Pf insects

From DIDEAS Wiki
Jump to: navigation, search

PF Users Navigation:

Edit

urgent

  • 10-19-09 : MC appears to have overflow in VQ
  • 10-12-09 : at boot, after commutation, the zero position of the VS advances
    • can use s,x,Z from sensors screen (3) to adjust. x decreases

important

  • 10-27-09 : had to disable the abs encoder error check, wasn't working in S3. Thinking that until we can ID which of the 2 encoders is 'bad' then this error check doesn't really help us.
  • 10-19-09 : the MC imp controller, and force fields scale factors don't change between MILE and MR encoders. ~3% error.
  • 10-19-09 : when encoder error, need to guess which encoder is still good. must servo to zero or hold zero current
    • partial fix - zero current / servo to zero - but still using incremental encoder.
  • 10-12-09 : read / write of EEPROM disrupts SM to MC communication - but recovers
  • 10-12-09 : SM UI commands to load/safe EEPROM defaults causes major malfunction, disabled for safety.
    • seems to be timing problem - not code bug as wrote 'do_command_146' which is called from tk_cmdset.c and statemachine_comm.c "(" in one case works....
  • 10-19-09 : if matlab not running at start, then LF needs to be restarted to make use of.
    • maybe the answer is to require -M <address> option and an error if not running


  • 9-29-09 : motor has minor, random, movement while in mode 3
  • 9-29-09 : motor moves 1/4 turn when entering mode 3
  • 9-22-09 : SWIFI at times sends garbage over radio. Cause unknown.
    • One instance of this was caused by WIPORT baud rate problem
  • 9-22-09 : boot sequence has commutation trouble. needs to limit # of retries

unconfirmed

  • 10-19-09: seems to work in S1&S2. 9-29-09 : addition of MILE encoder error checking or encoder type detect might have made use of non-MILE encoder motor not work

nuisance

features

  • 10-19-09 : need to send back MC state machine state (over swifi)
  • 10-19-09 : MC needs an current offset table.
  • 10-12-09 : when PDA used to adjust parameters, have them automatically saved into EEPROM
    • possibly, PDA action will also enable relaxed mode?
  • 9-29-09 : ability to support high speed (10khz) logging of specified signals
  • 9-20-09 : figure out how to reduce the distribution of the resistance measurements.

needs testing to be resolved

  • 9-29-09 : commanding mode 2 (without disabling ABS encoder) causes an ABS encoder error - causing MCSM to jump to state 80

resolved

  • 10-19-09 : (appears to be resolved when SWIFI provides enough data.) : 10-10-09 : lifefix :
<type 'exceptions.IndexError'> index out of bounds
File "/home/cbarn/src/pfcmd_python/lifefix_analysis.py", line 363, in process_cycle
valid_ndx = numpy.arange(start[0]+1, 2+start[0]+stop[0])


  • 10-18-09 : (fixed) 10-16-09 - all 3 processors we're using unsupported highspeed baudrates - the result was a different baudrate being used. changed baudrate for dataport to 1.25M and for SC to MC communication to 2.5Mbaud
  • 10-17-09 : resolved 10-11-09 : matlab on CB machine needs to be running.
  • 10-15-09 : resolved - 10-03-09 : lifefix - the swifi is likely not being flushed at the start of a cycle. at the end of the cycle the logging is stopped before it is known that the full cycle data has been received. (eg data could be help up in TCP buffer space.)


  • 10-8-09 (fixed) nuisance 10-07-09 : on LF, when using SWIFI mode, must run UDP_TERM on PC - else there are obscure error messages from run_life. And the swifi no longer works.
    • added code to swifi.py to write to the udp destination and trap an error. if there's an error, the text output stream is set to None.
  • (10-07??-09) : diff seems to be minor : 9-30-09 : determine drift of real-time clock
    • 9-30-09 : 9:57pm - set clock to with in +- 0.5 sec of GMT
    • 10-3-09 : 4:10pm - +0.765
Retrieved from "http://mediawiki.dideas.com/index.php?title=Pf_insects&oldid=5309"