Happy Roller Coaster Day!

Apparently this celebration of roller coasters extends globally.  So... enjoy some photos and ride your favorite coaster this weekend! 

 

IMG_0049.JPG

Heidi The Ride at Plopsaland De Panne, Belgium. 

 

IMG_0588.JPG

The turn around, and a beautiful Belgian morning. 

 

IMG_0911.JPG

Mystic Timbers, at Kings Island in Mason, Ohio, USA. 

 

IMG_0880.JPG

Working on the “upper idler” for Mystic Timbers.   

 

IMG_1152.JPG

InvadR at Busch Gardens Williamsburg, in Williamsburg, Virginia, USA. 

 

IMG_1072.JPG

The dragon train and those beautiful Millennium Flyer wheels.  

 

IMG_2415.JPG

Wicker Man, Alton Towers in Staffordshire, England, UK.   This is the first, second, and third fly-throughs in the soon to be themed Wickerman.

 

IMG_3359.JPG

A fancy model of the ride. 

 

And to round it out, me with a fan at the IAAPA Trade show. 

 

IMG_3295.JPG

A Rookie Mistake

In order to mass produce the anchorage for the roller coasters, I needed as special machine setup. Basically, it s a specially made vise that allows me to place small angle aluminum in a repeatable position to drill the correct holes.

 

IMAGE.JPG

The base plate of the vise is close to the width of travel for my little CNC mill, which is 4 inches.  My current setup for making “one-off” parts is to use one of my jewelers vises.  Seems like no problem, right?  Wrong.

 

IMG_4031.JPG

Starting with a facing cut to clean the top, I used my typical 1/4” end mill.  Everything was working fine.  But the next operation was to adaptively clear metal.  This is where the isssue arrose.  

 

IMAGE.JPG

Notice how the z-column is tilted?  You should, it’s pretty apparent.  Well, I didn’t take into account the solid jaw of the vise overhanging on the back of the table.  So, as it worked it’s way around the metal it happened to run into the mill itself.  This sucks in so many ways when this happens.  I mean, it hasn’t happened in years.  But it makes you feel like a moron.  How could this much oversight take place? 

Afterwards, I had to tram the mill (square it up) and we are back in business.  The problem is, how do I make my custom vise? 

Well, I will be using my fixture plate that I purchased way back in the day at the beginning of my machining endeavors. 

 

IMAGE.JPG

But, the vise is still there?  Yes, I am going to make the jaws of my mini vise and use the fixture plate for spacing.  And...another problem.  Since the holes on this fixture plate are 10-32 (common 3/16” screw size), I only have 5mm screws...close but no cigar to sound like an old man.  I will be purchasing some 10-32 hardware and will continue to mill my vise.   

If you follow along on Instagram (@therollercoasterproject), then you saw my post about putting all machining videos in a playlist dubbed “Machining” on YouTube.  I want to have the videos for those who like machining, and the primary videos of ride building separated.  

 

 

IMAGE.JPG

Multiple videos are coming, so keep your eyes peeled.  I’m not sure why that’s an expression in America.  It’s weird.  But videos will be out as quickly as possible.  Also, I somehow fail to record my machining failures.  I always move the camera during the machine crashing... Oh well.   

-Coast on  

Ride Building Material

If you've been following along on our Instagram page “@therollercoasterproject” you may have seen some work with the table saw.  What I have been working on is creating a few cross-cut sleds for the table saw and band saw.  What is a cross-cut sled you ask?

 

IMG_4027.JPG

Now, this isn’t my image but it is a good example of a completed cross-cut sled.  The sled allows the user to cut repeatable and varying size parts.  How does this stack up for mini roller coasters?  I’m glad you ask.  

For our building materials, we have been using basswood for most of the structure and track.  This can be relatively expensive with a wooden coaster since there are so many components.  But, now that we have a thickness planer and a table saw, the doors are open to making our own building lumber!   

To sum it up, each 24” stick of 1/4” basswood is $0.55, and based on the desired amounts per stick we can get 3 8” sections.  Now, if we have a sheet of basswood at a quarter of an inch, we can cut up to 94 of the 8 inch sections.  The price per sheet is $15.80, and if we do the math we save a tremendous amount.  But what if we use pine boards for our structure?  An 8 ft pine board is roughly $8 at one inch thick.  Now that we have a thickness planer, we can get it down to our desired thickness and cross cut each 1/4” strip.  This saves so much money and allows for more control and inventory. 

 

IMAGE.JPG

These oak strips are cut to fit in the slots on my Ridgid table saw.  Once I place some 3/4” birch plywood on them and square it to the blade, you have an accurate cutting method.  

 

IMAGE.JPG

The oak board was actually twice as thick as shown in the picture above.  By using our thickness planer, it was brough to correct dimensions. 

The roller coaster will be made of pine!   

To Maker Faire or Not?

Now that I’m back in Atlanta, Georgia, I have the opportunity to return to the Atlanta MakerFaire. 

 

IMG_4026.PNG

Prior to moving to Cincinnati a few years ago, I had displayed roller coaster related gadgets two years in a row.  I basically “hired” a friend or my girlfriend for the weekend to help me out at the booth.  Being that I was still learning how to machine and fabricate ride designs, I didn’t have substantial ride structure to display.

This year, however, I am poised to display some working rides and full scale production.  I will also have some merch to sell/raffle off.  Of course, these are my hopes.  Just as before, I have packed my life full of events, trade shows, trips, and now graduate school.  But I will stay committed and focused. 

My goal: 

  • One fully operational test track with banked turns and magnetic brakes. 
  • A station section with platform. 
  • Two full length trains
  • one miniature ride test track
  • five metal classic TRCP cars for sale
  • shirts galore! 

Most of this is going to be documented via YouTube and this website.  So, if you’re interested and in the Atlanta area, then attend! October 27 & 28 at the Atlanta Freigth Depot.

Video Music

I once worked for a radio station in a major market, and one thing I hold near and dear to my heart is music in video.  Having some music really takes a personal video, and transforms it into a real production.

IMG_4020.JPG

YouTube has a decent music/sound library, but I’ve been searching for a larger selection.  Perhaps more variety.  Having come a long way from my days of reusing “Maple Leaf Rag” in my videos, my last upload had a bit of a change.  Though I like the Dixieland Jazz and classic ballpark and amusement park music, having song selections that play on what’s being displayed is key.

IMG_4019.PNG

Epidemic Sound is a subscription based repository of thousands of songs and sounds.  They allow you to place any and all song on to Youtube videos, and they won’t remove it should you ever cancel your account.  I am not being paid by them, I’ve just been researching

It might be worthwhile for me since I want to focus 80% of the efforts into video content as I progress, and 20% social and blog.  

Has anyone out there had successes with Epidemic Sound or any other service?  Leave comments, since I’m curious and am looking to pick some good tracks for videos.

The Next Video

 

 

From the previous posts and videos, I have been uploading the building of ride assembly tools.  In the new video,  to come, there will be a bit of a saga.  What do I mean?  I will cover a test track from design to fabrication.  To keep with the whole “brevity thing,” for any of you Big Lebowski fans out there, I will do a portion on each video.  Going forward with the larger builds, I will do the same process.  Sadly, these things take much time and can become intricate.  Let’s see if it works!

Aside from bent building and ride layout design, there will be car and train assembly and mechanical components.  The test track will have a hill, banked turn, and ultimately a magnetic brake section.  For things like the magnetic brakes, I want to throw in a bonus video of calculating magnetic fields and dampening power.  I know, lots of stuff always on the horizon.  However, these next few videos will have ride building and finally testing. 

As for the cars, I have designed the car to mount a GoPro HERO 4.  In fact, the center of rotation, called the heart line by some Swiss manufacturers (B&M), is actually at the center of the lens.    

Either way, more things to come soon!   Be sure to check out the YouTube channel and subscribe to see all the rides in motion!

IMG_0007.JPG

Correcting The Mistakes

As with all fabrication and manufacturing, if you make a mistake cutting something it can’t usually be undone.  When it comes to subtractive manufacturing, like CNC machining, then you really can replace the metal once it’s milled away.

 

IMAGE.JPG

On the eve of building some ride sections, I had the idea to modify the snazzy building tools I made last week.  Since the outside post dimension is what I choose to govern the ride width, I milled out the slot a bit wider than the original 0.25 inches.

 

 

IMAGE.JPG

Since these parts were already milled, I had to accurately locate the part in all coordinates.  This was a good time to use my z-axis touch off setter.  It worked well.   The only drawback is that it’s 50mm in height, and I am limited with my z-axis travel on my little mill.

Now the error on my part... 

I didn’t mill deep enough when widening the pockets on the parts.  So, I tried to manually control the mill.  Sadly, I happened to run the end mill into the finished edge.  This now makes my out-to-out distance 4.78” instead of the design 4.75”.  I know, it’s only a 32nd of an inch, but the upper section is dead on at the latter of the two numbers.  It should be okay for some test bent sections and a model track, but for the final rides it has to be redone. 

In summary, don’t mess up.  It sucks tremendous ass, and it takes more time.  Focus on not rushing and always triple check measurements.  Especially if they are created in a computer program to be run by another computer.  Garbage in equals garbage out.  However, I will have a video out soon showing some ride building.  Later there will be a video of making the foundation anchors, too. 

 

IMAGE.JPG

A custom work vise for my foundation anchors!

Building Tools and Ride Structure

This week has been focused on the setup of the bent building tools.  

In the linked video, I machine the full sized bent building tool.  The post spacing is 4.5 inches, which is 1/24th the size of a modern wooden coaster post spacing.   

The last thing I will have to machine to make the ride structure repeatedly is a drill guide.  The drilling guide will provide correct locations for my small hardware: 1/16” nuts and bolts.  There will be a lot of hardware on this coaster, much like the real rides.  I mean, this is a real ride, just ridiculously small. 

After working on this for so long, sorry guys, I noticed the importance of creating the tools I will need to build my creations.  However, once the tools are made I can make as many of these rides as possible.  I shall cover the globe!  Yes, globe, meaning spheroid or ellipsoid.  None of that “flat Earth” bull shit on this page.  I mean, really?  People are still questioning this fact?

well...

Aside from that, checkout the video.  Thank you guys and gals for your support! 

IMAGE.JPG

Bent Building Tools

This week I have been focused on creating the necessary bent building tools to ensure accurate and repeatable parts.  From the video last week, I machined a miniature version of the "bent builder" tool I will be using for the mini coasters.  However, this week is focused on the larger scale ride which is 1/24th the size of the real thing.  In other words, it's massive.  

The post spacing is 4.5 inches.

IMAGE.JPG
IMG_3995.JPG
IMAGE.JPG

You'll notice that the size of the stock being machined in the pictures above is much larger than what was featured in last week's video.  One thing to make sure of when working with large parts and a smaller vise is keeping the part against the parallels and that you have adequate clamping pressure.

The parallels are machine ground to be flat within 0.0001", at least these particular ones.  

This week's video will be focused on the fully bent building tool, and there will be a threaded rod with two support rods added.  Ideally, this will help create structure rapidly and accurately.

IMAGE.JPG

When you get tired of watching the mill doing it's thing...

IMAGE.JPG

You can see, somewhat, the chip pan filling up with more and more aluminum.  Once there is a significant amount, I will recycle it.  The key is not to cross contaminate one metal with another.  

In all, the mill is running well and I have had no fails this week...but there's still time.

Another fixture to build this week, which will be in next week's video is my anchor vise.  This self-made vise will allow me to make my anchor angle aluminum parts rapidly.  Since there will be so many posts, I will need to make a great deal of them.  I will also have to make a cutting jib to use on my band saw.  But, one thing at a time.

FullSizeRender.jpg

In the picture above, you can see the difference an old (left) versus new (right) end mill produces.  The larger part is, and I mean this, the best thing I've machined so far.  Back in my 3D printing days, I refined my process fairly well.  Now, it seems that I am finally getting the hang of machining.  The one thing I didn't account for on the new end mill is that there is a slight corner radius; however, the part is smooth and shiny as can be.

Questions As of Today: Part 1

As things come back into full swing, I’ve had a few questions asked.  I will post a blog response as more of a summary to what has been asked on the various platforms.   

One of the most commonly asked questions pertains to how I created the data to build the ride or to build these rides.  To sum it up, I derive measurements and angles to place wooden members in the correct places based on the ride’s designed layout.  I know, vague.  Well, think of it like designing a building: you start from the top and work your way down.  As the building gets heavier the farther down you go, more structure is required to support the increasing loads.  With that being said, from a developed ride centerline you can provide a nice cushion of structure to support the varying loads.   

As for the actual physics and math to analyze these forces, I have a few videos on basic ride calculations on YouTube.

Now, for the angles and measurements...well they come from me or can come from you!  The crazy thing about science and engineering is that you need to make initial assumptions and test them to see a result.  As it pertains to a roller coaster, you are looking for a desired feel or experience.  To achieve the aforementioned ride experience you create numbers to match (or best you think).  Then you analyze what happens along the ride.  If these numbers represent what you’re shooting for, then you create the supporting aspects of the ride.   

To put it basically: set your parameters and make it fit within.  You’ll get all your angles and measurements, trust me.   

 

 

IMAGE.JPG
IMG_2932.JPG

Testing helps evaluate measured data versus theoretical data.  Heidi approves.

The week ahead

After a weekend of thought and making machining code, we have the week ahead of us.  As stated in the recently uploaded video, this week’s YouTube extravaganza will focus on building ride structure.   

To build ride structure, you have to focus on more than just the above structure.  We have to have a way to secure it to our “ground” which will be plywood.  Previously, I had CNC milled wood with a circular hole, filled it with glue, then pressed the rectangular posts in the pocket.  This wasn’t bad, but I would prefer a more repeatable and authentic style of anchorage system. 

 

IMG_2530.JPG

Prior to working in the amusement industry, I had ideas of placing angle metal (aluminum most likely) to connect the structure to the foundation.  The picture above is during the construction of Wickerman at Alton Towers.  Having worked with these anchors now, I will be implementing them into my ride construction.  This is for the full sized mini ride, by the way.  Yes, it was a weird sentence.

To make these anchor brackets hundreds of times, I will be making a drilling and cutting jig to repeat the process.  Ideally, it should connect the posts correctly and once the structure is standing it will be aligned vertically and horizontally or called “racked.” 

I’ll have more updates throughout the week!

Making YouTube Content

As some of you probably know, I use YouTube display my builds/machining, but I’ve been sporadic in the past.  Last night I wrapped up editing and narrating my latest video: check it out on YouTube.

IMG_3975.JPG

 

In order to build these mini coasters, I decided to machine a precision jig which allows accurate structure spacing.  My later videos will be heavy in ride building, and I think they are more of what people want. 

I plan on uploading videos each Friday and want to set out to complete one portion of a larger project in each video.  Upon looking over my previous videos, I’ve noticed that they need more story telling/direction.  Last night I did some voiceover work, and with the setup I have it turned out pretty good.  I was a tad bummed that my mixing board won’t exactly work with my computer since updating the operating system, but I can record separately and piece the files together. 

This week was a bit of a struggle machining since I haven’t done it in a while, but I’m confident that later videos will have more content and focus a little less on the milling/turning of parts.  The cars will be machined in a few weeks, now the structure takes priority.  Hope you enjoy! 

 

 

The First Part

Last night was one frought with successes and failures.  The first part in this long process of creating miniature roller coasters was successfully milled.  That is, after a few days of failures.  However, growth comes from setbacks.   Or a large pain in one’s ass.

 

IMAGE.JPG

You’ll notice from the picture that much has changed from the last few posts.  Most notably, the use of a vise instead of superglue and a sacrificial plate.  Nothing against the superglue, but I know this method repeatedly works. 

Now the issues... 

I used whatever stock I had in supply, and it’s dimension was roughly 3/4 inches.  That works all well and good until you square the stock for milling and your closest parallel is just large enough to cause the end mill to hit the vise.  (David hangs head in shame).  

 

IMAGE.JPG

Somehow, I managed to not get the happenings on film or take a picture of it.  I have much of the aftermath that will be present in Friday’s YouTube video.  This failure caused the end mill to chip, thus, it’s basically worthless.  S***.  

 

IMAGE.JPG

I destroyed one of my aluminum cutting end mills, but fortunately I had a three flute steel cutter on hand for the win.  In the picture above you can seee where it collided with the vise and part.   

Before pushing forward with machining, I altered the design and g-code in Fusion 360.  After updating everything, it was a complete success.  I even ramped up the spindle speed. 

 

IMAGE.JPG

This part had many firsts, of which milling dual sided pockets with my 1/8th inch aluminum cutter with lots of coolant.  Basically the area to place the wooden supports to keep them straight.  The other first was chamfering the part once completed.  It worked successfully with my quarter inch spot drill cutter.

The part can now be used to build ride bents, and the next video will show it in action!  As said before, the video (with bloopers) will be up tomorrow at 10 AM EST and subsequent videos will be up on Fridays. 

 

 A little excessive?  This lens makes some beautiful photos and videos.  But, don’t forget to pour one out for the fallen end mill.    

A little excessive?  This lens makes some beautiful photos and videos.  But, don’t forget to pour one out for the fallen end mill. 

 

IMAGE.JPG

Milling Fails

My shenanigans wouldn’t be complete without milling failures.  As I began filming and running the code for the first part, the superglue happened to fail and the part come loose.

 

IMAGE.JPG

Quickly, I stopped the mill and assessed the situation.  Since my last milling attempt, I placed a center screw to hold the aluminum plate and milled out a relief pocket to remove finished parts.

 

IMAGE.JPG

I figured it was a mishap and decided to try again.  I placed new Frogtape and superglue, applied pressure, and hit “cycle start.”

 

IMAGE.JPG

 ...and another failure.

About this time, I was becoming frustrated and had collected a lot of footage of these subsequent mishaps.  I am going to throw them into this week’s video.   

My thoughts now are: cleanliness of mating surfaces, and pressure applied for glue to dry.  However, I’m not sure if either of these will change what’s happened two times so far.  Going forward, I will just mount my vise and clamp the parts normally.  This will also be in the video.   

 

IMAGE.JPG

Slowly but Surely

Since the last time I machined anything was a while ago, I had to look through my availabile material.  I found various sizes of aluminum stock.  Most of the stock I have is for larger car concepts, but I did happen to order three feet of smaller sized aluminum which matches my needs currently. 

 

IMAGE.JPG

I had to fire up the Dewalt cold cut saw, again, and it worked well to cut the metal to approximate lengths.  It’s a good idea to know what size of stock metal you’re going to machine to setup the machine tool paths through the CAM software.  The whole process will be available in a video coming soon. 

 

IMG_3948.JPG

The video this week will be centered around building the ride structure.  The 3D model in the picture about is a base form/jig to help build the bents precisely.  They will not be perfect, but will fit in an appropriate range.  Much like the real thing.  Stay tuned! 

First Thing’s First

The first machining project to tackle now that the mill is up and running is to make bent building forms.  The bents, on a wood coaster, are the frame style supports.  Now, we had previously machined the structure components, but this time we will be making them by hand to speed up the process.  This also doesn’t cover the mill in saw dust. 

The metal forms will allow proper alignment and bolt hole locations.  From the bent drawings, I will have proper dimensions to each member of the structure.  Keep in mind, the rides are all built with similar components found in real rides...just smaller.  This means, bolts, washers, nuts, anchor rods, etc... 

I’ll have more updates this week and a video of bent building once the forms are made. 

 

IMG_2259.JPG

A few lift hill bents from Wickerman at Alton Towers. 

Milling Machine Test

Last night I decided to test the milling machine with a simple 1 inch square and a 3/8ths on an inch circular pocket (hole).  Since I didn’t feel like setting up a machining vise and aligning it, I used a trick that I’ve seen on ClickSpring on YouTube and NYC CNC: superglue and painters tape. 

 

IMG_3933.JPG

Keep in mind this was simply just to see if the ballscrew calculations were correct (I changed the number slightly from the previous post since the steppers I have only have two points).  My goal was to machine the square and measure how close to 1 inch it actually would be.  I was hoping to be within 0.001”. 

The idea of using superglue allows quick stock placement and strong work holding.  I began by placing an aluminum plate on the mill table and machined it to be flat and smooth.  Then, I placed painters tape down to make a surface to place superglue. 

 

IMAGE.JPG

Based on John Saunders of NYC CNC, I chose Frogtape due to increased strength over 3M Blue.  The next step was to place tape on the raw stock (I had already machined it flat, too).  Finally, you apply superglue to the part and apply pressure to secure it to the fixture plate.

 

IMAGE.JPG

Now, it should have gone smoothly and completed in about 20 minutes.  I didn’t really do much work to ensure proper tool speed, but it did hold perfectly.  Aside from the loud chatter, probably due to no support in the middle of the aluminum plate, it was perfect.  Until... 

For some reason, my CNC software decided it wanted to stop near the very end of machining the square.  I had enough space to test the square dimension, but never got to the circular pocket. 

 

IMAGE.JPG

Albeit frustrating, I was happy to measure it and get far better than expected.  But there was one issue: I forgot to machine a relief hole to easily remove the part that was superglued.  Oh well, a heat gun will help burn away the superglue.   

 

IMAGE.JPG

Now, I can begin machining roller coaster parts.  I’m going to start with forms to build ride structure.  Be on the lookout on the YouTube channel for more updates! 

Calibrating the Mill

When I decided to upgrade my Taig milling machine from a 1/2” lead screw to a more accurate 12mm ball screw, I figured there would be some conversion work.  However, as I delve deeper into Mach3 (the mill control software) I’ve had a few issues.  The main issue is coming up with the correct conversion so that when I machine something, the dimensions are correct and now off by some strange factor. 

 

IMAGE.JPG

Upon looking online, there are many posts about checking the actual travel as opposed to a calculation to verify measurements.  The most basic was to place a machinist ruler and physically check the travel.  Now, this works for about one inch, but not over longer travel.  

The next approach is to test using a dial indicator to produce a more accurate measurement.  I would either mount the indicator to the spindle and have the table travel a specific distance.  Near the end of travel there would be a solid surface, usually a gauge block, to check how much more it travels into the surface.  In other words, if I made it travel 6 inches and I know if passes that mark by roughly half an inch, then I could place the dial and get the exact reading.  That number would then be input to Mach3.  It hasn’t worked too well with a 1” indicator, so I purchased a 2” version recently.  

Now, if you don’t want to check measurements that way, you can calculate the appropriate steps per revolution of the ball screw.  Since I am using 12mm diameter screws with a 2.5mm pitch (distance from one thread to the other), I must account for this variation.  From what I’ve read, each stepper produces 200 steps per revolution.  I’m not sure if that’s clear across the board, maybe someone could clarify that.  My Mach3 setup is in Imperial (in) and the screw needs to be converted.  So...

25.4mm/inch / 2.5mm = 10.16 inch

This number is then multiplied by the steps per rev (200): 

10.16 x 200 = 2,032

And since there are apparently four poles that allow rotation of the motor, you multiply by four. 

4 x 2,032 = 8,128

This number is then entered in to Mach3 per axis and should provide proper calibration.  I’ll try this and verify with the dial indicator and hopefully all will be accurate.  Please comment if you know any other way or if there are any corrections to what I am doing.  

The next task for me was to tramm the mill, or basically square it up so I have a flat working surface. 

 

IMAGE.JPG

To ensure proper squareness of the mill spindle and mill table, I used my Mitutoyo indicator which checks variations in 10,000ths of an inch, also known as tenths.

 

IMAGE.JPG

After incremental adjustments and moving the table along the length of travel, I ended up with the mill being roughly 0.0005” square.  This is great and will allow me to make accurate parts!

Back to Filming

Now that things are ramping back up, video production is back.  This morning I uploaded a teaser video , albeit overly dramatic.  Would you expect anything less?

You can find the video on our YouTube page or on our channel HERE

 

IMG_3919.JPG

I will release a video every week, and they will vary, but there will be a process to making rides.  Each video will cover different aspect from start to finish: lift mechanics, brakes, structures, and trains.

For now, enjoy the teaser reel and be on the lookout for new blog and social media posts for backstage happenings.  

 

IMG_3920.JPG

Thoughts Over The Weekend

Since I’m on the verge of this long adventurous process of machining little coasters, I had a few thoughts this past weekend.  One thought was “where to buy material in bulk?”  Typically I have used a few different sources: McMaster-Carr or Online Metals.  The prices vary depending on alloy type, but they have both been relatively cheap.

Now, I may be jumping the gun a bit, but it’s a good idea to plan for increased material costs throughout larger production.  The wood I will be using is also on the list of thoughts. 

When it comes to wood, it’s always better to find a wholesaler rather than frequenting the big box stores.  Wholesale is going to be my route, but I’m not too familiar with ordering from one.  If you guys have any advice on either metal or small scale lumber, please leave comments or reach out to me.

Also, tonight I will be testing the milling machine and possibly releasing a video or two this week.  Check back soon! 

 

And if you haven’t yet, visit the Coaster Shop page and check out or pretty nifty shirts. 

-David