BECKHOFF ET1100 - Manual (Page 7)

General Issues
Slave Controller – Application Note FAQ 3
3 General Issues
3.1 Where can I find documentation updates?
Documentation updates are available for download at the BECKHOFF website:
http://www.beckhoff.de/english/download/ethercat_development_products.htm
3.2 ESC clock source accuracy: Is 25 ppm necessary?
Since standard PCs can be used as a master, the clock source deviation of the master’s Ethernet
PHY can be up to 100ppm. This is tolerated by the ESCs (according to IEEE 802.3) if the FIFO size is
default. Nevertheless, EtherCAT requires 25 ppm clock accuracy for the EtherCAT slaves in order to
enable FIFO size reduction.
FIFO size reduction is possible in any slave with high accuracy except for the first one directly
attached to the master. This can reduce network latency significantly, so, 25 ppm is mandatory.
Additionally, the clock accuracy has influence on the DC settling time and the forwarding latency jitter.
3.3 Why should port 0 never be an unused port?
Refer to the special features of port 0: If a frame travels through a slave with port 0 automatically
closed (because it is unused, the link state is fixed at no link), the Circulating Frame bit is set in the
frame. If this frame comes to a second device which has also port 0 automatically closed (because
unused), the frame will be destroyed.
So, any network containing more than one of the slaves with port 0 unused will experience impossible
communication. Additionally, in case of network trouble, masters might try to close ports in order to
isolate faulty network segments and to maintain communication in the rest of the network. If port 0 is
not used, the master is connected via port 1, 2 or 3. The master might try to close ports 1, 2, or 3
because it expects to be located at port 0, which results in network isolation.
Refer to the ESC datasheet section I for more information on port 0 special behaviour.
3.4 Link/Activity LEDs shows strange behavior
LED is on without anything attached, LED is off with other device attached. If all ports are connected,
communication with the slave is possible, although following slaves are not seen.
This happens if the LINK_MII polarity is incorrect, i.e. the link information is inverted for the ESC. The
communication with the slave is possible if all ports are connected, because the ESC detects that no
port has a link, and then it automatically opens port 0. You can check this in ESC DL status register
0x0110: no physical link at all ports (although you can obviously read through port 0).
3.5 Can slaves communicate without SII EEPROM / invalid SII EEPROM content?
Yes! EtherCAT communication is possible; all frames can be forwarded without a problem. Register
reading/writing is possible, but process data exchange is impossible. The ESC blocks Process Data
memory until the EEPROM is successfully loaded (typically no PDI is configured unless the SII
EEPROM is loaded successfully). You can do the initial programming of the SII EEPROM via
EtherCAT (e.g. for hundreds of slaves with
“empty” SII EEPROM in parallel (broadcast).
3.6 Do I need the complete XML ESI description for simple PDI read/write tests?
No, you do not need a complete ESI for PDI read/write tests. Only the first 8 words of the SII
EEPROM need to be written (e.g. via the network with values from the pinout configurator, the IP core
configurator, or the ESI EEPROM CRC calculator). This enables the PDI interface. Such a device
cannot be correctly identified by TwinCAT, but basic frame exchange (register view etc.) is possible.