CAN Newsletter magazine
To overcome CAN limitations, Marin successfully used the TCS-10 CAN switch from TK Engineering to split the CAN network into segments. This gave them the option to have a longer CAN, higher data rates, more flexible network topology and resistance to faults in the CAN network. A case study.
When Maritime Research Institute Netherlands built larger and more complex simulations and test facilities, they realized that they were limited by the size of the CAN network and by the length and speed of the CAN used to control the test equipment. To overcome these limitations, Marin successfully used the TCS-10 CAN switch from TK Engineering to split the CAN network into segments. This gave them the option to have a longer CAN, higher data rates, more flexible network topology, and resistance to faults in the CAN network.
Many Marin model tests are performed with electrical servo-powered ship models. In the most simple setup, for example a propulsion test, this involves only a propeller. Maneuvering and free sailing requires additional rudder servo motors. Other actuators such as fins, tunnel thrusters, winches, and podded propulsors are used as well, most of the time combined together. These servo drives and motors are typically developed by Marin itself to offer the best solution for a test. Most of the time of this equipment must be waterproof, it must have very little backlash but it also must be flexible to (de-)mount in any unique ship model.
The servo motors in the ship models are controlled by a software algorithm (auto pilot). This software algorithm provides set points for the propeller rpm and rudder angles based on the motions and position of the model. Position is measured by a 3D infrared camera system; motions are either a derivative or in most cases measured by a rate gyro. Each signal is recorded and handled by a PC running the auto pilot. The resulting set points are transferred via OPC to Marin software called BSS (Basic Steering System). BSS acts as an interface between the auto pilot and CAN. CAN is the Marin standard serial bus system for addressing servo drives. It gives us the advantage of a robust, high-speed bus system, which can be used over greater lengths in a rugged environment.
When CAN was first introduced at Marin in 1999, ship models were not as complex as they are today. Generally, throughout the last decade, an increase in actuators and desired bus speed can be seen. Today, models with 16 fast actuating servo motors are no exception in for example an offshore rig. This increased number of servo drives (or CAN nodes), and also the higher software processing speeds, required a higher bandwidth CAN to the point that the bandwidth of the CAN is becoming a bottleneck.
In Marin’s offshore basin both waves and wind can be generated. To simulate the latter, wind fans are used. These are all servo motor equipped fans with an integrated servo drive connected to CAN. This is where problems first arose. In order to create a homogeneous wind field, a battery with up to 55 wind fan servos can be used. In a model test a wind spectrum can be applied. This results in a dynamic wind field with for example sudden gusts as it would be in real life. Therefore, all wind fan servos continuously receive new set points.
In the past, all wind fan servos were connected to one long CAN network, with a PC acting as commander. In this setup the network load became critically high, sometimes over the edge of what is allowed. To further complicate things, if one servo drive would fail, as a result the entire network would fail, ultimately resulting in a complete breakdown. During a model test these breakdowns would create a lot of extra time pressure and costs. Other facilities also suffered from the high demands in performance, network load problems with free sailing models, but also distance vs. maximum bit rates (with 125 kbit/s as the Marin standard).
With CAN being a daisy chain, until recently there was no other option than connect one drive to another, with a terminator at the very beginning- and end of the network. This affected flexibility (which for Marin is very important with a changing setup for every test). In case of network load problems there was little more to do than lowering the network bit rate but this inconveniently interferes with the model test. In some cases, there even was an additional PC placed with its own CAN network and software couplings to relevant Marin systems as a quick fix. This however is very complex and time-consuming. In general, CAN at Marin provided a lot of possibilities, but as time went by the increasing demands left the system with room for improvement.
The solution was using the TKE CAN TCS-10 switch to split CAN into segments. Marin first learned from the switch entering the market in 2015. The product did seem to offer an ideal solution for the problems we were facing. At that very moment a large overhaul of the Offshore Basin was already in the planning, so a CAN upgrade could very well be combined with this overhaul.
News and reports