From: Jack Correia <jcorreia@fiona.umsmed.edu>
  To  : rasmb@bbri.eri.harvard.edu
  Date: Sun, 28 May 95 12:14:20 -0500

Re: Time Out Problems

        In response to John Steckert's reported time out/lost data/scan 
hangups I had 5 replies.  Most of them simply agreed it occurred although it 
was mostly considered an annoyance of the XLA.

        1)  R_SERVO TIME OUT and Radial_Rate errors can be attributed to the 
        slit servo and replacing the servo probably will make them go away!

        2)  R_SERVO TIME OUT errors do not necessarily cause data to be lost.

        3)  Tom Laue identifies the cause of R_SERVO TIME OUT errors to 
noise         on the DAB (data aquisition board) exceeding the resolution of 
the         A/D used to determine the slit position.  A fix involving 
altering         the hardware on the DAB solves the problem.  Tom's 
breadborad design         will eventually make to the service people.  {You 
don't know what         experimental uncertainty is until Tom has the guts 
of a DA system         laying on the floor in front of you while your sample 
spins in the           model E!}

        4)  Lost data files is attributed by Tom to a serial communication   
      problem between the computer and the XLA.  Some clones use a shallow   
      buffer on the UART (?4-word) which does not send a proper hanshake     
    back to the XLA to stop sending.  This causes a buffer overflow at       
  the computer!  Replace your UART card one that has a deeper buffer         
and the problem will go away.  Some computers like Dell or IBM or            
Compaq are OK because of the hardware they use.           

                This actually suggests a new survey - who has lost data      
   problems and what computer do you use.  The case where a set of scans     
    is missing may be a timeout error where a longer time interval occurs    
     between a particular data scan, but all the scans requested R01 to      
   R20 (?) are present.  Here in MISS we have an IBM and never see lost      
   data, but occasionally see an apparent skipped time interval!

        5)  0.001 cm, 1 av.; 0.01, 4 av; 0.005, 5 av; 0.025, 1 av (get your  
       averaging in dcdt.  {What a surprize, no agreement among         
centrifugationists!}

        6)  A problem I have never heard of is described by Cait Macphee.  
If         not in the XLA controler when data is thrown to the computer the 
data         is lost!  The viewer being used at the time is Desqview.  I 
suggested         this is a foreground background allocation problem in 
windows.  We           routinely have Bo's ultrascan in the foreground so we 
can watch         aquisition, while the controler is on the screen, but in 
the         background.  Even for equilibrium runs overnight I bury both 
icons so         the screen saver comes on, and data aquisition is OK!


Thanks to all who responded!  I hope Tom's fixes fix all your troubles.

  
-------------------------------
Dr. Jack Correia              |
Dept. of Biochemistry         |
Univ. of Miss Medical Center  | 
2500 N. State St.             |
Jackson, MS, 39216            |
(601) 984-1522                | 
fax (601) 984-1501            |
jcorreia@fiona.umsmed.edu     |
-------------------------------