PDA

View Full Version : What Shooter Design was Most Consistent


mikegrundvig
03-06-2012, 01:30 PM
Sadly, our first competition of the year went very badly. Our vision system is accurately telling us the distance to the target in matches (+- a few inches most of the time) and we were able to control the RPM of our shooter very accurately as best we can tell. The PID loops appeared very reliable once tuned. Sadly, every shot was hugely variable so we were never able to really dial in the shooter itself. We have removed the entire shooter and hope to redo it before the next regional.

Our shooter has four 8" wheels, two on top, two on bottom. Each is run by a 550 motor to a max of about 2200 RPM. Honestly, far more shooting power than is practical. We had azimuth (horizontal rotation) as well as elevation (shot angle) in the system but we ended up pinning the elevation at around 60 degrees and just varied the RPM because we had mechanical issues with the elevation motors. Due to our rather obnoxious drive system that can't be easily changed we are going to have to retain the azimuth control to a limited degree as I doubt the robot can spin in place effectively

I believe our inconsistency came from our shooter having mechanical problems - feeding variability, lots of flex in the design, it had some bends in it due to crashes, far too much compression, etc. What I'd like to know is what other shooter designs were people using that are consistently accurate while still being simple enough for reliability.

At our regional the best teams appeared to be using a "hooded shooter" with two wheels on a single axle and limited compression. I'd love to see other solidly reliable ideas too.

As of right now, I think our best bet is a new shooter we can dial in to take advantage of our consistently good vision data.

Thanks!

-Mike

s_forbes
03-06-2012, 02:11 PM
...
I believe our inconsistency came from our shooter having mechanical problems - feeding variability, lots of flex in the design, it had some bends in it due to crashes, far too much compression, etc. ...


From our testing, this was the biggest culprit leading to shot variability. There were hardly any tweeks needed to the shooter after we got balls feeding consistently and made sure every guide/support/contact point for the ball was very rigid.

Our shooter is a similar configuration to yours, but with smaller 4" wheels (smaller wheels = smaller support structure needed = easier to make lightweight and stiff). We don't have any experience with the single wheeled shooters for this year, but I'm sure the above advice still applies.

Ether
03-06-2012, 02:20 PM
Our shooter has four 8" wheels, two on top, two on bottom.

Do you have an exit chute to guide the ball as it exits the spinning wheels ?

Or is the last thing the ball touches the wheels?

If the latter, that could be your inconsistency problem.

http://www.chiefdelphi.com/forums/showpost.php?p=1108019&postcount=22

Chris Hibner
03-06-2012, 02:45 PM
Are you running your upper wheels and lower wheels at the same speed? If so, I would recommend running the upper wheels much slower than the lower wheels. If the two sets of wheels have the same linear speed, the ball will exit with no spin. That creates a lot of aerodynamic buffeting on the ball resulting in an erratic trajectory. Knuckle-ball pitchers have been using that to their advantage for years since erratic ball flight is great for fooling batters. However, erratic ball flight is not good for shooting baskets consitently.

A hooded shooter is easy to make, but it takes more than double the speed of a shooter with wheels on both sides of the ball.

mikegrundvig
03-06-2012, 03:53 PM
From our testing, this was the biggest culprit leading to shot variability. There were hardly any tweeks needed to the shooter after we got balls feeding consistently and made sure every guide/support/contact point for the ball was very rigid.Yeah, we knew it was a problem but didn't have the time to fix it. I believe it might be one of the bigger culprits as well.

Do you have an exit chute to guide the ball as it exits the spinning wheels ? Or is the last thing the ball touches the wheels? There is no exit chute currently. I believe we have far too much compression as well when I saw how little our competitors had. That video was very helpful.

Are you running your upper wheels and lower wheels at the same speed? Yes, we played with this extensively. Even with significant back spin we saw no increase in repeatability. When we knuckle-balled the shots intentionally we did see some lateral drift that went away when the ball was spinning.

Thanks much!

-Mike

pfreivald
03-06-2012, 04:01 PM
Our shooter design is much like yours. What seems to work for us -- though it hasn't been tested in an actual competition yet, it showed great repeatability regardless of whether or not the ball was beat up or brand new in practice -- is compressing the heck out of the ball immediately before it is fed to the shooter.

RB73
03-06-2012, 04:51 PM
Our shooter is very accurate. We have 2 six in wheels on the bottom and a curved plate on the top/back. We have the wheels powered by a CIMulator for the banebots motor (we max out around 4300 RPM). We read the RPM and get that within about a 50 RPM range of consistency before shooting and we use marks/objects on the field to line up. We average about 7/8 shots when lined up. We don't have any way of reading our distance or of changing the direction of the shot.

Ether
03-06-2012, 04:58 PM
RB73 wrote:
We have the wheels powered by a CIMulator for the banebots motor

When you say "the" Banebots motor, which Banebots motor do you mean?

We read the RPM and get that within about a 50 RPM range of consistency before shooting

"get that within..." does the driver do this manually, or do you have a closed-loop controller?

akoscielski3
03-06-2012, 04:58 PM
So we had alomst the same design. What we changed to make it accurate was put a piece of tread across the bottom two wheels (ours were 6in). We also are only spinning the bottom wheels for a shot from the key. The top wheels are not moving at all. The bottom wheels are at full power, giving it the back spin. We are also using a 775 motor on a 4:1 reduction. Maybe make some of these changes and see how it works?! :D Our shooter was really consistent, we went undefeated this week and won our first regional :)! Good luck with your shooter

RB73
03-06-2012, 05:02 PM
When you say "the" Banebots motor, which Banebots motor do you mean?



"get that within..." does the driver do this manually, or do you have a closed-loop controller?




1. The RS-775 motor, here is a link to the gear box: http://banebots.com/pc/FIRST/S24K-C1-7

2. We have a visual display for the device operator. She can read the actual RPM of the shooter of use an "At Speed" box that will turn green when the shooter is at the desired speed.

RB73
03-06-2012, 05:16 PM
Also, If you are worrying about the single drive shooter being too weak, stop worrying. Our team, 2866, was making 3 pointers from the coopetition bridge at the DC 2012 Play Day/Scrimmage.

Ether
03-06-2012, 05:42 PM
We have a visual display for the device operator. She can read the actual RPM of the shooter of use an "At Speed" box that will turn green when the shooter is at the desired speed.

does the driver adjust the speed manually, or do you have a closed-loop controller do that?

RB73
03-06-2012, 05:47 PM
does the driver adjust the speed manually, or do you have a closed-loop controller do that?




We have preset speeds for the shooter. The device operator chooses what speed the shooter needs to be for certain locations and then shoots the ball when the RPM is stable.

CalTran
03-06-2012, 05:53 PM
Thank you for the help. We'll try some of the advice suggested to us tonight (We took the shooter and the necessary electronics as our 30lbs off the robot.
Next question would be what you think would be the best approach to analytical testing? Some of our programmers are getting overwhelmed with exactly what to test. Ideally, we would like to pin it down to one independent variable (Whether it be the top rpm and change the bottom rpm to the accuracy, or set both rpms and find the corresponding distance, or something of that sort) and test from there.
Furthermore, how much of an exit chute would be necessary for the accuracy to improve? As our robot weighed in at 120 flat, I do not believe we would be able to add much to the chute without significantly reducing something elsewhere (Which, at the moment, we are unable to do because of the fact the robot is bagged)
When we attempted to create an entrance chute to reduce the feeding variability, it simply resulted in a drastic reduction of range (Down to less than 5 feet)

Finally, would it be worth it to scrap the quadwheel shooter all together and design a compression hood shooter in time for Oklahoma Regional (March 29th-31st)

Jeffy
03-06-2012, 06:11 PM
Thank you for the help. We'll try some of the advice suggested to us tonight (We took the shooter and the necessary electronics as our 30lbs off the robot.
Next question would be what you think would be the best approach to analytical testing? Some of our programmers are getting overwhelmed with exactly what to test. Ideally, we would like to pin it down to one independent variable (Whether it be the top rpm and change the bottom rpm to the accuracy, or set both rpms and find the corresponding distance, or something of that sort) and test from there.
Furthermore, how much of an exit chute would be necessary for the accuracy to improve? As our robot weighed in at 120 flat, I do not believe we would be able to add much to the chute without significantly reducing something elsewhere (Which, at the moment, we are unable to do because of the fact the robot is bagged)
When we attempted to create an entrance chute to reduce the feeding variability, it simply resulted in a drastic reduction of range (Down to less than 5 feet)

Finally, would it be worth it to scrap the quadwheel shooter all together and design a compression hood shooter in time for Oklahoma Regional (March 29th-31st)

Sup buddy?

I suggest you re-evaluate what tasks you need the shooter to accomplish. You may find some of the early season assumptions are not true, or not necessary.


If you decide a rebuild is in order, You have time. AND:
Please use smaller wheels :D. My vote is a single axle shooter made with something that "centers" the balls nicely and a very high shot angle. Mix in a nice feeder that puts the balls in with consistent spin, and I like it.

If you need to bounce an idea off of someone/something, feel free to give me a call (skype design sesh tonight? :P). I have far too much to list in text and far to little time to do so.

pfreivald
03-06-2012, 06:15 PM
Next question would be what you think would be the best approach to analytical testing? Some of our programmers are getting overwhelmed with exactly what to test. Ideally, we would like to pin it down to one independent variable (Whether it be the top rpm and change the bottom rpm to the accuracy, or set both rpms and find the corresponding distance, or something of that sort) and test from there.

For non-fender shooting, we found that top rpm = bottom/2 was nearly close to perfect every time. I hope that helps when you tune things.

When we attempted to create an entrance chute to reduce the feeding variability, it simply resulted in a drastic reduction of range (Down to less than 5 feet)

We squeeze the heck out of the ball just before we feed it to the shooter, using some bent polycarbonate and rivets (that is, not much weight). I can't say whether or not it works for you, but it seems to work for us. Consistency in feeding to the shooter is critical for accurate shooting as far as we can tell.

Finally, would it be worth it to scrap the quadwheel shooter all together and design a compression hood shooter in time for Oklahoma Regional (March 29th-31st)

I don't know that this is answerable by anyone on this forum... It depends on what your team can get to work well between now and then!

CalTran
03-06-2012, 06:20 PM
I don't know that this is answerable by anyone on this forum... It depends on what your team can get to work well between now and then!

I suppose the question would be how easily did other teams find dialing in a compression hood shooter? And if they happen to have experience from previous years, is it easier to dial in a compression hood rather than a quad wheel shooter? How much time and testing did it take other teams to map out the rpms?

V_Chip
03-06-2012, 06:47 PM
I'd suggest using a catapult! :)

Wayne TenBrink
03-06-2012, 07:34 PM
1918 uses two aluminum drums, 4.5" long x 3.5" dia, weighing about 2.5 lb each. Each drum is directly driven by an RS775-18 via a BB 4:1 gearbox. Wheel speed is monitored, but control is open loop for now. Balls exit at around 65 degrees (fixed angle). Azimuth is controlled with the chassis - no turret. Camera assist for aiming. The bottom wheel runs wide open, and the top is throttled to maximize backspin and control distance. Drivers have preset speeds, but can scale and/or adjust the speed with the classmate if needed. The current drums have a smooth OD and are coated with belt dressing (which increased distance significantly, but wears over time). We compress the balls about 2". Balls enter the shooter head while travelling in a straight line through a feeder chute set at a 50 degree angle (moving belt below, parallel moving trigger "shoe" above, close fitting guides on either side). The system is very rigid. It worked well enough to achieve a Teleop OPR over 10.5 (and win at Gull Lake with the help of 85, 1677, and 4327) despite an early string of 4 or 5 matches where we didn't score any.

We plan to replace the smooth OD drums with "toothed" OD profile shortly in an attempt to minimize the effects of unpredictable surface friction coefficients and ball compression modulus. Reducing variation (or being able to score despite it) was and remains a high priority.

Ether
03-06-2012, 07:52 PM
1918 uses two aluminum drums, 4.5" long x 3.5" dia, weighing about 2.5 lb each. Each drum is directly driven by an RS775-18 via a BB 4:1 gearbox. Wheel speed is monitored, but control is open loop for now. Balls exit at around 65 degrees (fixed angle). Azimuth is controlled with the chassis - no turret. Camera assist for aiming. The bottom wheel runs wide open, and the top is throttled to maximize backspin and control distance.

No pictures on your web page. Bummer :(

TeamSpyder1622
03-06-2012, 08:11 PM
We have 4 6-inch 2008 kit wheels (two on top two on bottom) and it has been extremely consistent. While on the practice field we were able to make 20-30 in a row without missing. As long as you let the wheels get back up to the same speed (takes a little longer with two sets of wheels) then it has seemed to be just as accurate as the one-sided shooters

GRT808
03-06-2012, 09:10 PM
During the six week build season we had a double axle shooter nearly identical to the one you described. However, although it could shoot balls across the field it also was very inaccurate. We slapped together a hood in about three hours from the KOP and some polycarbonate, no fancy or precise measurements, an installed it in our robot at SD regional. It worked beautifully and launched us into the position of highest scoring robot (not including the bonus) at the competition. We used a bit of PID for a preset speed but other than that our driver used dead reckoning which worked because the shooter was so consistent. I also noticed that us and Code Orange (another stationary shooter that scored high) scored much faster than the turreted shooters (probably won't stay like this in the coming weeks).

If you can go with a hood I'd highly suggest it.

Wayne TenBrink
03-06-2012, 09:11 PM
No pictures on your web page. Bummer :(




We're not going to win any website awards this year. We posted some pre-bag video http://www.youtube.com/watch?v=D-q-Db6RMIQ. I haven't seen any pictures or video from Gull Lake. We're trying to get something posted.

MrRiedemanJACC
03-06-2012, 11:11 PM
Wayne 2" of compression? That is good to know what you are doing. I second the week 10 Farming....


RB73, how much compression are you putting to the ball?

CalTran
03-07-2012, 09:10 AM
So, our team is looking into rapidly designing a shooter under the pretense "CAD it twice, cut once". If any team would be willing to post up some high detail CADs of just their hood shooter then it would be much appreciated. I've already started going through the CD Media, but if someone could direct me to specific ones they feel are good to emulate, that would eb much appreciated. My email is calvintran01@yahoo.com

DavidGitz
03-07-2012, 09:44 AM
So, our team is looking into rapidly designing a shooter under the pretense "CAD it twice, cut once". If any team would be willing to post up some high detail CADs of just their hood shooter then it would be much appreciated. I've already started going through the CD Media, but if someone could direct me to specific ones they feel are good to emulate, that would eb much appreciated. My email is calvintran01@yahoo.com

Check out FRC-Designs http://www.frc-designs.com/index.html
We posted our CAD of our complete robot. Our shooter is a single 6" KOP wheel powered by a BB-550 w/ a CIM-U-LATOR and a fixed polycarbonate hood with open-loop control. We shoot around 85-90% power from the key. I couldn't tell you an exact number for compression but probably around 1". We found it very consistent. I had seen many teams that were using 2 wheels and thought we would need to install a second one to help with left/right alignment but we never had a problem with that. Our shooter is a little less consistent than what we had found at our lab due to the new balls, but we were able to dial it in Thursday morning and we had a great OPR (~18).

If you watch/saw the elimination rounds at GKC you will see that we missed most (all?) of our hybrid shots. This was due to our alliance partners determining that their machine was more consistent with shooting from the center and we hadn't spent any time adjusting the power values for shooting from the left/right corners due to the new ball issue.

aldowyn
04-03-2012, 09:47 AM
Our shooter (3507) was one of the most consistent at the OK Regional. It has two wheels right next to each other, both above the ball when it shoots. What's different about ours than most is that the ball is in contact with the wheels for about half a rotation. Our feeder is also quite consistent, having guides to put it into the shooter just right, and then the two wheels pull it into place if it's out. Finally, we have a shield on the back end to propel it forwards properly.