Difference between revisions of "DEWBOT IX Drive Train"

Revision as of 03:14, 19 December 2012 (view source) MaiKangWei (talk | contribs) (→‎Pivot Module 9) ← Older edit		Latest revision as of 01:33, 19 February 2017 (view source) MaiKangWei (talk | contribs) (→‎End-of-Season Perspective)
(48 intermediate revisions by the same user not shown)
Line 1:		Line 1:
−	==October swerve module work==	+	[[Image:2013_pivot.jpg|right|frameless|upright=1.1|alt=2013 pivot (Solidworks)]]
−	Over the past few weeks prior to 17-Oct-2012, we have been modifying DEWBOT VIII's swerve modules to test cost- and weight-saving design changes. Thus far we've made 2 changes each on 2 pivots and have begun testing them as the front (intake & bridge arm side) on DEWBOT VIII ''Prime''.	+	[[image:DB9_SCH_130315_csm-10.jpg|right|frameless|upright=1.1|alt=DEWBOT's got places to go!]]Due to the anticipated strong defensive nature of Ultimate Ascent, the team decided on 11-January to utilize [[4-Wheel Pivot Drive | pivot drive]] again. In addition, to maximize agility, this drive would utilize our new Ocelot control code.
−	;Changes
−	*'''#35 to #25 Wheel Drive Chain:''' the wheel drive chain has been changed from #35 to #25, maintaining the same gear ratio. The #35 9T ''Transfer Axle Sprockets'' and 24T ''Wheel Drive Sprockets'' were changed to 12T and 32T, respectively. The maintained an integer #25 chain link distance and thus did not require any new machining. 1640 has never used #25 chain, but given the weight savings we're interesting in investigating its practicality.
−	*'''Bearing to Bushing Coaxial Drive Shaft:''' The 1" low-friction, high-durability ball bearings on the ''Coaxial Drive Shafts'' have been switched with bushings, one plastic-on-brass and the other plastic-on-plastic. (This did require remaking the cage top plate and module bottom plate.) The bearings have never failed on any generation of pivot, but at $18/ea (x8), they represent a major cost investment both in real dollars and Bill of Materials quota. (The entire FIRST competition generally cannot cost more than $3,500 + Kit of Parts). They're also represent a major weight-saving opportunity, as most of the rest of the module is running out of reasonable diet options.
−	;Test Results	+	==Issues with the 2012 pivot design==
−	*'''Coaxial Drive Shaft Bushing Failure:''' The 2 pivots were put to the test at driver practice on 18-Oct-2012, primarily in bridge-balancing and slalom runs (the shooter was malfunctioning). Unfortunately, one (the front left, plastic-plastic bushing) steering motor failed catastrophically in less than a hour, smoking visibly for almost 10 minutes and getting hotter than any steering (or other) motor we have ever had. The other test pivot (front right, plastic-brass bushing) was the second hottest motor we've ever had, though it did not emit smoke. Though the motors were checked periodically throughout the practice, the heat seemed to spike from mid-high normal to catastrophic in a matter of minutes, albeit during a rather difficult slalom course. Initial assessment attributes the failure to additional friction from the ''Coaxial Drive Shaft'' bushing.
−	**'''Update''': Upon disassembly, the plastic-on-plastic bushing module's failure has been attributed to steering motor gearbox lockup. However, based on the observed plastic-plastic stiction, we have decided to terminate this option anyway. The plastic-on-brass bushing module's failure is due partially to a re-assembly error causing undue vibration in the steering motor. This has been remedied and the module will be re-tested.
−	11-27-2012 Update. The Igus Bronze bushing module failure was from poor assemble. Will be re-mounted and tested again.
−	*We have not observed any issues with the wheel drive chain.
−	==Steering Motor Mounting==	+	Our 2012 robot, [[DEWBOT VIII]], participated in more competitions, demos and driver practice than any robot before it.  In addition, barrier crossing (which we did often) was abusive to pivots and the robot in general.  That said, the 2012 pivots generally performed very well, but the following issues were observed:
−	[[image:Steering_motor_121120.jpg|350px|right|thumb|Deaver steering motor mount with conceptual polycarbonate motor guards]]In 2011 & 2012, the design for mounting the steering motors onto the pivot modules turned out to be unsatisfactory, due to the extreme effort required to change them when damaged.  While usually not a competition issue (we just swap pivots), the shop repair time was too long.  This problem was not so obvious in 2011 because the BaneBots RS540 motors used for steering in 2011 has internal cooling fans and a robust mount and as a consequence seldom failed.  In 2012, we adopted the smaller, lighter RS395 motors for this job.  In addition to their reduced size and weight, 1) they lack an internal cooling fan; 2) the front-face cooling vents are blocked by the P60 back plate; and 3) the front-face threaded mounting holes are in very thin metal and are easily stripped.
−	Gary Deaver modified the mounting by utilizing (4) 6-32 1/4" hex spacers into which both the front and back plates of the P60 gearbox are bolted (independently).  The threaded holes in the P60 backplate need to be bored out (0.144") for 6-32 bolt clearance. This mount will allow quick removal and replacement of the steering motor/P60 backplate assembly.	+	:# Pivot tubes were not secure within the pivot tops and shifted over time.  Small and mostly angular shifts resulted in loss of calibration.  Large shifts caused pivot failure.  This problem was resolved during the season by welding the pivot tubes to tops.
		+	:# The BaneBots RS-540 motors used in 2011 were not legal in 2012, so RS-395 motors were used instead.  These motors are sufficiently powerful, but have very weak and easily damaged mounting points.  The gearbox mounting plates for these motors also cover the front cooling ports, leading to easy burn-out. We literally ‘’burned’’ through a lot of RS-395’s in 2012.
		+	:# Steering motor replacement required a significant disassembly of the pivot module.  This was annoying in 2011 with the RS-540 motors, but became a crisis in 2012 with the 395’s requirement of frequent replacement.
		+	:# Pivot module bottom plates tended to bend in service.  The 5<sup>th</sup> bolt hole for securing this to the robot chassis was almost never used.
		+	:#Set screws securing the 18T and 56T HTD5 pulleys to the CIM and coaxial drive shaft, respectively, did not reliably secure these pulleys.  This never resulted in a loss of drive power, but was troubling.
−	Polycarbonate steering motor guards are also under consideration to prevent damage to the delicate motor mounts.  In addition, we are investigating how this guard can 1) prevent undue vibration in the RS395s; 2) incorporate improved motor cooling measures; 3) incorporate a a more secure electrical plug (the tab connectors often become detached).	+	==Value Engineering==
−	We also intend to mill away part of the RS395s' P60 backplates to allow better ventilation through their front-face cooling vents.	+	[[image:Value_Engineering.jpg|300px|right]]Value engineering seeks to widen the gap between the value a product provides to the customer and its cost.  Value and cost in this context includes both monetary and non-monetary measures.  For example, addressing the issues listed above increases the value of the product to the customer (drive team & pit crew).  Cost includes monetary cost (measured following FRC cost guidelines) but also mass, manpower and machine time.
−	==Steering Motor Cooling==	+	Each Autumn, we conduct a value engineering exercise on our pivot drive design to understand what changes we would make if we were to select pivot drive again.  The results of this exercise are summarized in the accompanying Table.
−	Here follow a thread of concepts for improved cooling of the RS395 steering motors.
−	A common point to all of these concepts is milling channels into the rear face of the P60 backplate to open access for the cooling air openings in the motor's face.  The current backplate blocks these openings.  P60 backplates for the RS540/RS550 motor come with these milled channels for this purpose.	+	==Developments on the 2012 pivot design==
		+	[[2013 Pivot Development | Development]] of the 2013 pivot design took place through a number of physical tests and engineering exercises during the latter half of 2012.
−	<gallery widths=173 heights=190 perrow=4>	+	The following design changes were made:
−	Image:steering_motor_fan.jpg|Small fan mounted on top of motor guard	+	:*Switched back to BaneBots RS-540 motor (legal again) in lieu of RS-395
−	Image:San_Miguel_1.jpg|Concept for a cooling duct
−	Image:San_Miguel_2.jpg|
−	Image:steering_motor_fan_4.jpg|Small fan mounted on a blower box to force air into the motor face cooling air openings
−	</gallery>
−	<gallery widths=250 heights=250 perrow=3>
−	Image:RS395-1.jpg|BaneBots RS395 motor face
−	Image:RS395-2.jpg|BaneBots RS395 motor back
−	Image:RS395-3.jpg|BaneBots RS395 motor side elevation
−	</gallery>
−
−	==Pivot Module 9==
−	{|align="right"
−	|[[image:module_9a.jpg|upright|thumb|]]
−	|[[Image:module_9b.jpg|upright=0.7|frameless|right|]]
−	|}
−	We are looking at the following changes:
	:*Replacing 35 chain with 25 chain - 9T & 24T sprockets replaced with 12T & 32T - 44 link 25 chain reduces pivot cage height		:*Replacing 35 chain with 25 chain - 9T & 24T sprockets replaced with 12T & 32T - 44 link 25 chain reduces pivot cage height
−	:*Change wheel to AndyMark 4" HiGrip Wheel (am-2256) - 32T sprocket secured w/ 10-23 x 5/8" SHCSs - No sprocket spacers needed	+	:*Change wheel to AndyMark 4" HiGrip Wheel (am-2256) - 32T sprocket secured w/ 10-32 x 5/8" SHCSs - No sprocket spacers needed
		+	:*Adopted the VexPro 32T sprocket
		+	:*Replace top 1" ball bearing race with Igus polymer bushing
	:*Narrow pivot cage from 4" to 3.184" - machined acetal spacer eliminated - shorter axles		:*Narrow pivot cage from 4" to 3.184" - machined acetal spacer eliminated - shorter axles
	:*Pivot tube designed to be welded into the pivot top		:*Pivot tube designed to be welded into the pivot top
Line 50:		Line 32:
	:*Elimination of (1) of (2) E-clips on the pivot tube		:*Elimination of (1) of (2) E-clips on the pivot tube
	:*Reduce spacer length between module plates to 1.75" (2" in 2011; 1.875" in 2012)		:*Reduce spacer length between module plates to 1.75" (2" in 2011; 1.875" in 2012)
−	:*Replace BHCSs securing module plates with FHCSs - shift bolt/spacer nearest steering drive pulley 0.05" for clearance
	:*Use Deaver steering motor attachment		:*Use Deaver steering motor attachment
−	:*Clamp collar securing the 56T HTD5 pulley	+	:*E-clip securing the 56T HTD5 pulley
		+	:*Reduce pivot mount points from (5) to (4) and replace locknuts with rivet nuts on frame
		+	:*7075 Al replaced 6061 in the pivot braces and the module top plate
		+	:*Redesigned module bottom plate to provide additional strength (and look cooler)
		+
		+	==Ocelot Drive==
		+	{{#ev:youtube|Pej5pkFvFdY|300|right|Ocelot Drive}}The most exciting change in the 2013 drive-train had nothing to do with the pivot mechanism, but rather with the software controlling it.
		+
		+	When we first developed pivot drive, we understood that it offered the potential for [[Media:Pivot_-_Crab_with_a_Twist_edit2.pdf|dynamic driving]] in addition to straight-forward crab & snake drives.  Up until now, however, we have not been able to realize this potential and have managed with static drive modes (where joystick position maps directly to wheel positions).
		+
		+	Senior programmer Dhananjay (DJ), with help from mentor Gary Deaver, wrote the LabView vi for Ocelot drive.  Programming Lead Mike M integrated the code and streamlined the wheel positioning to be resource practical on the cRIO.
		+
		+	A great job and great teamwork!
		+
		+	==Pivot Module [[media:Swerve_1640_2013.zip | CAD Design]] and [[media:Swerve_BoM_2013.xls | Bill of Materials]]==
		+
		+	==Calibration - no change from [[DEWBOT VIII Drive Train#Pivot Calibration | last year's]]==
		+
		+	==Chassis & Wheel-base==
		+	Chassis was once again welded 6061 Al.  Mostly 1" x 2" x 0.130" U-Channel.
		+
		+	Wheelbase is 21" (nominally wide) x 22.25" (nominally long).  Chassis exterior dimensions are 27.75" x 27.75".
		+
		+	==Post-Championship Perspective==
		+	Now that the robot's been through 44 qualification and 28 elimination matches, how have things help up?  Remarkably well.
		+	*Lost one drive train chain during the last match at [[DEWBOT IX Hatboro-Horsham | Hatboro-Horsham]] (this did not seriously impede performance).  Failure occurred because the master-link became disconnected.
		+	*The 4" HiGrip wheels have worn.  Some of these have been replaced.
		+	*(2) steering motors have burned out.  Neither failure occurred in a match.  Steering motor failures occurred when steering motors were constrained while the motors were trying to turn the wheel to its set-point.
		+	* The angle sensor on the pit display pivot failed.  This has happened before.
		+
		+	We have modified the robot cart to allow the wheels to rotate (and spin) without constraint.  This change has the additional benefit of allowing easier and faster drive train checks between matches.
		+
		+	All-riveted chains (without master links) have been purchased and are replacing the originals (generally at the same time as worn wheels are replaced).
		+
		+	'''''Ocelot drive is awesome in competition!'''''  We were really hard to block and were able to play an active defense when appropriate.
		+
		+	==End-of-Season Perspective==
		+	[[DEWBOT IX]] participated in 11 competitions throughout the 2013 season and off-season; more than any previous [[FRC Team 1640 | 1640]] robot.
		+
		+	We were champions twice in-season ([[DEWBOT IX Mid-Atlantic Region Championship | MAR Championship]] and [[DEWBOT IX FRC Championship | Newton Division]]) and twice off-season ([[DEWBOT IX Battle-O'-Baltimore | Battle O' Baltimore]] and [[DEWBOT IX Duel on the Delaware | Duel on the Delaware]]) and were finalists at [[DEWBOT IX MidKnight Mayhem | MidKnight Mayhem]].  A really great season.
		+
		+	We played over 140 competition matches.  We also did an enormous amount of driver training.  This drive-train got used!
		+
		+	Out of ten (10) pivot assemblies produced, one (1) failed completely (at Battle O' Baltimore).  The weld failed at the critical tube/top junction.
−	===17-December update===	+	A second failed repairably (but was not repaired) when the miter gear on the bottom of the co-axial drive shaft came off.
−	[[image:module_9d.jpg|right|170px|thumb|]]
−	:Based on feedback, a few changes were made:
−	::*Switched to BaneBots RS540 motor in lieu of RS395
−	:::Reliability & avoidance of complex cooling & guard schemes with vanishing mass benefits
−	::*Returned to BHCSs (in lieu of FHCSc) to secure the plates
−	:::Machining avoidance
−	::*Adopted the VexPro 32T sprocket
−	:::Reduced mass & cost; improved design
−	:[[media:Pivot_Module_BoM_121217.xls|Bill of Materials]] is nearly complete.  Provisionally, it looks like we save 0.21lb (2.4%) and $44.21 (13.0%) vis-à-vis the 2012 pivot module.  Some of the cost savings is based on securing (2) CIM motors via FIRST Choice (the other two assumed to come in the Kit of Parts).	+	A third failed (and was repaired) when the needle bearings on the coaxial drive shaft seized (it is suspected that these were never lubricated during assembly).
−	:We are currently testing a modification in which the top 1" ball bearing race is replaced with metal and plastic bushings.  This could nearly double weight savings, and save significant additional cost as well.  This should work, but performance needs to be confirmed.	+	One additional steering motor failed off-season.
	----		----
−	[[Category:Robot]][[Category:DEWBOT IX]][[Category:Drive-train]][[Category:Pivot Drive]][[Category:Photo Galleries]]	+	[[Category:Robot]][[Category:DEWBOT IX]][[Category:Drive-train]][[Category:Swerve Drive]][[Category:Photo Galleries]]

---

Latest revision as of 01:33, 19 February 2017

Due to the anticipated strong defensive nature of Ultimate Ascent, the team decided on 11-January to utilize pivot drive again. In addition, to maximize agility, this drive would utilize our new Ocelot control code.

Issues with the 2012 pivot design

Our 2012 robot, DEWBOT VIII, participated in more competitions, demos and driver practice than any robot before it. In addition, barrier crossing (which we did often) was abusive to pivots and the robot in general. That said, the 2012 pivots generally performed very well, but the following issues were observed:

1. Pivot tubes were not secure within the pivot tops and shifted over time. Small and mostly angular shifts resulted in loss of calibration. Large shifts caused pivot failure. This problem was resolved during the season by welding the pivot tubes to tops.
2. The BaneBots RS-540 motors used in 2011 were not legal in 2012, so RS-395 motors were used instead. These motors are sufficiently powerful, but have very weak and easily damaged mounting points. The gearbox mounting plates for these motors also cover the front cooling ports, leading to easy burn-out. We literally ‘’burned’’ through a lot of RS-395’s in 2012.
3. Steering motor replacement required a significant disassembly of the pivot module. This was annoying in 2011 with the RS-540 motors, but became a crisis in 2012 with the 395’s requirement of frequent replacement.
4. Pivot module bottom plates tended to bend in service. The 5^th bolt hole for securing this to the robot chassis was almost never used.
5. Set screws securing the 18T and 56T HTD5 pulleys to the CIM and coaxial drive shaft, respectively, did not reliably secure these pulleys. This never resulted in a loss of drive power, but was troubling.

Value Engineering

Value engineering seeks to widen the gap between the value a product provides to the customer and its cost. Value and cost in this context includes both monetary and non-monetary measures. For example, addressing the issues listed above increases the value of the product to the customer (drive team & pit crew). Cost includes monetary cost (measured following FRC cost guidelines) but also mass, manpower and machine time.

Each Autumn, we conduct a value engineering exercise on our pivot drive design to understand what changes we would make if we were to select pivot drive again. The results of this exercise are summarized in the accompanying Table.

Developments on the 2012 pivot design

Development of the 2013 pivot design took place through a number of physical tests and engineering exercises during the latter half of 2012.

The following design changes were made:

• Switched back to BaneBots RS-540 motor (legal again) in lieu of RS-395
• Replacing 35 chain with 25 chain - 9T & 24T sprockets replaced with 12T & 32T - 44 link 25 chain reduces pivot cage height
• Change wheel to AndyMark 4" HiGrip Wheel (am-2256) - 32T sprocket secured w/ 10-32 x 5/8" SHCSs - No sprocket spacers needed
• Adopted the VexPro 32T sprocket
• Replace top 1" ball bearing race with Igus polymer bushing
• Narrow pivot cage from 4" to 3.184" - machined acetal spacer eliminated - shorter axles
• Pivot tube designed to be welded into the pivot top
• Drive miter gear shortened axially by 0.25" - further reducing pivot cage height
• Elimination of (1) of (2) E-clips on the pivot tube
• Reduce spacer length between module plates to 1.75" (2" in 2011; 1.875" in 2012)
• Use Deaver steering motor attachment
• E-clip securing the 56T HTD5 pulley
• Reduce pivot mount points from (5) to (4) and replace locknuts with rivet nuts on frame
• 7075 Al replaced 6061 in the pivot braces and the module top plate
• Redesigned module bottom plate to provide additional strength (and look cooler)

Ocelot Drive

The most exciting change in the 2013 drive-train had nothing to do with the pivot mechanism, but rather with the software controlling it.

When we first developed pivot drive, we understood that it offered the potential for dynamic driving in addition to straight-forward crab & snake drives. Up until now, however, we have not been able to realize this potential and have managed with static drive modes (where joystick position maps directly to wheel positions).

Senior programmer Dhananjay (DJ), with help from mentor Gary Deaver, wrote the LabView vi for Ocelot drive. Programming Lead Mike M integrated the code and streamlined the wheel positioning to be resource practical on the cRIO.

A great job and great teamwork!

Pivot Module CAD Design and Bill of Materials

Calibration - no change from last year's

Chassis & Wheel-base

Chassis was once again welded 6061 Al. Mostly 1" x 2" x 0.130" U-Channel.

Wheelbase is 21" (nominally wide) x 22.25" (nominally long). Chassis exterior dimensions are 27.75" x 27.75".

Post-Championship Perspective

Now that the robot's been through 44 qualification and 28 elimination matches, how have things help up? Remarkably well.

• Lost one drive train chain during the last match at Hatboro-Horsham (this did not seriously impede performance). Failure occurred because the master-link became disconnected.
• The 4" HiGrip wheels have worn. Some of these have been replaced.
• (2) steering motors have burned out. Neither failure occurred in a match. Steering motor failures occurred when steering motors were constrained while the motors were trying to turn the wheel to its set-point.
• The angle sensor on the pit display pivot failed. This has happened before.

We have modified the robot cart to allow the wheels to rotate (and spin) without constraint. This change has the additional benefit of allowing easier and faster drive train checks between matches.

All-riveted chains (without master links) have been purchased and are replacing the originals (generally at the same time as worn wheels are replaced).

Ocelot drive is awesome in competition! We were really hard to block and were able to play an active defense when appropriate.

End-of-Season Perspective

DEWBOT IX participated in 11 competitions throughout the 2013 season and off-season; more than any previous 1640 robot.

We were champions twice in-season ( MAR Championship and Newton Division) and twice off-season ( Battle O' Baltimore and Duel on the Delaware) and were finalists at MidKnight Mayhem. A really great season.

We played over 140 competition matches. We also did an enormous amount of driver training. This drive-train got used!

Out of ten (10) pivot assemblies produced, one (1) failed completely (at Battle O' Baltimore). The weld failed at the critical tube/top junction.

A second failed repairably (but was not repaired) when the miter gear on the bottom of the co-axial drive shaft came off.

A third failed (and was repaired) when the needle bearings on the coaxial drive shaft seized (it is suspected that these were never lubricated during assembly).

One additional steering motor failed off-season.

---