[etherlab-users] etherlab-users Digest, Vol 104, Issue 1

David Page dave.page at gleeble.com
Wed Feb 10 18:11:10 CET 2016 

ET1100 Section I.3 Frame Processing:

It is not possible to use unmanaged switches between these ESCs or between
master and the first
slave, because source and destination MAC addresses are not evaluated or
exchanged by the ESCs.
Only the source MAC address is modified when using the default settings, so
outgoing and incoming
frames can be distinguished by the master.
NOTE: Attaching an ESC directly to an office network will result in network
flooding, since the ESC will reflect any
frame – especially broadcast frames – back into the network (broadcast storm

If ESC DL Control register bit 0 is clear, then all frames are forwarded.

$ sudo ethercat -p0 reg_read 0x100 4
0x00 0x00 0x07 0x00

On my hardware bit 0 is clear after configuration by the etherlab master.

Clearly under certain configurations, it is possible to use an unmanaged
switch. It is also quite possible to use a managed switch to perform some
layer 2 routing. For a beginner, it is probably best to just add another
ethernet adapter.

    Best regards - Dave



On Mon, Feb 1, 2016 at 6:31 PM, Gavin Lambert <gavinl at compacsort.com> wrote:

> It is possible to have the master PC connect via standard switch to a
> standard Ethernet network, one connection of which is the entire EtherCAT
> slave network.  (It is *not* possible to have standard Ethernet devices
> interleaved with EtherCAT devices, at least not without dedicated bridge
> slaves designed for this purpose.)
>
>
>
> This does work, including with Etherlab – I sometimes have a switch
> between the master and the slaves in order to easily insert another PC
> running Wireshark to examine the exchanges – but it comes with some caveats.
>
>
>
> 1.       The EtherCAT messages are broadcasts at the MAC level, so will
> cause unnecessary traffic on other parts of the network.
>
> 2.       Also due to the above, you cannot have two such masters or slave
> groups on one network, even if each master is intended to talk to different
> slave sub-networks.  The base EtherCAT protocol does not contain any
> routing data, so if you need to support this case you’ll have to use a
> higher-level protocol such as ADS – which as David notes is not supported
> by Etherlab at present.
>
> 3.       If the switch is processing other packets in a store-and-forward
> manner (which is common) then the non-realtime traffic may adversely affect
> your realtime performance (especially if there’s a lot of it) and the
> master may think the packets are getting lost or mismatched.
>
> 4.       The most recommended way to operate the Etherlab master is with
> the customised drivers, which bypass the regular TCP/IP stack.  This means
> that the master PC won’t be able to send or receive any non-EtherCAT
> packets anyway.  It’s possible that this limit doesn’t apply when using the
> generic driver (I haven’t checked), but this comes with lower performance.
> (That may not matter to you, depending on your application and the desired
> cycle times and tolerance to latency or data loss.)
>
> 5.       If the first slave is disconnected or powered off, the master
> will still see this as a link-up network (due to the switch) and will start
> flooding the syslogs with missing packet errors instead of simply logging a
> disconnected state.
>
> 6.       The first slave will report errors when it receives a
> non-EtherCAT packet (it will not forward these on to the rest of the
> EtherCAT network).  Typically you won’t notice this at all unless you are
> reading the slaves’ error counter registers.
>
>
>
> *From:* David Page
> *Sent:* Tuesday, 2 February 2016 02:39
> *To:* etherlab-users at etherlab.org
> *Subject:* Re: [etherlab-users] etherlab-users Digest, Vol 104, Issue 1
>
>
>
> Hi Paul,
>
>
>
> The basic story is the EtherCAT slave hardware modifies the Ethernet frame
> as it passes through each node, and then returns the modified frame back to
> the host  -- essentially a token passing version of Ethernet. As such, the
> use of standard hubs or switches will result in a packet storm. The bus
> design assumes there is exactly one "normal" Ethernet host (at the master).
> The slave hardware does support a UDP encapsulated mode which is useful for
> hosts which cannot send and receive raw frames, but the above constraint
> still applies.
>
>
>
> It is possible with TwinCAT to talk over UDP (EtherCAT ADS protocol) to a
> remote EtherCAT bridge (e.g. Beckhoff BK9000) which then has another port
> dedicated to EtherCAT. The Etherlab master does not support ADS, though.
>
>
>
>
> https://infosys.beckhoff.com/english.php?content=../content/1033/tcadscommon/html/tcadscommon_intro.htm&id=
>
> https://www.beckhoff.com/english.asp?bus_terminal/bk9000_bk9050.htm
>
>
>
>
>
>     - Dave Page
>
>
>
>
>
>
>
>
>
>
>
> On Mon, Feb 1, 2016 at 6:00 AM, <etherlab-users-request at etherlab.org>
> wrote:
>
> Send etherlab-users mailing list submissions to
>         etherlab-users at etherlab.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         http://lists.etherlab.org/mailman/listinfo/etherlab-users
> or, via email, send a message with subject or body 'help' to
>         etherlab-users-request at etherlab.org
>
> You can reach the person managing the list at
>         etherlab-users-owner at etherlab.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of etherlab-users digest..."
>
>
> Today's Topics:
>
>    1. Communicating with Ethercat over TCP/IP network (Paul Mulligan)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Sun, 31 Jan 2016 14:59:32 +0000 (UTC)
> From: Paul Mulligan <mulligan252 at yahoo.ie>
> To: "etherlab-users at etherlab.org" <etherlab-users at etherlab.org>
> Subject: [etherlab-users] Communicating with Ethercat over TCP/IP
>         network
> Message-ID:
>         <25812475.4467849.1454252372172.JavaMail.yahoo at mail.yahoo.com>
> Content-Type: text/plain; charset="utf-8"
>
> Hi,
> I've read that Ethercat packets can be sent over a network by packing them
> into UDP datagrams. Is it therefore possible to connect the Ethercat bus
> with slaves to a network and have the master controller pc not directly
> connected to the slaves but connected to the same network and hence
> transfer the Ethercat data to and from the slaves over the network? I
> assume the timing performance will be worse, but how is this achieved?
> Thanks
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <
> http://lists.etherlab.org/pipermail/etherlab-users/attachments/20160131/1b83406c/attachment-0001.html
> >
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
> etherlab-users mailing list
> etherlab-users at etherlab.org
> http://lists.etherlab.org/mailman/listinfo/etherlab-users
>
>
> ------------------------------
>
> End of etherlab-users Digest, Vol 104, Issue 1
> **********************************************
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20160210/177ad75f/attachment-0004.htm>

More information about the Etherlab-users mailing list