Pepperl+Fuchs Blog

How Long Can an AS-Interface Network Really Be?

Posted by Helge Hornis on Wed, Jan 29, 2014

Ed. note: This article is co-authored by Fabio Di Bernardo ( Fabio is a Lead Development Engineer at Pepperl+Fuchs (Mannheim, Germany) and is responsible for the development of AS-Interface field devices and Safety at Work components.

Download this blog: How Long Can an AS-Interface Network Really Be? (PDF, 345 KB)

That depends! Personally, we are not fans of vague or evasive answers to this question, but curiously, the answer is true. And it is one of the more common questions concerning AS-Interface that engineers ask when they scan the Internet in hopes of getting help. What's really unfortunate, though, is that you will find many incomplete and even wrong answers. In this article, we will try once and for all to cover all length-related aspects of AS-Interface that users need to know about when designing their systems. Let's start with the simple cases and then build on them:

Is there really a 100 m limit?

Yes, there is such a limit.  What this means is that any AS-Interface system designed according to specification must have no more than 100 m of total cable. Figure 1 shows such a setup composed of a gateway, an AS-Interface power supply, and a few modules.

  • The '100 m total' means that when adding up all the cable that carries AS-Interface signals, their cumulative length will be 100 m or less.
  • It does not mean the longest path (in this case from the gateway to the module with address 7), and it also does not mean to include the length of the various cables used to connect sensors to the I/O modules; these are not used to transmit AS-Interface signals.

An AS-Interface system has one master

Figure 1: Any AS-Interface system must have exactly one master.  The master can be part of a gateway to an upper-level network (shown here) or a PLC-specific backplane scanner card. This network has one segment and must, therefore, have exactly one AS-Interface power supply.  When calculating the length of the segment, all cable that carries AS-Interface signals must be included. I/O cables between the modules and sensors and actuators do not carry AS-Interface signals and, therefore, do not count towards the 100 m limit.  Passive adapters used to connect AS-Interface devices with an M12 connection must also be included.

What is a segment and why does it matter?

Actually, the 100 m limit does not apply to the network but to a segment. It's important to understand the difference between segments and networks.  AS-Interface networks are composed of segments and segments need to obey -- at least for now -- the 100 m limit rule. Figure 2 shows a network made up of three segments.  Note the following:

  • The network has exactly one gateway (or scanner card)
  • Two segments are connected using a repeater
  • Each segment needs to be powered; thus the three power supplies

A three-segment network

Figure 2: A three-segment network can be up to 300 m long.  The network must have exactly one gateway or scanner card.  The segments are connected using AS-Interface repeaters.  Each segment must have exactly one AS-Interface power supply.  The power supplies can be located anywhere along the segment. In practice, the installation location of a power supply depends more on the location of existing cabinets, the availability of AC power, and general accessibility. In situations where the minimal voltage drop across the network is of importance, locating the power supply near the middle of the network is ideal.  Certain advanced gateways offer additional diagnostics functions (specifically, duplicate address detection on the first segment) when the power supply is connected to the gateway as shown here.  The first segment of the network is then powered through the gateway.   

As a consequence of this, the total AS-Interface cable length is 300 m.  This is a good time to mention one of those half-truths floating around.  Looking at some literature, web pages, and articles, you may find statements like "the maximum length of an AS-Interface network is 300 m and one gets there by stringing three segments together back to back." Nothing could be further from the truth—even though competing automation technologies keep feeding this myth and, we suppose, wish it to actually be true.  Let's look at this in detail.

How many segments can be interconnected?

AS-Interface is a deterministic, real-time solution designed to update I/O data quickly.  The repeater is a device that galvanically isolates two segments and allows data to propagate from one segment to the next.  Doing this takes a little bit of time, and in order to be deterministic, the maximum allowable time for a full data exchange, i.e., the time it takes for the gateway to send a packet to a module, plus the time it takes for the module to respond, plus the time it takes for those signals to propagate along the cable, must be limited.  When a repeater is used, the propagation time is extended by the time it takes the repeater to do its job.  This is the reason why the specification allows no more than two repeaters to delay the total propagation time.  In other words, no modules must be located more than "two repeaters" away from the gateway.  Stringing more repeaters back-to-back simply adds too much delay.  Using this two jump rule, we can see that constructing larger networks is easily accomplished.  Figure 3 shows a setup where five segments are combined to provide a total linear network length of 500 m.

Placing the gateway in the middle segment

Figure 3:  Placing the gateway in the middle segment makes it possible to construct this 5-segment network.

Note that in order to satisfy the two jump rule, the gateway is placed in the middle segment.  The 100 m limit still applies to each segment, but by correctly combining multiple segments, even longer networks can be constructed.  Figure 4 shows a valid star network with 900 m total length.  With this in mind, we get the following:

  • The network has exactly one gateway ( or scanner card)
  • Any number of segments can be connected using repeaters
  • A segment has a maximum length of 100 m
  • Any module on the network must be within two jumps from the gateway

Star-shaped topology

Figure 4:  Following the "no module can be more than two repeaters from the gateway" rule allows even larger networks to be designed.  In this case, a star-shaped topology resulting in 900 m of total network. 

How strict are those rules?

So far we've followed the specification to the letter.  But what if the application cannot be solved within those limits, or, if during the installation someone did not pay attention and violated the 100 m segment rule?  Can those rules be bent and still result in a network that operates reliably?  Addressing this question is where the "that depends" answer comes in.  AS-Interface offers exceptional installation flexibility as long as the 100 m segment limit is observed.

  • A segment is not terminated and can have any topology
  • The modules can be located anywhere along the segment
  • Any combination of modules can be used
  • A segment can have up to 31 modules (or even 62 when using newer extended address modules)

Conversely, if this level of flexibility can be restricted, it may not be necessary to strictly adhere to the 100 m limit.  For instance, if the topology is essentially linear and the number of modules within the segment is reduced, the total length of the segment can be increased.  How much additional length depends on a number of other factors, including the distribution of the modules and if those modules represent more of a capacitive or inductive load.  As you can see, bending the rules comes with its own set of issues, and most of these small modifications are not worth the trouble as they yield only a relatively small additional segment length.  Fortunately, with a bit of planning it is possible to successfully design segments that are much longer, allowing linear network lengths of 1000 m.

What does a terminator do?

AS-Interface, according to the specification, is an unterminated network allowing a segment to have any topology.  Figure 5 shows a segment where branches come off branches of branches.  Note that such a topology has many "open cable ends."  In most communication systems, open cable ends are a problem, as signals get reflected back into the cable, which in turn leads to constructive or destructive interference.  This is the reason why most networking technologies require some sort of termination at the cable end, limiting their topology to be more or less linear.

Ed. Note -  Ethernet is a strictly linear, point-to-point network. Using switches allows the implementation of other topologies, but then the connections from an end-device to a port on the switch is still linear.  Modern Ethernet hardware goes as far as offering a switch in the end-device itself.  This allows line structures and even rings to be constructed.  PROFINET, EtherNet/IP, and Modbus/TCP are based on standard Ethernet and follow the same rules.   

DeviceNet is a trunk-line/drop-line solution with limitations in terms of the maximum length of a single drop and limitations in terms of the cumulative length of all drops.  

PROFIBUS is strictly a linear network.  This is very important to know when using preconfigured cables with cable tees.  While a tee is allowed on the device side it must have essentially zero length.

AS-Interface allows highly branched networks

Figure 5: AS-Interface allows highly branched networks.  While not many initial layouts look like this, allowing such topologies makes adding last minute actuators and sensors possible.

For AS-Interface segments that are up to 100 m long, reflections do not cause problems, but once the length exceeds this limit the negative effects of reflections start building up.  Just like with any other network technology, a terminator can be used, allowing the total segment to extend past the specification limit.  The following must be considered:

  • Using an AS-Interface terminator, segments can be up to 200 m long
  • These segments must be essentially linear
  • Only one terminator must be used per segment
  • The terminator must be located as far from the AS-Interface power supply as possible
  • If modules are mounted using drop-tabs, the drops need to be short (2 m max, less is better), and there must be only one module per drop (this is what essentially linear means)

It is, of course, still possible to combine multiple segments.  Placing the gateway in the middle segment allows a total of 1000 m network length when aligned linearly and 1800 m when used on a star-shaped network, a far cry from the 100 m some people think is the limit for AS-Interface (see Figure 6).  When using a terminator, it is important to know that it must be "as far from the power supply as possible."  With a linear 200 m segment, this means the power supply ends up on one cable end, while the terminator is located at the other end. 

Using terminators on linear networks

Figure 6: Using terminators on essentially linear networks allows segments to be up to 200 m long.  This linear network has a total length of 600 m while a star-shaped configuration can be 1800 m long.  Note the locations of the power supplies.  For the terminator to work, it must be on the opposite side of the segment with respect to the power supply. While it is generally a good idea to consider the voltage drop along the network, doing this is critical when working with 200 m segments.

What is an advanced repeater?

The AS-Interface specification demands a repeater to offer a minimum level of performance.  Specifically, repeating a signal must be possible in 7 microseconds.  This demand is a direct result of the desired performance of the system and takes into account factors like response delays and propagation time along the network.  In fact, the specification requires that a slave response, i.e., the data packet sent by a module in response to addresses by the gateway, must never be delayed by more than 10.5 bit times. 

Read this blog post for a more in-depth discussion on the concept of bit time.

  • As it turns out, a bit time on AS-Interface is 6 microseconds.  If a module is two repeaters away from the gateway, the repeaters are responsible for 28 microseconds or 4.7 bit times response delay.
  • It is clear that no module can process a data packet from the gateway infinitely fast.  Here, the specification allows the modules between 1.5 bit times and 5 bit times of response time.
  • We also need to consider the speed with which electrical signals propagate along the cable.  This speed depends on the construction of the cable, and for AS-Interface flat cable, it is approximately 70 % the speed of light in vacuum.  Expressed in bit time, it takes 1/12 bit time to travel 100 m.   When traveling to and from a module, that cable must be traveled twice, 100 m AS-Interface corresponds to 1/6 bit time.

If we add up the worst-case times for a 300 m network, 4.7 bit times to travel across two repeaters, 5 bit times for the module to respond, and 1/2 bit time to propagate the 300 m  of network, we see that we are using up almost the full 10.5 bit times the master will allow.  Now imagine a linear network that uses terminators to allow each segment to be 200 m long.  This adds another 1/2 bit time in propagation, possibly resulting in responses that will be too late.  The solution to this problem is a faster repeater.  The advanced repeater reduces the delay time to 4.5 microseconds, such that the two repeaters needed to combine three segments will delay the signals by just 3 bit times.  While the added speed of an advanced repeater generally increases the speed of the network, using them in cases where the segment length is over 100 m is mandatory.

What about voltage drop?

Up to this point, we have discussed segment length only from the point of view of data communication across the network.  But AS-Interface is a data and power solution where the modules draw part of their operating current from the network.  This current draw results in voltage drop. Before deciding on setting up a 200 m segment, it is a good idea to check if the available operating voltage at each location where a module is mounted is still sufficient.  In cases where the modules draw a lot of current, the real limit may actually be the operating voltage; possibly long before the 100 m limit has been met. 

Take the following network as an example. Fifty extended addressing, 4-input modules are mounted 3 m apart on a linear network for a total length of 147 m between the first and the last module. This should be possible using a terminator, right?  Let's check.  Each module has 4 inductive sensors attached.  A quick measurement shows that the total current consumption per module (including the activated sensors) is 100 mA.  The total current that the AS-Interface power supply needs to make available is well below the 8 amp limit of the power supply used in this case.  Looks good so far. 

Now let's take a look at the voltage drop along the network. Using the cross section of the AS-Interface cable (1.5 mm2) and the primary voltage of the AS-Interface power supply (30 V), Ohm's law tells us that the last module will have an operating voltage of 21.8 V.  This may not be enough to power the module, and it is absolutely necessary to check with the manufacturer first.  If the module does not operate at this voltage, a repeater is necessary. If you don't feel like making these calculations by hand, contact us, and we will send you an Excel file that will do these for you.  Alternatively, consider downloading our AS-Interface Power Calculator.

Is a ring segment possible?

Most users considering ring topologies are interested in the redundancy this topology brings.  But there is another reason why a ring may be a useful solution.  Electrically speaking, a 200 m ring is similar to a 100 m linear network!  Think about it.  A signal that travels the ring will arrive back at the gateway after having traveled across 200 m of cable.  The same is true for a signal that goes 100 m on a linear network and is then reflected back at the open network end.  From a voltage drop point of view, things look even better as the shorter of the two cable paths to a module is always less than 100 m.  In the simple case of the module located at the vertex of the ring, each path is 100 m, and as a consequence, the equivalent cable resistance is 1/2 that of the linear 100 m cable.  The real advantage of the ring comes in when it is possible to replace a setup with many drops that would be 150 m, 170 m, or even 200 m long.  Figure 7 shows such a scenario.

AS-Interface terminator network

Figure 7: The terminator requires an essentially linear network. For that reason this branched layout with a 150 m segment length is not a suitable candidate for using the terminator.  Ideally, this network is split into two shorter segments connected through a repeater.  In situations where that is not possible, the modules may be connected by a ring.  Since the 200 m ring is electrically similar to a 100 m linear network, a terminator or repeater is not needed.

The highly branched nature of this network is not a good situation for a terminator. Connecting the module on a ring removes all reflection-generating network ends and removes the need for the terminator while keeping the electrical characteristics of the 100 m setup. 

When using rings, it is important to keep in mind that a ring must never contain a repeater, but it is always possible to combine multiple rings using repeaters.  To get a better visual of what I mean by that, consider watching this short AS-Interface Segments and Networks video. Using repeaters to combine five ring segments allows you to set up a network with 1200 m total network length.

How about shielding?

AS-Interface was designed to work reliably with the unshielded flat cable.  Using common wiring practices, a network can be expected to work reliably in all common industrial environments.  Tens of thousands of networks installed worldwide prove that AS-Interface easily deals with the noise levels found in typical installations.  But, what if the noise levels are exceptionally high?  Or, perhaps the cable can simply not be routed with sufficient separation from a high-voltage, high-noise cable?  In those cases, shielding, either by using shielded cable or by placing the AS-Interface cable inside properly grounded conduit may be considered. 

Before talking about the drawbacks of using shielded cable, let us mention that in all the years we have been working with AS-Interface, there has been only one installation where the electromagnetic noise level was so high that shielding was required.  Shielding the AS-Interface is the absolute last resort after all possible sources for problems have been evaluated and all other available methods dealing with noise have been exhausted. 

The reason is that shielding reduces the maximum possible segment length by 30 %.  Shielding the cable will increase the capacitance of the cable, and the only way to compensate for this is to limit the maximum segment length to 70 m.  Using shielded cable has many other installation-related drawbacks.  For instance, shielded cable is round and the installation advantages that come with piercing the flat cable are gone, increasing the installation cost and time.

Hope for the best and plan for the worst

In practice, terminators work very well and give engineers additional options.  Still, when bending (or rather breaking) the specification design rules, it is advisable to plan well and be prepared.  One of the best tools to have available when setting up an out-of-specification network is the AS-Interface Analyzer.  The Analyzer is the ideal tool to get a quantitative measurement of how well the network performs.  Any industrial communication system must be designed to deal with the loss of data.  And AS-Interface is no different.  The powerful built-in retry methods work hard to get data successfully across the network, giving the technology the necessary reserves to operate reliably in tough industrial environments.  But what if an overly aggressive layout uses up 75 % of those reserves?  It is a fact that once installed in the plant, the environment where these systems operate only gets worse over time.  Additional equipment and cabling installed later add to the electromagnetic noise level, requiring the gateways and modules to use additional operating reserves.  Hoping for the best is okay as long as one is prepared for the worst.  Here are a few things that can be done if the design was simply too aggressive:

  • Reduce the segment length (not necessarily all the way back to 100 m) and use a repeater
  • Evaluate if the network cable can be routed along a different, less noisy, path
  • Subdivide the network into two smaller networks using a dual-network gateway
  • Set up ring-shaped segments

 Learn more about AS-Interface by subscribing to TechTalk, our weekly podcast:

Subscribe to TechTalk

Topics: AS-Interface, Ask an Expert

Subscribe to Email Updates

Search our blog!

Loading Google Custom Search.....

Follow Me