Appendix B: TiCkS in the control system
Mail from CC, 20181015
Theres is one way to connect to the UCTS through UART (mini USB connector on the bottom of UCTS board) with minicom for instance. Shell command ang gui informations are explained in WR PTP cores user’s manual (Chapter 4 Using WRPC shell).
Mail from LAT, 20181211
A MOS server has been developed for LST/NectarCam (by UCM, Madrid, with modifications by CEA-Saclay)
Installed on tcs01, running a service called uctsd.service.
It should be listening at 10.1.4.1:48010
The XML file for the ucts configuration should be
- ::
<?xml version=’1.0’ encoding=’UTF-8’?> <UCTS>
<PC_IP_ADDRESS value=”10.1.7.250”/> <UCTS_IP_ADDRESS value=”10.1.4.4”/> <PC_MAC_ADDRESS value=”90:1B:0E:D2:3F:C7”/>
</UCTS>
UAExpert can be used to perform tests remotely
Mail from LAT 20180913
TIB NETWORK AND SLOW CONTROL
The TIB has to be connected to the slow control switch. It will ask for an IP for the MAC address 28:27:EB:6C:E1:E3 to the DHCP server. The DHCP server should give the IP 10.1.4.64. Then, you should be able to access the servers running in the TIB computer: 10.1.4.64:22 for SSH, 10.1.4.64:48010 for the OPC-UA.
This interface of the TIB is also used to send TIB timestamps to the EVB.
The TIB slow control works as long as it has power and network connection. It doesn’t matter if CLK and PPS are being distributed or not. However the read values can be absurd.
UCTS NETWORK AND SLOW CONTROL
The UCTS only has one interface to the outside world, and this is the optical fiber to the WR switch. For LST-1 there are only 3 physical elements in the WR network: the UCTS, the WR switch and a computer. The computer must have an ethernet interface in the WR network (i.e. connected to the WR switch) with the IP 10.1.11.250. From this interface, the computer runs three pieces of software:
A DHCP server for the WR network, in charge of providing the IP 10.1.8.4 to the MAC address 22:33:09:1e:54:69 (the UCTS).
An OPC-UA server which sends the low level slow control commands to the UCTS through the WR interface, while exposing the OPC-UA datapoints to the outside world, to be accessed by CaCo.
The CDTS server, in charge of receiving the timestamps generated by the UCTS, process them, and provides them to the EVB. If the EVB is also running in Osaka, the data can be delivered by any method internal to the computer (Julien Houles knows the details)
The computer configured to do these 3 functions at IFAE tests was Osaka. Being more specific, we used Osaka EM2, with MAC 90:E2:BA:86:25:65, assigning to this interface the IP 10.1.11.250. Osaka was the machine with one interface in the WR network and other interfaces in the other networks.
Mail from LAT 20181015
In order to connect to UCTS slow control, you had to connect to any Osaka’s interface but EM2.
In order to check the fiber, just power on the TIB and the UCTS and connect to the TIB OPC-UA slow control (10.1.4.64:48010) with CaCo or any generic OPC-UA client. Then, you can check several datapoints:
Alarmvector/PPSFailure: If it is “0”, it means that the clock and the PPS are reaching the TIB properly.
Whatever trigger rate in Rates datapoint: if you see reasonable values, it means that the clock and PPS are arriving properly.
Mail from MP/AV 20181022
Hi Taka, I am pretty sure that all ports are configured equal so you can choose either switch port to insert the SFP and use it for timing distribution. At least that was the configuration as we sent it to Madrid.
Just try it out.. worst thing which can happen is that the slave won’t lock.
Best regards, Achim
Hi Taka,
I add Achim to this list, since he provided you the switch, and can maybe comment on the configuration. I think in the latest versions the non-White-Rabbit connection (i.e. the “copper-SFP” which allows an RJ45 LAN ethernet connection) can be in any port. Achim could comment if he has made any specific configurations (VLAN, etc)…
Since the switch should be configured so that none of the ports are in “slave” mode (i.e., get their timing from an upstream WRS), then they should just try to send timing information along with ethernet on all the ports. So a non-white-rabbit connection should fall back to standard ethernet. I would think that it would “just work”, to switch the LAN connection, then.
Here is a FAQ answer on copper SFPs: https://www.ohwr.org/projects/white-rabbit/wiki/faqwr#Q-Can-I-use-copper-SFPs-in-WR-network-for-connecting-nodes-that-dont-require-WR-timing.
You can find a user-guide to the switch here, for example:creotech.pl/wp-content/uploads/2015/08/WRuserguide.pdf <http://creotech.pl/wp-content/uploads/2015/08/WRuserguide.pdf>
As described on page 9 and 16, you can try to log in to the White Rabbit Switch’s Web Management Interface (using a browser). That should tell you what the current configuration of the ports is:
where the ports will be identified with symbols as follows:
The interface allows you to change the endpoint mode
Endpoint Mode: Modify the mode for each port (wr master/wr slave/auto/none)
But as I said, I would hope this is not needed.