|
|
LAN-XI – One Cable Data Acquisition Propelled by Power over Ethernet (PoE) and Precision Timing Protocol (PTP)
By Lars T. Kroman, Karl Kristian Lundsgaard & Erik Ziegler, Brüel & Kjær, Denmark
06 Jun 2008
Cabling multichannel measurement systems has always been a challenge. Running transducer cables, power cables, synchronisation cables and data transfer cables from control room to measurement position is difficult, time consuming and often a potential source of error and added noise. In fact, for most multichannel measurements the time spent on setting up the system is far longer than the actual measurement time! Being able to acquire data from many channels, synchronise measurements and power the front-end via just one cable, dramatically reduces the cost of system setup. We have achieved this with our LAN-XI Data Acquisition System using standard LAN with Power over Ethernet (PoE) and Precision Timing Protocol (PTP). This article describes the main technical aspects of the two technologies.
|
Technology
|
|
|
PoE - Power over Ethernet | |
Old fashioned telephones were powered via telephone lines. This type of technology was also implemented to work on a Local Area Network (LAN) to power, for example, IP telephones, CCTV and Webcams. Power over Ethernet is a technology that allows power to be transmitted along with data on a standard LAN cable in an Ethernet network. The technology is specified in the IEEE 802.3af standard from 2003 and updated in the IEEE 802.3-2005 standard.
PoE basics
A PoE system consists of 3 main units:
- A special power supply
- Standard LAN cable
- The device that needs the power
The special power supply – called Power Sourcing Equipment (PSE) – can either be a LAN switch/hub or an in-line power injector.
LAN can be implemented with cables using either two or four pairs of twisted wires, with
or without shielding and with different levels of speed requirements. All PSE equipment must
take this into consideration when applying power to the cable. Providing power over a
long, high-speed LAN cable does, however, require a high-quality Ethernet cable, so you
should always use a CAT 5e or CAT 6 cable in order to avoid problems.
The power source injects between 36 V and 57 V – usually 48 V – into the cable and must be able to provide 15.4 W (max. 400 mA).
With a maximum cable length of 100 m (330 ft), the available power for the powered device is 12.95 W when accounting for cable loss.
 | | Enlarge image | | Fig. 1: Example of PoE “communication” between a power source and a powered device |
|
PoE “Communication”
When the power source – a LAN switch or a Power Injector – is turned on it starts a detection cycle in order to find out if there is a need for power (see Fig. 1 on page 20). If you have both a PoE switch/hub and a Power Injector present on the same LAN-cable, only one of them will turn on the power. The device that wants to be powered indicates this by placing a specified “signature resistance” (19 – 26.5 kΩ) and a “signature capacitance” (150 nF) between the powering pairs. This is recognised by the power source and power is turned on.
In order to prevent overloading as well as powering open lines, the Power Sourcing Equipment will turn off the power if the resistance between the powered pair is less than 15 Ω or greater than 33 kΩ or the capacitance is greater than 10 μF.
Power classes
The PoE standard specifies a set of different power classes:
0 : 0.44 W to 12.95 W
1 : 0.44 W to 3.84 W
2 : 3.84 W to 6.49 W
3 : 6.49 W to 12.95 W
4 : Reserved for future use
Power class 0 is the default, 1 – 3 are optional and class 4 is currently not allowed. By using the power classes, a Powered Device can indicate to the Power Sourcing Equipment how much power it requires (see Fig. 1).
The communication about the power class is done in a manner similar to that described for turning on the power.
If the powered device requires more power than the Power Sourcing Equipment is able to supply, the power to this line is disabled and the detection cycle restarted.
Intelligent power management
As can be seen from the above description of the main aspects of Power over Ethernet, this is an intelligent way of distributing power via a local area network. The obvious benefits of having just one cable to the device is supplemented by the intelligent power safety features built into the products complying with the IEEE 802.3af standard.
Available PoE sources
PoE Power Sourcing Equipment complying with IEEE 802.3af is available from the regular suppliers of network products in versions from 1 port injectors over 8 port desktop switches/hubs to 12 – 48 port switches/hubs for rack mount covering any measurement setup from one stand-alone LAN-XI unit over distributed measurement systems up to covering complete production line test cells. PTP – Precision Timing Protocol
Accurate timing has always been important in network applications – just think of how awkward it is to talk on a telephone when there is a delay on the line. Large networks have, for many years, used a protocol called NTP – Network Timing Protocol – providing the ability to synchronise clocks within milliseconds. With the development of a new standard protocol, – Precision Timing Protocol IEEE 1588 or simply PTP – specifying synchronisation with sub microsecond accuracy, it is now possible to synchronise multichannel measurements via a LAN network.
PTP basics
The fundamental of a PTP system is that a set of clocks present on the network are organised in a hierarchy of Master and Slaves. The Masterclock sends out synchronisation messages to all the other clocks – the Slaves – on the network and the Slaves then adjust their clocks accordingly. The Master clock is normally the most accurate clock present on the network and a special algorithm, “The best Master clock algorithm”, is used by all the clocks to determine which of the present clocks should be the Master clock.
When a new clock becomes present on the network it initially starts to listen for synchronisation messages. If the new clock does not receive any synchronisation message within a specified time, the new clock will assume the role of Master clock and start to send out synchronisation messages.
When a Master clock receives a synchronisation message from a clock with a better accuracy than its own, it will pass over the role as Masterclock to the better clock and become a slave.
 | |
Principle of synchronisation
The Master clock continuously sends out synchronisation message to the slaves, for example, every second (Fig. 2). The outbound synchronisation messages are time-stamped leaving the Master clock. When the Slave clock receives the synchronisation messages they are time-stamped as well. The Slave can use the first synchronisation message to adjust its local time depending on the time difference between the Master and the Slave.
Master and Slave clocks are now adjusted tothe same time difference from an offset related to the transmission time of the synchronisation messages between the Master and the Slaves. After receiving the first synchronisation message and adjusting its time, the Slave will randomly send a time-stamped Delay Request message to the Master clock that replies with a Delay Response message containing the Delay Request message time-stamp. Based on the four available time-stamps, the Slave can calculate the time it takes for a message to travel between the Master and the Slave. It is assumed that the travel time is the same in both directions of the network. This simplified description does not take oscillator errors into account. A simple servo implementation will handle that.
As all clocks drift, the clock adjust process is carried out on a regular but infrequent basis. This allows clocks to stay synchronised and at the same time helps to minimise “clock adjust traffic”.
In order to avoid the variable delay caused by the software in a communication protocol, the time stamping is implemented in the hardware of a PTP device. This is one of the major differences to NTP (Network Timing Protocol) that allows sub microsecond synchronisation.
PTP is easy to use
The PTP protocol is independent of the network topology and it is “self-adjusting” to actual system setup in terms of selecting the most accurate clock and adapting to the actual delay in the network. This makes it very easy to set up a measurement system using PTP to synchronise multiple front-ends.
 | | Enlarge image | | Fig. 3: Typical Phase deviation using a standard LAN switch |
|
Phase accuracy
Switches used in standard LAN do not include any special features to support the Precision Timing Protocol but special PTP switches are available from manufacturers of network backbone components.
As all synchronisation is done by the PTP devices on the network, a measurement system with PTP complying front-ends can operate on a standard network. In order to get reliable measurements from multiple channels, the Sample and Phase synchronisation is of great importance. The typical Phase deviation, measured at 25.6 kHz in a network using a standard LAN switch and PTP synchronisation, is less than 1 degree – more than enough for most sound and vibration measurements (Fig. 3).
PTP technology allows special PTP switches where PTP traffic gets preference on the network and is, therefore, routed through the switch with the smallest possible delay. Hence PTP technology accommodates future demands for better sample synchronisation and lower phase deviation in measurement application. One-cable data acquisition! | | Enlarge image | | Fig. 4: Simple Test Cell setup with LAN-XI front-end |
|
By implementing both PoE and PTP in the LAN-XI measurement front-end, we have made it possible to set up distributed measurement systems using standard LAN with obvious benefits such as:
- Reduced noise and cable cost due to shorter transducer cables
- Simple system cabling resulting in dramatically lower setup cost and less errors
- LAN switches replace expensive Patch Panels
- No need for extra power outlets
Read the latest industry news
|
|
