CAMAC Progress Report

                     Week Ending: August 16, 2002
                               Week: 81


Tom Folkers:

Tom Sargent:
     1. Bill and I went to the SMT on Monday to make sure things were 
        ready for the NSF site visit on Tuesday.  Monday night Bill 
        was able to excite the slew run-away problem which generates 
        floating point "nan" numbers.  This was when the sun was down, 
        so it wasn't using sun avoidance. So my earlier assumption 
        that it was an error in the sun avoidance code was wrong. The 
        code I had put in to detect the nans was in the sun avoidance 
        slew mode, so it wasn't triggered.   I then put more nan 
        detection code in the non-sun-avoid slew code.  Of course, I 
        couldn't make the error happen again after that.  
     2. I installed 3 more flat panel displays at the SMT so that 
        smtcont and smtoast computers have all new displays.
     3. We did several successful simulated "observations" of the comb 
        generator Tuesday.  The demo for the NSF sit visit worked OK.
     4. I looked at the nan problem when we got back. The fact that it 
        affects both sun avoid and non sun avoid slew modes simplifies 
        the problem. I suspected the velocity computation because it 
        is common to both algorithms.  I found that servo was 
        sometimes receiving multiple position reports from the DSP 
        which all had the same time stamp.  This would certainly 
        produce strange velocity results.  It turns out the DSP code 
        was written so that if it falls behind in making position 
        reports, it catches up by doing several at the same time.  
        This was changed to have it report an error if it misses 
        generating a report but to do no more than one report per 
        reporting period.  I haven't seen two reports at the same time 
        stamp since this change.
     5. While at the SMT we discovered that the sluggish track 
        algorithm behavior was fixed at lower elevations but it could 
        still happen at high  elevations. I increased the range over 
        which the correction is applied. This still needs to be 
        tested. 
     6. I had some trouble running wine (the Windows emulator used to 
        compile and link code for the DSP) on smtocont. It doesn't 
        want to be run remotely. I installed it on smtxdpy2 in the lab 
        and can now use it there.
     7. I began looking at the velocity computation in detail to see 
        how accurate it is.  It seems to have some noise and spikes in 
        it.  

Pete Hubbard:
     1. Finish spare differential driver and receiver boards.
     2. Update the drawings for servo and tracker  ADC,DAC,DIO 
        interface chassis.
     3. Working with Gene H. on SAMBUS problem.

David Forbes: