The robot's movements are uncoordinated. What went wrong?
Click on the blue text to enter Analog Devices , then click " Set as Star " in the upper right corner~
Industrial robotics and machine tool applications may involve the precise coordination of multiple axes of movement within a specific space to complete the task at hand. Robots typically have six axes that must be coordinated, and sometimes there are seven axes if the robot moves along a track.
Long press to enter the ADI Think Tank applet to learn more about ADI industrial technology~
In CNC machining, 5-axis coordination is common, but some applications use up to 12 axes, where the tool and workpiece move relative to each other in a specific space. Each axis consists of a servo drive, a motor, and sometimes a gearbox between the motor and the axis joint or end effector. The system is then interconnected via Industrial Ethernet, generally using a LINE topology, as shown in Figure 1. The motor controller converts the required spatial trajectory into a single position reference required for each servo axis, which is then transmitted cyclically on the network.
Fig. 1. Network topology of a multi-axis machine tool.
These applications run at a defined cycle time, which is typically equal to, or a multiple of, the fundamental control/pulse width modulation (PWM) switching period of the underlying servo motor drive. In this environment, shown in Figure 2, end-to-end network transmission latency is an important parameter. During each cycle, the motor controller must transmit the new position reference and other relevant information to each node in Figure 1. Then, there needs to be enough time left in the PWM cycle for each node to update the servo control algorithm calculations with the new position reference and any new sensor data. Each node then applies the updated PWM vector to the servo drive at the same point in time through a distributed clock mechanism that relies on the Industrial Ethernet protocol. Depending on the specific control architecture, part of the control loop algorithm can be implemented in the PLC, if sufficient time is required to implement after any relevant sensor information updates are received on the network.
Figure 2. PWM period and network transmission time.
Assuming the only traffic on the network is the periodic data flow between the machine tool controller and the servo nodes, the network latency (TNW) is determined by the number of network hops to the farthest node, the network data rate, and the delay experienced by each node. When using robots and machine tools, the signal transmission delay caused by the wire can be ignored because the cable length is generally relatively short. The main delay is bandwidth latency; that is, the time required to transmit the data to the wire. For the smallest Ethernet frame (generally suitable for machine tool and robot control), refer to Figure 3 for bandwidth latency for 100 Mbps and 1 Gbps bit rates. This is equal to packet size/data rate. For a multi-axis system, a typical data payload from the controller to the servo consists of a 4-byte speed/position reference update for each servo and a 1-byte controller update, which is 30 bytes for a 6-axis robot. Of course, some applications contain more information in the update and/or have more axes, in which case the packet size is larger than the minimum size.
Figure 3. Bandwidth delay for minimum length Ethernet frames.
In addition to bandwidth delays, other delay elements are introduced as the Ethernet frame moves through the PHY and dual-port switch that serve each network interface. These delays are shown in Figures 4 and 5, which show the portion of the frame moving through the PHY into the MAC (1-2), where only the preamble and destination portions of the frame need to be timed as the destination address is analyzed. Path 2-3a shows the interception of the payload data at the current node, while path 2-3b shows the frame's journey to the destination node. Figure 4a shows only the payload transmitted to the application in 2-3a, while Figure 4b shows the majority of the frame being transmitted; this illustrates that there may be subtle differences between Ethernet protocols. Path 3b-4 shows the outbound transmission of the frame, through the transmit queue, through the PHY, and then back to the cable. This path does not exist in the line end nodes shown in the figure. This assumes cut-through packet switching, rather than store-and-forward, which has higher latency because the entire frame is accounted for in the switch and then forwarded.
Figure 4. Frame delay: (a) dual-port mode frame delay and (b) line terminal node.
Figure 5 shows the latency elements of a frame as a timeline, which depicts the total transmission time of a frame through one axis node. T BW represents the bandwidth delay and T L_1node represents the delay of the frame through a single node. In addition to the delays associated with the physical transmission of bits across the wire and accounting for address bits for destination address analysis, PHY and switch component delays are other factors that affect transmission delays within the system. As the bit rate on the wire increases and the number of nodes increases, these delays have a greater impact on the overall end-to-end frame transmission delay.
Figure 5. Frame transmission timeline.
Analog Devices has introduced two new Industrial Ethernet PHYs designed to operate reliably in harsh industrial conditions over a wider ambient temperature range (up to 105°C) with excellent power and latency specifications. The ADIN1300 and ADIN1200 are designed to address the challenges mentioned in this article, making them ideal for industrial applications. With the fido5000 real-time Ethernet, multi-protocol embedded dual-port switch, Analog Devices has developed a solution for deterministic time-sensitive applications.
Table 1 lists the delays introduced by the PHY and switches, assuming that the receive buffer analysis is based on the destination address and assuming a 100 Mbps network.
Table 1. PHY and switch delays
For example, factoring these delays into a line network of up to 7 axes and factoring in the total payload into the final node (3a in Figure 4), the total transmission delay becomes
The 58 × 80 ns represents the remaining 58 bytes of payload after the leading and target address bytes are read.
This calculation assumes that there is no other traffic on the network or that the network is able to prioritize time-sensitive traffic. It is somewhat protocol dependent and the calculated values will vary slightly depending on the specific Industrial Ethernet protocol used. Recalling Figure 2, when reducing the cycle time of a mechanical system to 50 µs to 100 µs, the transmission of the frame to the farthest node can take up nearly 50% of the entire cycle, resulting in less time available to update the motor control and motion control algorithm calculations in the next cycle. Minimizing this transmission time is important for optimizing performance because it allows for longer and more complex control calculations. Given that the latency associated with wire data is fixed and bit rate dependent, using low latency components such as the ADIN1200 PHY and fido5000 embedded switches will be key to optimizing performance, especially as the number of nodes increases (for example, a 12-axis CNC machine) and the cycle time decreases. Moving to Gigabit Ethernet can significantly reduce the impact of bandwidth latency, but will increase the proportion of the overall latency caused by the switch and PHY components. For example, the network transmission latency for a 12-axis CNC machine using a Gigabit network is approximately 7.5 µs. In this case, the bandwidth element is negligible and using the minimum or maximum Ethernet frame size does not make any difference. Network latency is roughly evenly split between the PHY and the switch, and the value of minimizing latency in these elements is highlighted as industrial systems move to Gigabit speeds, control cycle times decrease (WEtherCAT ® shows a cycle time of 12.5 µs), node counts increase as Ethernet-connected sensors are added to control networks, and network topologies continue to flatten.
In high-performance multi-axis synchronized motion applications, control timing requirements are very precise, deterministic, and time-critical, requiring the minimum end-to-end latency, especially as control cycle times decrease and the complexity of control algorithms increases. Low latency PHYs and embedded pass-through switches are important components for optimizing these systems. To address the challenges described in this article, ADI has introduced two new robust Industrial Ethernet PHYs, the ADIN1300 (10 Mb/100 Mb/1 Gb) and the ADIN1200 (10 Mb/100 Mb).
ADIN1200
-
10BASE-Te/100BASE-TX IEEE ® 802.3 ™ compliant
-
MII, RMII, and RGMII MAC interfaces
-
100BASE-TX RGMII latency transmit: <124 ns, receive <250 ns
-
100BASE-TX MII latency transmit: <52 ns, receive <248 ns
-
EMC test standards
-
IEC 61000-4-5 surge (±4 kV)
-
IEC 61000-4-4 Electrical fast transient (EFT) (±4 kV)
-
IEC 61000-4-6 Conducted Immunity (10 V)
-
EN55032 Electromagnetic radiation disturbance (Class A)
-
EN55032 Conducted Emission (Class A)
-
Unmanaged Configuration Using Multi-Level Pin Bonding
-
EEE complies with IEEE 802.3az standard
-
Packet start detection supports IEEE 1588 timestamp
-
Enhanced Link Detection
-
Configurable LEDs
-
Crystal/clock input: 25 MHz
-
25 MHz/125 MHz synchronous clock output
-
Small package and wide temperature range
-
5 mm × 5 mm, 32-lead LFCSP
-
Specified operating temperature range: -40°C to +105°C and −40°C to 85°C
-
Low power consumption
-
139 mW (for 100BASE-TX)
-
MAC interface VDDIO power supply voltage: 3.3 V/2.5 V/1.8 V
-
Single-supply operation using 3.3 V VDDIO
-
Integrated power supply monitoring and POR
京公网安备 11010802033920号