Fall 2001 R. Millan-Gabet. Excerpts from IOTA Engineering Run Notes ---------------------------------------- Oct. 02 ------- - discovered that LD power supplies (switching) introduce huge spikes (~100du) in the PICNIC signal. They also grind some internet connections in the lab to a halt! They will be replaced by non-switching power supplies, and corresponding motors. Oct. 21 ------- - new LD supplies & motors have been installed for LD1. Oct. 22 ------- - LD testing: Tested both LD1 and LD2 using Quadra (DelayLine.Config needs to be modified to address one LD or the other from the Quadra). Aligned HP metrology laser on LD2 Both work, in servo and absolute modes. H and -L limits also OK on both. Far limits not tested, but broken according to MGL. Linux server software: /root/srv1 and /root/srv2 work, although MGL reported erratic behaviour. LD2 still has switching supply, and slows down even UDP communications enormously. Supply will be replaced soon. LD2 index always shows up as "->>>>>>>>>". We don't care though, since OT will control LDs soon. /home/ldelay/additionals/additionals/ldelay.script seems to work, but seems to require proper stopping of services before starting. * This is an example of output when it did not work: Starting ldelay services: kernel CPU speed is 233000864 Hz Running on CPU=586, FPU=1 (AT6400 OS does not seem to fully load) rmmod: module at6400 is not loaded Using /lib/modules/2.4.2-2/misc/at6400.o rmmod: module hp10885a is not loaded Using /lib/modules/2.4.2-2/misc/hp10885a.o [root@ldelay ldelay]# Opening connection to hardware drivers... Couldn't connect to /dev/at6400_0! * And this is an example of output when it did work: Starting ldelay services: kernel CPU speed is 233000864 Hz Running on CPU=586, FPU=1 (AT6400 OS loads fully) rmmod: module at6400 is not loaded Using /lib/modules/2.4.2-2/misc/at6400.o rmmod: module hp10885a is not loaded Using /lib/modules/2.4.2-2/misc/hp10885a.o [root@ldelay ldelay]# Opening connection to hardware drivers... Allocating socket ... Got socket! Binding socket ... Oct. 23 ------- - VxWorks LD control: Pete working on it. Succeded in establishing UDP communication between VxWorks and Linux server. Linux server is now in local net. There are now a bunch of temporary hub connections that should be cleaned up/organized once this is all sorted out. note: MGL has now replaced the switching power supply for LD2. - Misc: EP realized that the network connection between our different machines might be wrong: usually requires 50 Ohm termination at one end, and grounded 50 ohm termination at the other, which we don't do. According to web docs, this could cause severe slow downs in the net connection. Oct. 24 ------- - Matrix AO board: Since it was observed that the AO outputs oscillate, possibly due long flat cable (non-twisted pairs, unshielded) from Matrix board to interface, we did the following: move interface and piezo transmitter box on top of VME rack, with very short flat cable from Matrix to interface, and long BNCs from there to Physik piezo amplifiers, hoping that the better properties of a transmission line will prevent the oscillation. Unfortunately this did not help at all! We still get big 0-10V oscillation of the AO voltage outputs (channels 6 and 7), unless the output is terminated with 200ohm load, and large 20mV noise. We will consult with Matrix on this tomorrow. Note: currently channels 6 and 7 seem to be configured for 0-5V output range. We think this should be configured for 0-10V output, since that is what the Physik amplifiers we will use need for 0-120um piezo stroke. - DAQ: We tested whether the DAQ_Scheduler really stores scan data to disk when instructed to do so, by running the IDL readout code (setting save_data->TRUE): it works! We verified this by writing an IDL test program to read and plot the data from disk. Aside from that, we further tested that the FPGA readout generally works by running the 6-pixels readout and moving a well focused light spot from a fiber tip vertically on the array, and verifying that signal appears in the plotted scans as expected. Oct. 25 ------- - Matrix AO board: Re-configured channels 6 & 7 for 0-10V output (on board#1 of 2) Verified that board is configured to local sensing, as it should Today's conversation with Matrix (now Thales) tech support not very useful, the expert guy was not in and will call tomorrow. - CCDs and electronics moved a bit to new location to make room for classical 3-beam combiner components. WAT working on setting up classical 3-beam combiner. WAT also improving interferometer for telescope alignment. Star-tracked on 3 telescopes (using OT) therefore alignment to CCDs is good again. - FPS wrote SUN shell scripts to start everything that needs to be started to run the VxWorks/OT automatically. FPS has improved the OT star tracking GUIs, to make them more intuitive to use. FPS wrote OT panel to activate the telescopes TV-cameras. FPS wrote VxWorks/OT control of LDs. Linux server moved to network containing iota17 (SUN) and VxWorks. No longer accesible from computers hooked to our other (external) network. Its IP address is now: 192.168.51.18 (but from iota17 it is also known as ldelay). - PICNIC FPGA readout essentially complete and bug-free (only averaging of loops and reads is still buggy). Next will focus on readout details, noise characterization and noise reduction. We have also transfered the EP experimental code into the official locations and now works with the Oct2001 workspace, and are working on providing useful IDL display routines based on building block functions that perform the actual low-level readout tasks. note: /home/iota/iotamc/vxworks/daq/ was archived in: /home/iota/iotamc/vxworks/daq/Archive/ and its contents replaced with: /home/epedrett/scan/ public location of new IDL DAQ programs is: /home/iota/Idlstuff/DAQ_Oct2001 Oct. 26 ------- - MGL talked to tech support about Matrix AO board oscillation problem: they are not surprised at all, their board is not meant to drive a long cable! We must do something about this such as matching impedances using series resistors, or adding a video amplifier. Oct. 27 ------- - NICMOS: In preparation for FLUOR's arrival, I tested the NICMOS system, at the level of reading a signal fed to the ADC card. It works fine. Hopefully nothing blew inside the dewar this summer and they are all set. - Star tracker VxWorks/OT: FPS fixed problem of oscillation of star on CCD. FPS implemented star searching using piezo mirrors (instead of siderostats). Oct. 28 ------- - FPS has written anti-collision code for SDs Attempts to test this were frustrated by latest disaster: two of the hoses feeding air to the SD carts (sides for SD1 and SD2) have been chewed up (one split in half, the other has 9 big holes). Best explanation so far is that mouse is responsible for this. Until this is fixed, SD carts cannot float or move. - WAT and SR work on telescope B focus. Looking through alignment telescope, star image much improved: goes through about 1 arcsec focus uniformly, with no coma or astigmatic shape. Oct 29 ------ - Backed up iota17:/home/iota/iotamc/ to /home/iota/Backups/iotamc.2001Oct29.tar.gz Also put backup in /jmonnier/data/ CfA account. - MGL replaced SD air tubing damaged by mouse. - At JDM's request, given the difference in transmission between telescopes A and B observed in Oct2001 visible/fiber run, I looked at the inscriptions on the side of the windows at the lab end of the pipes: beam: Fixed: #2 infrasil, xxx has a small crack at the edge oriented backwards (vacuum side, air side) compared to label! (I re-mounted it according to label, turned it around) beam: Delay1: #10 infrasil 301 (contrary to CaF2 label on pipe) air, vacuum orientation: OK beam: Delay2: #3 infrasil 301 lambda/4 MgF2 air, vacumm orientation: OK - PICNIC: Adjusted pre-amp bias to better match detector range to ADC range. Started at VB3 = 3.86V Ended at VB3 = 3.78V: reset=+5V, saturation=-6V: OK