View Full Version : pic: MOEnarch Swerve Drive
nathannfm
02-11-2014, 19:13
[cdm-description=photo]40920[/cdm-description]
Did you have any problems with the belt wrap?
nathannfm
02-11-2014, 19:28
Did you have any problems with the belt wrap?
The prototype used chain and we have not had any problems with that. The belt can be tensioned by sliding the CIM. If we do run into problems with the wrap we can very easily switch to chain.
The prototype used chain and we have not had any problems with that. The belt can be tensioned by sliding the CIM. If we do run into problems with the wrap we can very easily switch to chain.
I was actually wondering about how much belt engagement you are getting. In the pic it looks like you only have around 90* of belt engagment on the driveshaft.
Scott Kozutsky
02-11-2014, 20:00
I was actually wondering about how much belt engagement you are getting. In the pic it looks like you only have around 90* of belt engagment on the driveshaft.
http://www.chiefdelphi.com/forums/showthread.php?t=126357
that gearbox has only slightly more than 90* of engagement but uses both thinner belt and smaller diameter pulleys. Given that, I would probably trust this arrangement.
Andrew Lawrence
02-11-2014, 20:00
10 - 20 ft/s is a pretty big gap there, Nathan :P
Sdp pi suggests at least 6 teeth engaged on a pulley for full use of the belt strength, it looks like have sufficient engagement for that belt pitch.
I like how they packaged both motors in the drive module, it looks good.
nathannfm
02-11-2014, 22:07
10 - 20 ft/s is a pretty big gap there, Nathan :P
We can use any size up to 44 teeth, flexibility man. Through our testing and based on the game we will determine our optimal speed.
billbo911
02-11-2014, 23:46
Is it a fair assumption that you will be using 8 CIM drive with this.
If so, how are you planning to keep from popping your main breaker?
Are you using current monitoring to keep yourself in a safe zone?
Tom Line
03-11-2014, 01:19
I would consider changing the location of your digital encoder for a couple reasons.
First, as you tension the belt by moving your cim, it will change the engagement on the Digital encoder.
Secondly, generally speaking those things don't appreciate a lot of side-loading.
Third, if your belt slips, your wheel will no longer be aligned with the encoder reading.
I'd consider a small lightweight bracket that places it directly over the primary shaft of the module itself, so you'll always know the true position of the module. This also relieves any side-pressure on it, and it can be coupled with something as simple as surgical tubing. It will probably be a wash weight-wise.
nathannfm
03-11-2014, 05:32
Is it a fair assumption that you will be using 8 CIM drive with this.
If so, how are you planning to keep from popping your main breaker?
Are you using current monitoring to keep yourself in a safe zone?
4 CIM + 4 MiniCIM (assuming the motor count rules stay the same) but yes, current monitoring is the plan, I know there are a few teams this year who had this many CIMs on the drive without problem, but they did not have the additional draw of the 4 rotation motors so we may be forced down to 4 CIMs. We will be testing with 4 CIMs to start then if we can scrounge up enough Jags, all 8. The new control system should make this much easier. We are also concerned about weight, so no guarantees but we wanted to include all 8 mounting positions to cover our bases.
I would consider changing the location of your digital encoder...
I am not sure if you can tell but the encoder is also on a slot so we can adjust it along with the CIM and make it just kiss the belt so there is almost no side loading. If the belt skips im pretty sure we are just gonna go to chain, we had enough headaches dealing with skipping belts this year :P
We did consider a bracket like you describe but it was just another part to manufacture and would also put the encoder in a more vulnerable location. If there is some unforeseen mechanical problem with it's current location we may go this route.
I know there are a few teams this year who had this many CIMs on the drive without problem, but they did not have the additional draw of the 4 rotation motors so we may be forced down to 4 CIMs.
Team 2587 ran 4 CIMs + 4 Mini-CIMs all year for around 70 matches + lots of practice time. Here are details about our drive system (http://2014.discobots.org/node/94).
We did not experience breaker issues while also running an intake motor & compressor.
We also did not perform any type of current sensing.
nathannfm
03-11-2014, 09:17
Team 2587 ran 4 CIMs + 4 Mini-CIMs all year for around 70 matches + lots of practice time. Here are details about our drive system (http://2014.discobots.org/node/94).
We did not experience breaker issues while also running an intake motor & compressor.
We also did not perform any type of current sensing.
Good to know, depending on the final speed we pick I suspect it will be fine, we shouldn't be trying to pivot the module while stalling the CIMs anyway. Also, nice video, I'm surprised more teams didn't use a pneumatic shooter like you guys, looks quite effective. I couldn't help but notice in the close ups, are you guys using Churros for drive axles (http://www.chiefdelphi.com/forums/showthread.php?t=130208)? Did you have any problems with that?
We did not experience breaker issues...
Here's why:
4" Versa Wheels at 12:72 gear * 18:40 sprocket
I couldn't help but notice in the close ups, are you guys using Churros for drive axles (http://www.chiefdelphi.com/forums/showthread.php?t=130208)? Did you have any problems with that?
Many of the pictures are from the 6-week build period. Eventually during the event we changed to black hex VEXpro shafts.
Churros are not recommended drive axles, but they work well for a quick prototype. We did have enough black hex VEXpro shafts on hand at the time.
Adrian Clark
03-11-2014, 16:19
I am not sure if you can tell but the encoder is also on a slot so we can adjust it along with the CIM and make it just kiss the belt so there is almost no side loading. If the belt skips im pretty sure we are just gonna go to chain, we had enough headaches dealing with skipping belts this year :P
We did consider a bracket like you describe but it was just another part to manufacture and would also put the encoder in a more vulnerable location. If there is some unforeseen mechanical problem with it's current location we may go this route.
I don't think your encoder position will yield good values because it isn't properly engaged with the belt. With the encoder pulley tangent to the belt I don't think the encoder will rotate in perfect relation to the belt. When you have a belt wrapped around a pulley, there is always at least one tooth that is fully engaged with the pulley. This ensures smooth movement. When a pulley is placed tangent to the belt you don't have a tooth constantly engaged with the pulley, because of this there will be some sliding action as the pulley rotates and your encoder values will not reflect the true motion of the belt. This effect is exaggerated when you decrees the size of the pulley.
This is all conjecture, but I highly recommend testing this type of encoder setup or it could come back and bite you.
-Adrian
nathannfm
03-11-2014, 17:00
I highly recommend testing this type of encoder setup or it could come back and bite you.
Thanks, I had not thought of this, I will make sure to keep a close eye on it.
Here's why:
+1. That kind of reduction will result in an extremely low speed, around 5 fps according to JVN design calc. Not really worth a 4 cim 4 minicim drive IMO, as even in high gear mecanum you're only running 12fps...
Free speed of CIM+MiniCIM motor combination (http://ether.comeze.com/FRC/share/CIM+MiniCIM.png) geared 1:1
= (343 + 198) / (343/5310 + 198/6200)
~ 5604 rpm
vehicle speed (with traction wheels) @ motor combo free speed =
(pi*4*5604) / (720*12.75*40/18)
~ 3.45 feet/sec
nathannfm
19-11-2014, 21:42
Getting there!
https://www.facebook.com/video.php?v=769331319795075&set=vb.104577359603811
AdamHeard
19-11-2014, 21:43
Getting there!
https://www.facebook.com/video.php?v=769331319795075&set=vb.104577359603811
The diagonal cut roughtop is killing me! Straight cut (for maximum cleat effect) is the only way to go in my mind.
Congrats on getting to the movement milestone though!
nathannfm
19-11-2014, 21:47
The diagonal cut roughtop is killing me! Straight cut (for maximum cleat effect) is the only way to go in my mind.!
You noticed that ehh?
We actually tested this last year with a force gauge, it takes ~15% more force to pull/push the robot side to side or front to back (with locked wheels) because when you put it on a diagonal more nubs are touching the carpet per contact patch area.
Looks like a happy happy joy joy moment. Some thing does not sound right when the module was rotating and driven. Those are wide wheels and may cause problems with amp draw on both drive motors and steering. Pay close attention to the tuning of the steering pid. You can have to much traction. For a fall back you might want to consider 1-1/2 or 1-1/4" Colsons. Next, get 4 of them on a frame and get programming. Then you will be half way there. Driver training is the last half. If you get it going before kick off, do you want to meet and do some swerve on swerve driving?
Note for the drive encoder, Do you really need quadrature? Wouldn't a tachometer solution be good enough? I do not now what the under side of the coaxial pulley looks like but, could you put some reflective tape strips on it and use a optical reflective sensor? We used a magnet solution last year and will use optical this year.
it takes ~15% more force to pull/push the robot side to side or front to back
Would you please post your data and test procedure?
nathannfm
21-11-2014, 02:14
Would you please post your data and test procedure?
Sure, all from memory but it should be good enough.
First I had to build The Bathroom Scale of Science (https://scontent-a-lga.xx.fbcdn.net/hphotos-xfa1/v/t1.0-9/1920567_10152245196233559_1795743674_n.jpg?oh=7e9f d527e2f623d2805365ded41f0c6c&oe=54D24D5D), the "force gauge" we used to test this because the real one we had in the lab maxed out at 50lb. The bars that stick out from the sides are the handles and the X that crosses the face of the scale has 4 ropes that converge at a shackle that was placed around a 1x1 frame member on the robot in the bumper zone. The wheels were then locked by jamming a scrap between the spokes. The robot was pulled with the ropes remaining as horizontal, and the scale remaining as vertical as possible to the ground. A reading of ~120lb was observed on the scale. The wheels were then switched to those with diagonal tread, the same procedure was followed and a reading of ~140lb was observed. A ~15% increase.
We also used a similar procedure to demonstrate how ineffective trying to drive out of a T-bone is. This time the wheels were unlocked and I pulled sideways acting as an opposing robot pushing from the oppisite side as our driver floored it (and the robot started to drive in a circle around me with a radius of the rope (~5ft)) you could see the scale instantly drop from a reading of ~140lb to something like 60lb (I don't remember the exact number but it was something drastic)
We again used the test to determine that 2" wide 4" dia wheels were ~15% better than 1 inch wide wheels. Take 4 1" wide wheels, place them on the four corners of the robot, lock them, drag. Place 4 more 1" wide wheels on the same axles, lock them, drag.
*All tests performed on actual FRC carpet
nathannfm
21-11-2014, 02:45
Looks like a happy happy joy joy moment. Some thing does not sound right when the module was rotating and driven. Those are wide wheels and may cause problems with amp draw on both drive motors and steering. Pay close attention to the tuning of the steering pid. You can have to much traction. For a fall back you might want to consider 1-1/2 or 1-1/4" Colsons. Next, get 4 of them on a frame and get programming. Then you will be half way there. Driver training is the last half. If you get it going before kick off, do you want to meet and do some swerve on swerve driving?
Note for the drive encoder, Do you really need quadrature? Wouldn't a tachometer solution be good enough? I do not now what the under side of the coaxial pulley looks like but, could you put some reflective tape strips on it and use a optical reflective sensor? We used a magnet solution last year and will use optical this year.
It was (MOE cheer right after the video cut out :D ) If you were referring to the pulsating of the sound I think that was because most of the noise was coming from the bevel gears and their location was changing relative to the mic, but I will look into it. Do you grease yours? We also have not played with tension of the rotation motor belt, we will look into that next meeting (we did notice it affected the speed of rotation between modules) We will definitely be keeping an eye on the amps and have Jags on every motor to monitor that, The RoboRIO should also make that easy when it arrives. We will hopefully get driving by Monday. Our programmers already have a basic crab drive program and are working on fancier moves now. I think we would love to have some swerve on swerve driving! You and your team have already been a great help this summer, I will ask the drive team about this at the next meeting and get back to you.
As for the encoders we debated this, but we haven't had very reliable results from magnetic sensors in the past, and there is not much room to mount an optical sensor under the pulley. We also had all the encoders we plan to use in stock so we just went with them.
The robot was pulled with the ropes remaining as horizontal, and the scale remaining as vertical as possible to the ground. A reading of ~120lb was observed on the scale.
Was this the breakaway force, or the min force to continue motion?
Was the height of the horizontal rope above, below, or at the height of the wheel axes?
nathannfm
21-11-2014, 13:41
Was this the breakaway force, or the min force to continue motion?
The force to continue motion, but surprisingly the difference was only a few pounds. Nothing like the dramatic example of the spinning wheels.
Was the height of the horizontal rope above, below, or at the height of the wheel axes?
It was about 2 inches above, in the middle of our bumper (one bumper removed for test to attach shackle to frame)
TheLoveGouru
24-11-2014, 23:39
Hey Everyone!
This is just a quick update on the progress of the MOEnarch drive system.
We have mounted the four modules onto the square chassis and it's working smoothly.
This is the Facebook link to the video:
https://www.facebook.com/video.php?v=771671739561033&set=vb.104577359603811&type=2&theater
As you will notice, the robot doesn't have any encoders on the modules just yet. We are looking to set those up by the end of next Monday and have a robot-centric swerve drive ready to go before Winter Break.
Sounds like a swerve, kind of moves like a swerve, it must be a swerve. Welcome to the 3 degrees of freedom club.
vBulletin® v3.6.4, Copyright ©2000-2017, Jelsoft Enterprises Ltd.