All Notes (89)
Ver 7.16 software
Batteries fell out at 880 AH, gradual (~1-2 day) fall to 8.4 V and steady there for 930+ AH. Vehicle will be shut down shortly.
Recovered by Colin from Ascension Island fishing charters about 25 nm west of Ascension.
After batteries fell, we swam one more night and then drifted.
!!! F_MAX_WORKING_DEPTH 500 !!! (longterm_put)
4 yo's 25 to 300 m
New sbdlist.dat with m_battery_inst and some other sensors backed off to keep size respectable.
Batteries dropped today. Seeing how strong they still are underwater. Positioning for recovery. Turned science computer/sampling off.
Anyway, we have slowed down about 20% since the cleaning and have about 36 days left on battery at the current draw. That is about 20%/82 days = .25 % / day, the half life of speed is 200 days.
Sampling initial yo only now. Power usage was a bit too high sampling all the time.
Underwater reset, 32 software watchdog, ran and aborted initial/lastgasp.
Couple single yo's for testing, back to 4 yo's. Power drain is high may need to change sampling scheme.
Take 5% off battery prediction for 'OTIS effect' or 'end of life effect'
~ 965 AH
Changed TBDlist.dat to
12 7 6
so it will collect the first 3 yo's
Changed sample01.ma to sample CTD state 7, up's and downs for the ridge experiment.
If file size is too big we can revisit.
!!! ALTIMETER ON !!!
yo to 20 m off bottom
Updated TBDLIST.dat to 12 sec state 7, from 12 sec state 7 every other
Allows us to do downcast upcast combinations better.
CTD turned on every downcast.
instant gain (low power) .75 from .5, was a bit too low for how much fin we need.
This is more information for the previous 400 cc note:
-235 / +165 cc D/C
no_comms set to 16 and 20 hours
Lowered SBD size, added x_fin_max
instant gain = .5
f_fin_offset 9 degrees
Flying well, minus initial course correction, lowering igain from .004 to .004/5=0.0008. Should lower the over-correction from ramped up ierror
Tuning the glider with its new fin settings. Disabled heading deadband and set hardover hold off to 12 sec.
Segment 01160006 is trying 400 cc of buoyancy drive.
2 digifin aborts since the cleaning. Both were with the fin at > + 25 degrees. Conclusion is the fin sticks when at, near, or beyond that point.
MBD pulled for 2nd abort segment, unsure about 1st at this time.
use + digifin and exit reset did not put the fin back in service. Fin ultiamtely drifted to -.62 radians. I set f_fin_safety_max to .65 in order to attempt to be in control of glider(iridium) and still have the fin 'in service'. f_digifin_movement_retry_max and setdevlimit would not help in this regard. setnumwarn did, set it to 50, regained iridium on glider while fin was still trying to move to 0. Then slowly inched it back to 0.
Limited fin to .3 radians immediately, and set that in longterm.dat. Reset the vehicle.
Changed back to smaller SBD
Flying well, downcasts dominated by instant error + integrated falling out due to hardover. Upcasts steady state with integrated. Average fin has returned to original value of 10 degrees (fin offset could help here)
low power instant gain = .75
hardover holdoff = 30 sec (from 120)
Triple 1km to 125 m
-210 / 140 cc
CTD downcast only
Single 1 km dive:
-230 / 120 cc , +- 26 deg, 1000 cc surfac10 (nothing_commanded)
Disabled u_heading_deadband = -1 to get a better steady state situation with integrated error.
loadmission hdg_def.mi - reset fin gains
sent 500 m test dive yo20.
Ran 60 m test dive, then 2 500 m dives. Vertical speed has increased significantly. Setting up to run single full depth dive.
Nearing intervention/cleaning, so set for 1 dive w/ ctd on. The following adjustments were made at this past surfacing:
Transfer .ma files:
surfac40.ma - changed no comms surfacing to 4 hours just in case glider does not come up after 1 yo
surfac41.ma - redundant no comms, set to 5 hours
yo20.ma - number of halfcycles from 6 to 2 (1 dive to 1000 m for CTD)
sample01.ma - intersample time from -1 to 0; turns CTD on
goto_l10.ma - waypoint being changed to just west of Ascension island
Transfer .mi file hdg_def.mi. This resets all fin gains and offset to default values. This will not be used until after the cleaning.
Transfer new sbdlist.dat to add a little more pressure data. sbd load sbdlist.dat.
Put c_science_on - physically turns science on.
Trying this again; this time without the oops.
ru29.mi mission set correct fin offset. Wrong gains were also set; standard gains were set instead of _low_power_ gains. ru29.mi mission corrects that as well.
New waypoint set at 05 45 W, 16 S.
get m_why_started returns 32;
# 32 -> Software Watchdog
ru29.mi setup to set current sensor values. Fixing f_fin_offset from -5 to -.0873.
1k_n2.mi set to transfer as well, setting undervolts to 9.5 as per TWRC
Also no comms at 14 hours
RU29 reset underwater again. All systems turned on at ~700 m depth; brought voltage down to about 9.753 v. Undervolts abort is 10.0.
Reset all current settings; should put these in an "RU29.mi" mission so we can loadmission if this happens again.
Speed decay based on this plot, vertical velocity 'mostly linearly' corrected for ballast and pitch:
Output from curve fit of both dive and climbs together:
General model Gauss1:
fittedmodel3(x) = a1*exp(-((x-b1)/c1)^2)
Coefficients (with 95% confidence bounds):
a1 = 16.44 (16.36, 16.53)
b1 = 9.708 (-3.555, 22.97)
c1 = 408.5 (374.8, 442.2)
High Density Data Segment for steering:
Glider back to, which seemed to work the best:
P = .75, I = .002, D = 5
Performed steering tests all week.
P = .25 to 1, I = .0004 to .004, D = 5-10
None performed all that well suggesting increasing vertical velocity may be the best way to improve steering on current vehicle.
f_fin_offset -5 deg
PID = .75, .004, 5
Trying to lower the instant gain and increase the dgain to smooth out the steering. We think we are building for a fin_offset case of -5 to -10 degrees.
35 degree dives
gain_inst = 1
gain_int = 0.004
gain_der = 3
fin_offset = 0
Integrated error turned back on for low power mode:
Done Approx 6/24 PM EST. Value to follow in an update.
Used data from when decay (vertical velocity) started to just before hiatus, about 2 months, fit order 2 poly to it:
p_dives = [-5.6316827720749e-05 -0.0221610403831366 15.4403949223222];
OR about -1.5 cm/s @ 60, -3.5 @ 120 days
Both dive and climbs now set to 30 deg.
Apparently up vs down discrepancy on steering was from u_alt_min_depth not being set and glider not going into low power mode.
Problem still exists for fin hardover --> integrated reset.
u_alt_min_depth controls low power state, not -1 for dive to altitude, setting u_alt_min_depth to 1000 to enable low power mode.
Adjusted waypoint back to St. Helena. Continuing work on steering parameters.
Also adjusting waypoint to 17 S
Changing dive pitch to 30 degrees (.524 radians) on next surfacing
sensor: u_low_power_hd_fin_ap_gain 1.0
Waypoint changed to St. Helena Island
At next surfacing:
sensor: u_low_power_hd_fin_ap_igain 0
sensor: x_hd_fin_ap_is_hardover added to sbdlist.dat to verify if glider knows fin is hardover.
CTD off in sample.ma
25 deg, dive/climb AP, 3 deg deadband
Vertical velocity vs pitch angle findings:
Dives: ~ 4/10 = .4 cm/s / degree
Climbs: ~ 4/10 = .4 cm/s / degree
25 30 35
D 13 14.5 17
C 12 14 16
sample01.ma back to state 3 (no climbing data)
CTD: state_to_sample 7 diving|hovering|climbing
# Dive Angle Arguments (using AP)
b_arg: d_pitch_value -0.52 (30 degrees)
b_arg: c_pitch_value 0.52 (30 degrees)
c_science_on 180 7
m_science_on 180 7
Changed waypoint to:
# LONGITUDE LATITUDE
Slow things down a little to analyze better.
Overnight: 5/30 into 5/31 AM
Reset the steering gains back to defaults:
sensor: u_low_power_hd_fin_ap_gain(1/rad) 1.0
sensor: u_low_power_hd_fin_ap_igain(1/rad-sec) 0.004
sensor: u_low_power_hd_fin_ap_dgain(sec/rad) 0.00
f_fin_offset still set to 8 degrees
Steering and vertical velocity have not been doing very well. Not much success in steering at reduced pump throw, -165 / + 185 cc, so returned it to +-1000cc and it flew better overnight.
We are trying to hammer down the reasons for this, is it vertical velocity loss, fin fouling?
Quick analysis is that vertical velocity is lowered but Silbo has flown slower and not had too many problems.
Incremental depth tests show glider was able to dive OK. Partial MBD was pulled from 400 m dive and complete 60 m dive.
Glider returned to +-1000 cc @ 25 deg.
System reset at Wed May 15 13:30:33, glider ran initial.mi underwater and surfaced for the initial mission surfacing. Due to 7.9 bug and bug applicable mission file, initial.mi, glider ran yo loop until aborting 2013-05-25T06:51:52 for GPS.
Did not pull any log files, the smallest was 2 MB. Power usage was ~ 50 days at 2.5 AH/day, or 125 AH.
Iridium outage 5/2 EST PM
Climb pitch angles decreasing over time.
Power surge was from increased battery usage. Fixed battpos working fine for now.
CTD upcast turned on for 2 segments and back off again.
Surge in power usage. Not sure if it was related to pitch as the battery has been moving more than before.
Trying fixed battpos:
.4 climb, .47 dive
Adjusted sbdlist.dat to try to shrink size a bit, but increased battpos measurements.
02/24 22 GMT set u_low_power_hd_fin_ap_dgain back to 5 from 10. 10 was too much for low power mode + noisy compass direction.
Result of steering/compass test: Compass held steady current and steering went almost back to baseline segment: ru29-2013-020-0-6-sbd(00580006)
Full power mode steering/compass test for 1 segment:
put f_fin_offset 0
put u_hd_fin_ap_gain 1
u_hd_fin_ap_gain = 1 1/rad
put u_hd_fin_ap_igain .02
put u_hd_fin_ap_dgain 5
put u_low_power_cycle_time -1
Waypoint to heading to 305 T for one segment to see if the currents change back to what they were(compass test). Then back to ~ 275 T.
P = .3, I = 4e-4, D = 10
Result: currents remained slack, which is a good thing, back to W waypoint
CTD off upcasts (only on initial downcast)
Integrated error oscillations, disabling integrated for now, lowered derivative to 5. Let's see how it does and then start to raise the gains until good.
Difficult getting u_heading_deadband to work. Requirement that integrated = 0 when the heading deadband is large (we are still stabilizing), rather than small (we are nearing steady state and want integrated to go straight) might be better?
Staying deep to stay out of 'light zone', inflecting at 100 m (from 20 m). Let that run 2 segments and then disable upcast CTD to see the power increase on each change individually).
u_pitch_ap_deadband 5 degrees
Using u_heading_deadband + f_fin_offset looks promising:
(lowering instant controls (P,D) just to try to use less fin, and increasing integrated because it is zeroing out now, will see if it integrates quick enough to correct heading deviations) (we also may be approaching the limit of the fin's 1.x degrees of movement)
dgain = 10 was too much, trying 5
Glider hung up last surfacing, will try the following for next surfacing:
u_heading_deadband 0.087 (5 deg)
Slight heading AP adjustment:
South 500 m test complete, true heading west 1 km yo, then back to NW waypoint and triple yo's.
Steering very well given compass.
1 south segment, 500 m x 1 yo
Steering tests conducted at:
kp = .1
ki = .0004
kd = 2
Took a bit long to get on center and had some 30 deg drifts. Next segment/overnight I will try the following:
kp = .2
ki = .0004
kd = 7.5
(on another note, its possible we can use heading_deadband integrated error reset coupled with fin_offset, but thats another test)
Scott Pitch Tests (all due north):
ru29_2013_029_0_28 fixed battpos .4 in forward - typical wobble we have seen in the past on downcast (pitch battery didn't move so not pitch battery related)
ru29_2013_029_0_30 - actually not too bad, +- 60 deg at its worst, but still a decent wobble on downcast; pitch was set to 26 deg AP
ru29-2013-035-0-0 - no integrated error, instant gain still .25, fin offset 10 degrees
A bit more under control, eliminated integrated error gain to get a starting point.
Due west segment flew well. Trying some gains to see if we can fly NW any better:
f_fin_offset .14 = 8 deg (this will keep the glider straight, integrated error cheat)
u_low_power_hd_fin_ap_gain .25 (with noisy compass maybe it won't throw the fin so much)
1 yo due WEST, test heading, then back to NNE triple yo's
Pitch AP upcast 26 degrees
3 yo's @ 20 deg down, 25 deg up
D .365 / C .385 in battpos
D -165 / C 185 cc oilvol
sbdlist.dat returned to smaller size
removed sci_ctd41cp_timestamp from TBD because it was out of sync and not useful
Current correction OFF
Low Power Mode ON = 30 sec
Performed spin tests:
pitch = +-26 deg: ru29-2013-028-2-0-sbd(00650000)
pitch = +-30 deg: ru29-2013-028-3-0-sbd(00660000)
pitch = +-20 deg: ru29-2013-028-4-0-sbd(00670000)
Results are slight problems 330-100 deg @ 20 deg (unrealistic spin rates when traversing those headings)
Heading problems during downcasts starting 1/27, found out just off South Africa is a weak horizontal magnetic field.
-165 / + 185 cc
Low power mode seemed to work. Back to double yo's at 950 m just to be safe.
Next steps: shrink SBD/TBD, full depth, trim heading
First Low Power Test:
u_alt_min_depth 1000 (didn't need to do this, may undo it)
added x_low_power_status to SBD
-150 / + 200 cc (noticed glider was surfacing a bit slow near surface, will need to find the sweet spot)
+ 175, .49 in / -175, .46 in = 350 cc
(28->22 deg dives) (25->18 deg climbs, will need to adjust this a bit)
Changed mission to 2 x no_comms and 1 x surface start_when = 1 (nothing commanded)
Cleaned up ma's to use same pump throw as yo's except for no_comms which uses max pump throw.
Shrunk SBD by reducing m_pressure and m_pitch
Good steering results: slightly decreased heading variation, and now 0 deg heading offset
b_arg: d_pitch_value(X) .46
b_arg: c_pitch_value(X) .49
Lowered throw to +175 / -175 cc for a total of 350 cc, hopefully this helps the pitch angle become a bit shallower since battpos doesn't alter pitch angle much at all.
Non-low power heading test
u_heading_deadband = -1
u_hd_fin_ap_gain = 1
u_hd_fin_ap_igain = 0.02
u_hd_fin_ap_dgain = 5
3 yo / 10 hour no_comm
1000 m yo, not sure how it got a 400 m yo.
8 hour missions
North test, then more west.
4 hour surfacing
+-200cc pump throw
CC turned off!
Single segment due south, then west
Later tonight N, E, and back to west
Queued new surfac40 moving from 1.5hrs to 3hr missions change will take place at next surfacing.
Deployed by Dave Aragon and 3 Sea Technology Services Interns (JP, Ashley, Sinekhaya) aboard the University of Cape Town's Dive Boat operated by Pieter Truter.
Shore Support Haskins
TWRC delivery configuration