Azimuth motor torques

dec07

az motor names/locations
az motor torques vs velocity (by month)

23jun08: on 11jun08 mot82 torques read non-zero when power off.
21feb08: az motor torques for group 1 and group 2.
05feb08: motor torques vs az after amp pots adjusted.
jan08: variation in az motor torques.
Dec07: az motor torques vs az position

Motor names/locations (top)

There are 8 azimuth motors numbered 11,12,41,42,51,52,81,82. They are located at:
az motor locations

When tracking, one pair of each motors on each azimuth side does the most work (backlash compensation)

CW motion: 11,12,51,52 do most of the work
CCW motion: 41,42,81,82 do the work

When slewing, the compensation may be turned off??

az motor torque vs velocity by month (top)

Each month the motor torques are binned in velocity (-.5 to .5 deg/sec with 51 bins). The average for the month is then computed for each bin. The average torques versus velocity are plotted for all the motors and then for motor groups 1 and 2 (a motor group is a diagonol set of motors). A final plot of the counts for each velocity bin is included. The data includes all motion of the telescope: tracking and open loop slewing.

Az torques vs velocity by month
year	jan	feb	mar	apr	may	jun	jul	aug	sep	oct	nov	dec
2008	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf
2007	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf
2006	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf
2005	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf
2004						.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf	.ps .pdf
2003							.ps .pdf

processing: x101/agc/mottq/aztqbymon.pro

11jun08: Az mot82 torque reads when shutoff in auxmode.

On 10jun08 around 2300 hours azimuth motor 82 had an amplifier failure. The Azimuth was switched to aux mode 1 where only motors 11,12,51,52 were in use. The motor contactor for motors 41,42,81,82 was shut off. The amplifiere was replaced 11jun08 after 8am.
The plots show the azimuth status for 11jun08(.ps) (.pdf) :

Page 4: bottom shows mnCont2On low. This is from selecting auxmode.
Page 5: shows the azimuth amplified status. 41,42,81,82 are not ready because the az 2 motor contactor is open. Amp 82 has a fault.
Page 8: shows the motor torques vs time. The top trace is motor 82. It was measuring a torque even though the amplfier had no power. The other 3 amplifiers of this group were not showing a torques.

The amplfier did have a fault so maybe these readings were erronous.

Looking at the historical torque data for the azimuth drives:

prior to 11jun08 mot82 torque was always close to 0
11jun08 morning: mot82 torque read non zero even when faulted amplier had no dc bus power.
11jun08 after amplifier replaced, mot82 torque now reads non zero

SUMMARY:

From nov07 till 11jun08 the az mot82 torques have been close to zero
11jun08 morning: mot82 torque read non zero even when mot82 faulted amplier had no dc bus power.
11jun08 after amplifier replaced, mot82 torque now reads non zero
Did we fix the az mot82 torque readings when we replaced the amplifier on 11jun08 or are the current readings artificial (as were the readings for 11jun08 in the morning).

processing: x101/080623/chkaz.pro

21feb08: Az motor torques for motor groups 1 and 2. (top)

The azimuth motors can be run with only 4 motors using auxiliary mode. Aux mode splits the motors into 2 groups: group 1 uses motors 11,12,51,52 while group 2 used motors 41,42,81,82 (be careful with the only display messages which tell which group was disabled rather than which group was used).
On 21feb08 we measured the motor torques for individual motor groups. The setup was:

Enable 1 motor group
Move the azimuth between az 350 to 390 at .4 degrees per second.
We went back an forth in program track mode (PI loop on) and then repeated that in rate mode (open loop).

The plots show the motor torques when running with motor group 1 or group 2 (.ps) (.pdf):

The motor torques are color coded. The horizontally dashed lines bracket a motion in program track (red) and rate mode (green).
Top: the motor torque when motor group 2 was used (motors 41,42,81,82)
2nd: the azimuth position during the group 2 test.
3rd: the motor torques when motor group 2 was used (motors 11,12,51,52)
Bottom: the azimuth position during the group 1 test.

Notes on how the az drive system works: (see dcs block diagram of hagen.. my az version)

the PI velocity loop is active for ProgTrack and for Rate mode. A position loop is also active for ProgTrack.
The PI velocity loop uses the average speed of m11,m12,m41,m42 (one side). The motor speed for m51,m52,m81,m82 is not used or reported in the speed average.
The backlash compensation between 11,12 and 41,42 as well as 51,52 and 81,82 is not enabled (K2,k4 amp boards 1,2 is 0).
The torque bias bit (K2,k4 amp boards 3,4 is 0).
The bending compensation is added to the system. Check if it is zero or not.

Conclusions:

Rate mode and program track mode give similar torques
The motor torques for group 1 have similar values around 2.
The motor torques for group 2 have very different values

motor 81 (inside track) has a torque of about 4 while 82 (the corresponding out track has a torque less than 1.
motors 41,42 have a mean close to 2 by they have more variations than the motor 1 group.

When grp 2 active only speeds from 41,42 are used for PI loop motion. check to make sure the speed output of amp is ok.

Try to find out where the difference is coming from:

Monitor the 4 motor values while in group 1 and group 2.

Does the difference come from DCS output:

Put dcs board on extender (universal amp card #4 for motors 51,52,81,52. look at

at the speed input of each motor,
the inner outer difference (amp card 3 relay K1,K3).

Check the torque read backs at the amplifiers to make sure the plc is reading them correctly.

If all the inputs look the same and the read backs differ then Switch amp for motor 81 and 82.

If motor 81 torque remains high then it may be the motor cable, motor, or resolver.

processing: 080221/chkaz.pro

05feb08: Az motor torques vs az after adjusting amplifier pots. (top)

The azimuth amplifiers have pots that can be adjusted to change various parameters. The pots with there "correct" settings are;

Stability: should be full CCW.
CurLimit: should be full CW. This can limit the current/torque that the amp provides.
CmdScale: should be full CW. This can scale the input command voltage.
Balance: not used (according the klaus's documentation). This is a zero offset adjust for the command.

We found that some of the motors had there curLimit not at the default position and some of the cmdScale pots were not all the way CW. We adjusted all 8 to be the default value.
After adjusting the pots, we tested the torques by moving the azimuth. The setup was:

Dome at 10 deg, ch at stow.
Az moves 270 to 630 and then back at .3 deg/sec in tracking mode.
Az moves 270 to about 190at .3 deg/sec in rate mode (no PI loop).

The plots show the torques during the az swings (.ps) (.pdf):

Page 1: az swings in program track mode (PI loop on):

Top: Torque vs az for CW motion motor group 1 (mot11,12,51,52)
2nd: Torque vs az for CW motion motor group 2 (mot41,42,81,82)
3rd: Torque vs az for CCW motion motor group 1.
Bottom: Torque vs az for CCW motion motor group 2.
CW motion uses group 1, CCW motion uses group 2.
Motor 82 is not providing any torque.

Page 2: az swing in rate mode.

There is only a CW swing.
The results are similar to the program track data.

Conclusions:

The torques look correct except for motor 81.
Az 130-160 has a dip/bump in torques. m11,12 to down while m51,52 go up.

This is close to the location where the encoder rack gear outrunning is a maximum.
At 150 the radius is a minimum so the motor 11,12 has a shorter distance to travel while 51,52 are moving over a longer path. If we had the individual motor velocities we could see the difference.

processing: x101/080205/aztorques.pro

jan08: variation in az motor torques. (top)

The azimuth motors were causing problems in dec07 after the painting project. Some adjustments were made to the amplifiers at this time. After these adjustements, the torques measured by some motors showed more variation than before:
The plots show the azimuth motor torques (.ps) (.pdf):

The plots show the azimuth motor torques for 12 days binned in velocity (-.5 to .5 deg/sec with 51 bins) and then averaged over each bin.
The top plot shows all 8 motors (color coded). Plots 2 and 3 show group 1 (11,12,51,52) and group 2 (41,42,81,82) separately. The bottom plot has the number of measurements in each velocity bin (0 velocity has been removed).
The first 6 pages show data from 2008,07,06,05,04, and 03.
The 7th page shows the torque difference Jul03 - otherDates to see how things have changed.
The final page blows up the torque diffence jan07 - jan08 (split into two groups).

Conclusions:

The torque difference plot (page 7) shows that the torques are no longer as evenly distributed between the motors as they used to be.
Comparing the jan08, dec07 monthly torques (see above table) it looks like the change occurred sometime between dec07 and jan08.
The problem could be:

The torque readout is changing.
Some component in the dcs that computes the torque output has changed/failed.
The D/A voltage output to the amplifiers has changed (for a given input commanded input)
The command scale for the amplifiers has changed
The amplifier compensation boards have the wrong attenuator values
Some amplifiers are failing
My guess is the spread is coming from the command scale being misadjusted.

processing: x101/080130/chkaz.pro

Az motor torques vs az. (top)

On 09dec07 6 azimuth swings were done with the dome at 15 degrees za and the carriage house at stow. The azimuth moved at .1 degrees/second. The motion went from -90 to +270 degrees azimuth. 3 of the swings were clockwise (CW) and 3 of the swing were counter clockwise (CCW).

Motor locations:

The measured torques during the 6 swings.

The plots below show the torques during the az swings. The 3 torques for each direction have been averaged together. I've plotted the inner (black), outer (red) motors on the same frame. Each page show 90 degrees of motion starting at -90 and going to 270.

Torques during the 3 clockwise swings (.ps) (.pdf):

The clockwise spins use 11,12,51, and 52 for driving. motors 41,42,81,82 have little torque.
At az=-33 there is an increase in torque for 11,12 and 41,42.

This is where the az bending occurred at az=327.
At this location the wheel flanges are press against the rails and there is no grease to help it slide.

At az=128 there is a wider bump where the torques for m11,m12 increase.
m81 is not used for CW driving but it's torques are larger than the others in its group.

Torques during the 3 counter clockwise swings (.ps) (.pdf):

The count clockwise spins use m41,m42,m81,m82 for driving.
At az=128 there is an increase in torque (similar to the CW drive) in motor 41/42
At az -80 (+280) there is an increase in torques for m41,42.
The torque for motor 81 is 3 to 4 times larger than the torque from the other driving motors.

Conclusions:

There are torque increases at az =128 and az=327.

This is probably caused by the wheel flange rubbing on the rail.
We need more grease.
One of true az bendings occurred at az = 327.

The torque in M81 is too large.

The brake could be dragging.
The brake coil cable/connection may poor so we aren't getting enough current to lift the coil completely
We have been getting some brake failures. This may be the cause.

Processing: x101/071209/aztorques.pro

page_up

home_~phil