week we will be talking about using AI for defect detection and how we trained our own model.  Stay tuned! 😄