Bearing Blocks vs. Idlers

[MattC9]

Recently I have been looking to chain tensioning systems for drives and I have been designing in a movable idler sprocket into my drive, but recently I have herd that idlers decrease efficiency and take power away from your drive. I would like to run bearing blocks but I think they require a lot of machining. Do bearing blocks require a lot machining, and are they worth the extra effort when it comes to the over all efficiency? Also do idlers really take a lot of power away from the system, where you would want use a lot of machining resources for the bearing blocks?

If you have any other input tips or tricks that you have found while designing something like this I would love to hear it!

[AlecMataloni]

I can't answer the questions about whether or not it's worth it for your team to design and implement sliding bearing blocks, because I don't know your team.

In the past, my team has used bearing blocks attached to screws to tension our chain without decreasing efficiency: (http://www.youtube.com/watch?v=MgYnw...XrOpK1&index=2)

It's one of the best solutions for maximizing efficiency in a drivetrain, but the process is very machining-intensive. I'm sure your sliding bearing blocks, while not as complex to machine as our tensioning system, will have similar pros and cons that accompany any machined part that isn't flat.

I recently had a similar discussion with one of my friends, and he brought up a profound point.

"It's not about the numbers, it's what you can do with them."

I'm not going to pretend that I've done the math, so here's just a ballpark estimate: Say you're losing around 5% efficiency when adding 18t idler sprockets that you bought for 6 bucks each from VexPRO (great value, by the way).

Is it worth the machining time and design time to make your own tensioning system? Are you confident in your ability to get it right the first time, if you haven't prototyped in the offseason? If your machining time is limited, would time spent by making the bearing blocks be better spent by machining parts for your end effector? I'd ask more questions, but I think you get the point.

If you want my honest opinion, I think that our chain tensioning system is a bit convoluted and unnecessary. I would say that using idlers isn't a bad option at all. Heck, 1503 in 2011 and 1114 in 2012 (IIRC) used dead spaced (not tensioned) #35 chain runs on their drivetrains, and you know where they ended up. It's really a matter of preference.

If you know your limitations, you know your budget, you know your sponsor's capabilities, etc, and you decide that the boost in efficiency is/isn't worth your time and effort, then make that judgement call.

Remember that time is one of the most scarce resources during build season. Use it wisely.

[apalrd]

If you have a bandsaw, drill press or mill, and some student machinists who care about precision, you can make bearing blocks fairly easily. We've done a variety of systems for both live and dead axle drivetrains:

Live axle - We used a aluminum plate with three holes in a straight line, and mounted two such plates with a bearing in each and bolts holding the remaining two holes through spacers and the frame rail. We tensioned it using cams, which were a piece of plate cut in a spiral with a slot for a screwdriver.

Dead axle - We used a rectangular aluminum plate with two holes (one sized for the axle and another drilled and taped for a #10 bolt). A third hole was drilled and taped on end for a #10 bolt. The frame rail had a slot for the axle bolt (which was smaller than the axle itself, and taped into the end of the axle) and a slot for the #10 bolt. The bolt on the end hole was attached to an L bracket for support, and tighting the bolt pulled the block to the end of the slot (tensioning the chain). We would then tighten the other two bolts to keep everything in place.


A properly designed moving idler should not rob any significant power. Make sure the material you use is smooth, an use a bearing if you use a sprocket.

An example of our live-axle bearing blocks on page 4 of the dual drive powerpoint here. The red fixed cantilever bearing block assembly could be used to build a non-articulating drivetrain, without the green lift forks.

[BJC]

Originally Posted by MattC9

Recently I have been looking to chain tensioning systems for drives and I have been designing in a movable idler sprocket into my drive, but recently I have herd that idlers decrease efficiency and take power away from your drive. I would like to run bearing blocks but I think they require a lot of machining. Do bearing blocks require a lot machining, and are they worth the extra effort when it comes to the over all efficiency? Also do idlers really take a lot of power away from the system, where you would want use a lot of machining resources for the bearing blocks?

If you have any other input tips or tricks that you have found while designing something like this I would love to hear it!

I'm guessing that you heard this from the Wildstang drivetrain video.

In 2011, 33 used a bearing block setup which pulled the chains tight. This drivetrain was around 85% efficient which is pretty good.

In 2012, 33 ran "loose" chains. We had delrin skids which we shimed up to tention the chains but not to the point where the chain was tight around the sprockets. This drivetrain was around 92% efficient which is extremely good.

Basically, either is exceptable for FRC applications if done correctly. A bearing block setup is inherently more efficent but as proved by our 2012 robot idlers can be made more then good enough. There isn't much in the way of tentioning thats simpler then bolting 1/4" delrin slabs between wheels.

Edit: Curses! Beaten by Andrew.

[MichaelBick]

Originally Posted by AlecMataloni

I can't answer the questions about whether or not it's worth it for your team to design and implement sliding bearing blocks, because I don't know your team.

In the past, my team has used bearing blocks attached to screws to tension our chain without decreasing efficiency: (http://www.youtube.com/watch?v=MgYnw...XrOpK1&index=2)

It's one of the best solutions for maximizing efficiency in a drivetrain, but the process is very machining-intensive. I'm sure your sliding bearing blocks, while not as complex to machine as our tensioning system, will have similar pros and cons that accompany any machined part that isn't flat.

I recently had a similar discussion with one of my friends, and he brought up a profound point.

"It's not about the numbers, it's what you can do with them."

I'm not going to pretend that I've done the math, so here's just a ballpark estimate: Say you're losing around 5% efficiency when adding 18t idler sprockets that you bought for 6 bucks each from VexPRO (great value, by the way).

Is it worth the machining time and design time to make your own tensioning system? Are you confident in your ability to get it right the first time, if you haven't prototyped in the offseason? If your machining time is limited, would time spent by making the bearing blocks be better spent by machining parts for your end effector? I'd ask more questions, but I think you get the point.

If you want my honest opinion, I think that our chain tensioning system is a bit convoluted and unnecessary. I would say that using idlers isn't a bad option at all. Heck, 1503 in 2011 and 1114 in 2012 (IIRC) used dead spaced (not tensioned) #35 chain runs on their drivetrains, and you know where they ended up. It's really a matter of preference.

If you know your limitations, you know your budget, you know your sponsor's capabilities, etc, and you decide that the boost in efficiency is/isn't worth your time and effort, then make that judgement call.

Remember that time is one of the most scarce resources during build season. Use it wisely.

This is exactly how you should look at building a robot. It all boils down to whether the boost in performance is worth design, programming, and iteration time down the road.

That said, bearing blocks can be super easy(at least on a tube frame. I haven't done a sheet metal drive so I couldn't say). Take a look at some of the CADs 973 has posted. Their bearing block is super easy to make.

[Travis Schuh]

Originally Posted by BJC

In 2011, 33 used a bearing block setup which pulled the chains tight. This drivetrain was around 85% efficient which is pretty good.

In 2012, 33 ran "loose" chains. We had delrin skids which we shimed up to tention the chains but not to the point where the chain was tight around the sprockets. This drivetrain was around 92% efficient which is extremely good.

How are you determining your efficiency numbers? Am I picking up that you are doing custom testing? (If so, do you mind sharing your methods?)

Thanks,
-Travis

[apalrd]

Originally Posted by Travis Schuh

How are you determining your efficiency numbers? Am I picking up that you are doing custom testing? (If so, do you mind sharing your methods?)

Thanks,
-Travis

We measure the speed loss. Robot speed / theoretical robot speed.

We use this number as efficiency. We should also be able to figure out how much power we are loosing. But, I don't think we ever actually calculated that. We were primarily interested in acceleration and top speed (in that order).

The numbers are approximations, due to test noise and limited sample size.

Other data showed us that our 2012 drivetrain was more efficient, such as the significantly increased coast down length over 2011.

[IKE]

As a side bar, apalrd & BJC learned a very important lesson in their testing.
Some design for top speed. Some design for acceleration.
Time to distance though can be key in some games (pick and place games).

For shorter distances, acceleration is more important. For longer distances top speed might be more important. We initially started the season with an emphasis on top speed assuming full field width runs (about 24 feet) and full field length runs (about 50 feet). In reality, with the barrier in the middle slowing you down, and the fact that you typically went from 1 side to the middle, the typical sprint was more in line with 10-15 feet. Because of this (and other reasons), the guys regeared for MSC and the Championship to accelerate better.

[apalrd]

Originally Posted by IKE

As a side bar, apalrd & BJC learned a very important lesson in their testing.
Some design for top speed. Some design for acceleration.
Time to distance though can be key in some games (pick and place games).

For shorter distances, acceleration is more important. For longer distances top speed might be more important. We initially started the season with an emphasis on top speed assuming full field width runs (about 24 feet) and full field length runs (about 50 feet). In reality, with the barrier in the middle slowing you down, and the fact that you typically went from 1 side to the middle, the typical sprint was more in line with 10-15 feet. Because of this (and other reasons), the guys regeared for MSC and the Championship to accelerate better.

After exams, I was planning on writing a paper about this and the simulation we did to back up/verify the change, including the battery modeling.

[Gregor]

Originally Posted by apalrd

After exams, I was planning on writing a paper about this and the simulation we did to back up/verify the change, including the battery modeling.

I'm very much looking forward to this.

[Littleboy]

Continuing on what IKE said, if you are going long and short distances a lot in one match, shifting may be helpful. You can have one gear maximized for long distances, and another gear for shorter distances. This way, no matter which distance you are going, it will usually be fairly efficient time-wise to get there.

[apalrd]

Originally Posted by Littleboy

Continuing on what IKE said, if you are going long and short distances a lot in one match, shifting may be helpful. You can have one gear maximized for long distances, and another gear for shorter distances. This way, no matter which distance you are going, it will usually be fairly efficient time-wise to get there.

That's fantastic in theory. There are several issues that come in to play:
-Shift times/shift lurch
-Ratio spread
-Pushing

Pushing for long periods of time (e.g. pushing an entire alliance up a bridge for a triple) requires the robot to be traction-limited to 40a/motor or lower to avoid tripping the individual motor breakers. For a 4xCIM drivetrain, the highest speed you can get that at is around 5.5fps using high traction wheels on carpet. That's not a great speed to play most short games at.

Ratio spread of commercially-available transmissions limits your available choices in shifting. AM shifters are either 4:1 or 2.56:1 ratio spread, Vex ball shifter is 2.27:1. Since the vex ball shifter did not exist at the time, it was either 4:1 or 2.56:1 spread. To optimize a 2-speed for long/short gameplay (instead of driving/pushing), the ideal speeds for generalized FIRST games are somewhere in the vicinity of 8fps and 13fps, which is roughly a 1.6:1 ratio spread. You can't get that using parts from AndyMark.

Shift time and shift lurch are important when auto-shifting. Autoshifting can increase acceleration time. The general upshift logic looks for the point where the acceleration curves in low and high intersects, and shifts at or just before there. Downshifting uses totally different logic, with totally different goals, but it gets you back into low to accelerate. This is usually not a problem when pneumatically shifting, but we were servo shifting, and the shift time wasn't a huge problem, but the left and right sides would shift asynchronously and the robot would lurch to one side, which is not good. Our 2011 robot auto-upshifted, which we were happy with. It could also auto-downshift but we disabled that because it was too difficult to calibrate well with the driver interfaces we were using - A coastdown shift as we approached the rack would change the turning performance, which was not desired. I believe we could solve this issue with better calibration and the Halo-type drive we use now, but we weren't running Halo in 2011.

Conclusion: The real solution is a three speed transmission that can quickly shift between second and third. However, that's just too much work, so we'll stick with the AM or Vex 2-speeds and optimize the final drive to get the best driving performance in high gear, while keeping low traction limited.