The computer control system for the 1m consists of the following computers:
The current method of commmunication between the comoputers is via commands over sockets. On the DOS machine, this is enabled using networking software from NetManage, their PC/TCP Software Development Kit. This allows for socket communication, which was implemented circa 2004. There is a set of 5 manuals, and software comes on a set of 3.5 floppies and some CDROMs. The serial and key number can be found in the LaTeX source of this manual.
The old method of communication between the computers was via disk files that are seen by each computer on an NFS mounted disk on control1m. In fact, the software for telescope and guider PCs are also located on the control1m disk. Most computers used PC-NFS Version 5.1; the installation disks can be found in the 1m control room; the PI computer was still on version 4.0. Note that if one wants to change the license of PC-NFS, the only way is to delete the existing version and reinstall.
The power to the computers (and several other things) is controlled by a network device that has 16 outlets that can be switched on and off via network commands. There is a separate 8 outlet network power strip located on the azimuth disk.
The APO network uses the following:
As of December 2010, APO switched the 1m computers over to a local network, with addresses obtained via DHCP.
A new telescope computer was ordered March 2006 from Industrial Computers Inc. This contains a backplane with 5 ISA and 7 PCI slots and a single board (PICMG slot) computer with integrated display, Ethernet, etc, plus an 80 Gby disk, CDROM, and floppy drive. The SBC is an LGA 775 with Pentium 4 processor. Ethernet is via an Intel 82562EZ 10/100. It uses DDR2/533 memory, currently a 256Mby SIMM.
Installation of DOS on this computer proceeded as follows: DOS installed
using standard 3 MS-DOS installation disks. After installation, computer
hangs on boot while loading HIMEM.SYS. Reboot from floppy and replace
HIMEM.SYS with Windows version (usr/local/1m/dosfiles). Network software
was switched from PCNFS to PCTCP to allow socket communication. To get
network to work, drivers for 82562EZ (Intel PRO 100) were downloaded
from Intel (PRODOS.EXE); this file didn't execute on DOS and needed
to be unpacked on a Windows machine. This included both ODI and NDIS
drivers. I could get the ODI driver to load (LSL, E100BODI, IPXODI),
but was unable to get the PCTCP packet driver interface (
PCTCPODIPKT)
to work with it. I then installed the NDIS driver. First, I did a PCNFS
installation, but of course the new device is not in the installation
list. I then modified LANMANPROTOCOL.INI and put in a section for
the new device driver (E100B.DOS). The driver loaded successfully on
bootup. Commented out PCNFS initialization. Finally, I redid the PCTCP
installation and it saw the NDIS device. I used frame type DIX-ETHERNET
and PCTCP kernel worked!
During 2007-2010 we would get sporadic hangups of the telescope computer, sometimes going a long time without one, and sometimes having multiple hangups in rapid succession. In April 2010, the entire single board computer, including CPU and memory was replaced to see if this makes any difference. The replacement single board computer was bought from IPC Hammer and was an FS-979 (also possible known as PG-7791); it differs slightly from the original in that it has two different Ethernet controllers. A Pentium 4 CPU was purchased from MemoryLabs.
Unfortunately, this didn't seem to make a difference for the hangs. I then modified the software to update the telescope status less frequently ( 4 seconds), and this appears to have reduced the hangs significantly, although not eliminated them. Subsequently, I suspect that the culprit is the motor control card (PC38-6E); see information in the motor control section about the replacement of this card.
In July 2014, the computer started to freeze/crash shortly after reboots. I took the card out, and discovered that the two big fans in the computer case were not attached to power, so I connected them. I went back to the old SBC because I thought a fan on the board wasn't working, but it was the same with the old SBC (not the CPU fan but a smaller one on the board). I left the old SBC in there, so the FS-979 is back in the in the drawer in the office; not clear if there could be a problem with it or not! Note that when I connected the new card, I had to change the weather station to COM2 (in sysscf.new).
Linux (Redhat Enterprise 4-U1) was also installed on this computer for possible convenience.
The ISA slots are populated by:
The old telescope computer was an Industrial Computer Source
(tech support and customer service 1-800-480-0044,
sales 1-888-294-4558, original sales order 144151, 7/9/97) 200 Mhz Pentium
PC (7000-8MB chassis, motherboard 586MBH200, floppy drive, RAM, KB3
with touchpad, 2" CPU fan, power supplies Turbo-Cool 300 Slim with P8/P9
power connectors); three of these computers were ordered, one
for the telescope, one for the guider, and one spare. The operating
system is DOS. The program it runs is called TOCC.EXE and it lives on
E:
BIN
TOCC.EXE.
The telescope computer has a 166 Mhz processor, a 1.2Gbyte disk, and 8 Mbytes RAM. It has eight (??) ISA slots. These are populated by: video card (SIIG VGA), 3c509 networking card, PC38-6E (SN 902) motor controller (long card that controls telescope alt/az/rot and 3 focus axes), PC34 motor controller (controls dome), PC38-6 (SN 4053) card for filter wheel and guider, Technology 80 5312 encoder card (long card used with dome encoder), and a Computer Optical Products CP4016 motor/encoder controller purchased for use with rotator encoder. The Tech 80 card was used for the dome encoder, but in July 2005 we had problems and switched over to using the CP4016 card instead. A weather station is plugged into a serial port (COM1), and a voltage meter used for measuring the PI CCD temperature is plugged into the parallel port.
The networking is done using PC-NFS. The PC-NFS configuration is as follows, as reported by the NET ALL command:
This computer, eyeball, runs Linux (RedHat 9). It was purchased from Elite PC, and is a 600 Mhz Pentium III with 46Mby (!!) memory and a 80 Gbyte disk drive.
The control computer, command1m, runs Linux (RedHat Enterprise 4). It is a Dell Dimension 5150. See the LaTex source for this manual for account information.
On RHEL4, sometimes on the virtual desktop, the window manager stops allowing windows to be resized or moved; if this occurs, kwin -replace seems to fix it!
The camera control computer, ccd1m, runs Linux (RedHat Enterprise). It is located in the APO computer room. It communicates with the dome via a fiber link using a fiber repeater purchased from Icron. This computer controls the Finger Lakes guider camera (USB), three USB webcams, and the Leach camera (custom card). See the LaTex source for this manual for account information.
The Leach controller requires a kernel module to be loaded. This is compiled for a specific kernel version (2.4), so significant attention must be paid if the system were to be upgraded. In addition, the 2.4 kernel required some modifications to allow control of the Quickcam 4000 Pro webcams.
Power is controlled by two network controllable power switches.
The first commandable power controller is an APC MasterSwitch, part number AP9210, serial number wa9720838756, Ethernet address 00:c0:b7:92:01:5a; this has 8 plugs. It is statically configured as power1m.apo.nmsu.edu. It can be accessed via a serial port, via the WWW, or via SNMP. The regular usage is via SNMP; four machines can access it this way, and these machines are control1m, eyeball, tycho, and tocc1m. To access via serial (only necessary if the WWW password has been forgotten), one needs a female 9-pin serial connector cable. This cable should have a null model pin-swap. The Linux program minicom can be used to communicate over serial port, using /dev/ttyS0.
The second commandable power controller is an APC ..., Ethernet address 00:c0:b7:76:96:79; this has 16 plugs. It is statically configured as spare1m.apo.nmsu.edu. This has a serial port cable included with the device; serial communication as above.
To access via http, one needs a username and password. These are set the same as the observe account on control1m except only 8 characters are used for the password. In desperate situations, there is a backdoor password that was supplied by APC. See the APC manual for this device or LaTeX source for this manual for access information.
The ccd1m computer, located in the APO control room, is also on a controllable power switch.
Photpower (powerc1m) is a controllable power switch for the high speed photometer.
WE NOW NO LONGER HAVE A SEPARATE GUIDER COMPUTER. The guider stages are controlled by the telescope computer, and the camera is controlled by the ccd1m computer.
The guider computer was an Industrial Computer Source, identical to the system for the telescope computer
described above. The program it runs is called GCS.EXE
and it lives on E:
BIN
GCS.EXE.
The guider computer has a 200 Mhz processor, a 1.2 Gbyte disk, and 8 Mbytes RAM. Its ISA slots have an SIIG video card, a 3c509 network card, a long card used to communicate with the Spectrasource guide camera, and a long PC38 card for the guider stage and filter wheel control.
The networking is done using PC-NFS. The PC-NFS configuration is as follows, as reported by the NET ALL command:
The spare computer is an Industrial Computer Source, identical to the systems above. It has a 200 Mhz processor, 1.2 Gbyte disk, 64 Mbyte RAM, a CDROM, and a 3c509 network card.
The spare is no longer functional. The old control computer, loki, may serve as an adequate spare.
The PI control computer wass a Dell 486 running Window for Workgroups.