YT YouTube2Text

YouTube2Text → Transcripts → From Catastrophic Failure To Print-Ready RC Chassis

From Catastrophic Failure To Print-Ready RC Chassis — Transcript

by curv lab · 3,419 words · 489 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00For the last 6 months I've been working
0:01on designing a 3D printable remote
0:03control car chassis. I've tested
0:04multiple drive trains. I've gone through
0:06a bunch of different designs and
0:07iterations. I've gotten a bunch of
0:09really good feedback and I've finally
0:11arrived at a design that's easy to
0:12print, easy to assemble, just as robust
0:15as an RC car you'd buy at the store, and
0:17it absolutely hauls ass. So, if you want
0:20to see how I designed it, stick around
0:21for the rest of the video. But first,
0:23the 3D print files are available for
0:25download on my website curvelab.com
0:27along with step-by-step instructions in
0:29a build guide. And there's a complete
0:31hardware shopping list for all the parts
0:33that can't be 3D printed. With that,
0:35let's briefly review. So, in case you
0:36missed my last couple videos, I've been
0:38working on this chassis for a while. The
0:40first prototype was great, but a lot of
0:41the components that I used were still
0:43carryover from the original RC car that
0:45I harvested them from. So, in the next
0:47prototype, I wanted to redesign the
0:48drivetrain to be as 3D printable as
0:51possible. And in doing so, integrated
0:53steel ball bearings into plastic bearing
0:55races within the differentials. I built
0:58up a prototype and took it out to test
0:59on roads and off-roads. And pretty
1:01quickly there was a catastrophic failure
1:03where the center differential grenaded
1:05and threw out all the ball bearings onto
1:06the chassis plate. At the time I was a
1:08little miffed as to why it failed, and
1:10so I posed the question to all of you to
1:11try and analyze what happened. And
1:13thanks to your input, I did figure out
1:15what went wrong. So, first and foremost,
1:17I'm an idiot for not putting lube in the
1:18bearings. Pretty much half of the
1:20comments were like, "Bro, just put some
1:22kind of grease in there." So, thank you
1:23if you commented something to that
1:24effect. I get it that this is a critical
1:26design flaw. But a couple comments in
1:28particular highlighted what I now know
1:30to be the true cause of the catastrophic
1:32failure. The belt drive system needs
1:34tension to function properly. As the
1:36tensioners rotate up and increase the
1:38tension on each of the belts, it creates
1:40a clockwise torque on the axis of
1:42rotation of the center differential. And
1:44if we take a look at the center diff in
1:46section view, that force exerted on
1:48those belt drive pinions compresses the
1:50space for these steel ball bearings
1:52within the plastic races. And because
1:54the plastic is so so softer than the
1:55steel ball bearings, they dig in and
1:57bury themselves into the plastic race
1:59and bind up the whole system. Without
2:01tension, the belts don't work, and with
2:03proper tension, the bearings burrow into
2:05the plastic, as is evident from these
2:07welts. And unfortunately, no amount of
2:09lube or grease is going to solve this
2:11problem. Jason Miles identified this
2:12issue and proposed a solution. He says,
2:15"Add a second bearing outboard of the
2:17center diff belt pulleys, since
2:18currently they're in single shear." What
2:20Jason's talking about is that if I add
2:22another bearing just outboard of each of
2:24the belt pinions, tension on the belts
2:26will no longer cause the center diff to
2:28want to rotate, since they're now
2:30supported on the outside. And you know
2:32what? That would probably work, but it
2:33doesn't solve my core fundamental issue,
2:36which is steel ball bearings in soft
2:38plastic races. It's just a matter of
2:40time before they blow up. And even if I
2:42somehow found a way to redesign these
2:43differentials using off-the-shelf
2:45bearings, it doesn't change the fact
2:47that what started as a simple idea has
2:49spiraled into so much complexity that
2:51it's time to back up and re-examine some
2:54of my old assumptions and see if I've
2:55gone down the wrong path. So, I decided
2:57to go back and take a closer look at how
2:59I could possibly 3D print a drivetrain
3:01that's strong enough to handle the
3:02torque that's going through the system.
3:04Fundamentally, I need to find some way
3:06to connect the motor to the center diff,
3:08and then the center diff to the front
3:10and rear diff. My concern has always
3:11been since 3D prints are grown layer by
3:13layer, and there's inherent weakness
3:15between those layers, that a printed
3:17drive shaft has no torsional strength.
3:19The second issue is that even if you can
3:21print a shaft to be strong enough,
3:22you're still dumping all the power going
3:24through the system through a small bevel
3:26gear. And I just don't trust a 3D
3:28printed part to be able to handle that
3:30load. I trust these spider gears to be
3:313D printed because they only spin at low
3:33speed and intermittently. And the work
3:36is shared between four of them. So, if
3:37my goal is just to have the best and
3:39most robust 3D printable RC car that I
3:42can, maybe I should just source the
3:43diffs off the shelf. I found these GPM
3:46differentials, and there's a bunch of
3:47different sources on Amazon and on
3:49AliExpress listed here in the build
3:51guide, so they're easy to acquire if you
3:53want to build one of these chassis at
3:54home. And they come with a hardened
3:55steel drive pinion, which alleviates a
3:58ton of my anxiety around the robustness
4:00and longevity of the drivetrain. So,
4:02when they came in the mail, I
4:03disassembled them so I could reverse
4:04engineer the whole assembly in order to
4:06add them to my SolidWorks model. A super
4:08helpful tip for modeling spur gears in
4:10SolidWorks is to just dump all the specs
4:12you have into an LLM like Claude. Give
4:15it the number of teeth, the thickness,
4:17the module, the pitch diameter, or the
4:19outer diameter spur gear, and then ask
4:21Claude to generate the parametric
4:23equations that draw the spline and the
4:25involute curve for your spur gears. And
4:27gives you just all the global variables
4:29you need to plug into your SolidWorks
4:31model so that you can have a template
4:32file to create any size of spur gear
4:34going forward. These AI models are
4:36getting so powerful now that I would not
4:38be surprised if in the next year or two
4:40this entire RC car could be generated
4:42just with a prompt. But at least for the
4:43time being, I have to toil my hours away
4:45in CAD. So, now that we have the diff
4:47problem solved, how do we 3D print a
4:49drive shaft to be strong enough? Jello
4:51Mello had the idea to just print the
4:53shafts longitudinally, which is a great
4:55idea. Where printing the shaft
4:56vertically like a column gives you a
4:58bunch of these really weak little layer
4:59lines, flipping the column onto its side
5:02and printing it longitudinally takes
5:04full advantage of the strength of the
5:06filament rather than the adhesion
5:07between layers. However, printing a
5:09cylinder on its side requires support
5:11material to prop up the overhangs. So,
5:13instead I opted to slice the column in
5:15half and print each half without the use
5:17of support material with three wall
5:19loops and 85% infill so that it's
5:21extremely strong. Then the two halves
5:23can be glued together with super glue.
5:25Again, I used Claude to calculate
5:27exactly how thick this drive shaft would
5:29have to be while printed in PETG to
5:32achieve similar torsional strength to
5:33the original steel drive shaft. I
5:35explored some ways to couple that shaft
5:37to the cups on the center differential
5:39and the small hardened steel pin that
5:41drives the front and rear diffs. This
5:43little cross shape plugs straight into
5:45those cups and creates a super robust
5:47linkage between the two. On the other
5:48side of the drive shaft, there's a
5:50little tiny channel cut that fits a
5:51threaded insert and a set screw to lock
5:54it under the hardened steel pin of the
5:55front and rear diffs. I designed the
5:57chassis plate with a spine that runs
5:59down the middle to increase the
6:00longitudinal stiffness of the plate.
6:02That spine continues from tip to tail
6:04even under the differentials, where I've
6:06cut a small triangular channel to allow
6:08those differentials to slide fore and
6:09aft, so the drive shaft can be
6:11installed. On this version, I completely
6:13redesigned the motor mount. eccentric
6:15tensioner that I designed relied on the
6:17clamping force of a PETG part and
6:19vibrated itself out of alignment fairly
6:21quickly. The new motor mount slides
6:23along these channels to accommodate
6:24different size pinions and securely
6:26attaches using fasteners to the plate.
6:28The top bracket is printed in two pieces
6:30and glued together, and is secured over
6:33the center differential with threaded
6:34inserts and some screws. To start out,
6:36I'm running a 12-tooth pinion to get a
6:38little bit more torque at the wheel.
6:40You'll notice the fasteners are securing
6:41the motor just to the top bracket and
6:44not to the bottom. This is because the
6:45channel at the bottom is designed with
6:47an undercut to lock the motor in place.
6:49The differential cases were designed not
6:51just to contain the differentials, but
6:53to be a central bracket that holds a
6:55bunch of different moving parts in
6:56position. The underlying design is super
6:59simple. You take the volume of the
7:00differential and you subtract it from a
7:02block. This creates a nice snug housing
7:04that holds the drive pinion in concert
7:07with the differential. Where it gets a
7:08bit more complex is determining the
7:10exact geometry and alignment of all the
7:12moving parts that attach to the casing.
7:14The angle at which the control arms
7:16attach determine things like toe angle
7:19and caster angle, which could be a whole
7:20separate video in itself, so I'll spare
7:22you from the explanation here until
7:24another time. But one thing I do want to
7:26talk about is how I design complex
7:28kinematics like these suspension arms
7:30and the geometry of the steering rack.
7:31It seems like a daunting thing to
7:33design, and you never really know where
7:34to start. So what I like to do is just
7:36start in the simplest possible form with
7:39just lines sketched out representing
7:41each of the bars in the four-bar
7:42linkage. Here I'm looking at how the
7:44relationship between the length of the
7:46upper and lower control arms results in
7:48changing camber in the wheel along the
7:50full range of motion. If you were to
7:52design a full 3D part, in order to go
7:54back and change the geometry slightly,
7:55you'd have to redesign that whole part.
7:57This allows you to play with the
7:58mechanical systems and really quickly
8:00tune the geometry before locking in
8:02critical dimensions. I tested one such
8:05idea early on in the design process when
8:07I was considering doing push rod
8:09suspension to lower the total height of
8:10the chassis. But through mocking it up
8:13with just simple lines in SolidWorks, I
8:14was able to rule it out fairly quickly
8:16as a viable solution in this particular
8:18case. Before installing them onto the
8:20chassis, I added some 30K diff fluid to
8:22tune the amount of resistance each of
8:24the diffs have. And I should note that
8:26within the diff casing, I'm using white
8:28lithium grease this time so that all you
8:30folks watching don't skewer me in the
8:32comments for lack of lube again. I kind
8:34of felt like I was P. Diddy, only I was
8:35getting canceled for a lack of baby oil.
8:37The suspension towers here are a
8:39separate part because I wanted to be
8:40able to tune and test a bunch of
8:42different geometries as I searched for
8:44the right set of shocks. I sourced a
8:46waterproof 25 kg servo, which has more
8:49than enough oomph to steer the car at
8:50speed, but is compact enough to fit
8:52between the edge of the chassis and the
8:54drive shaft. The steering rack design is
8:56simple. It's another four-bar linkage
8:58where the arms pivot around these posts
9:00that are integrated into the chassis
9:01plate and are secured with a couple M3
9:03screws on top. I'm pretty dumb, but I
9:05learn from mistakes, so I'm using
9:06off-the-shelf bearings for this drive
9:08pinion instead of 3D printed ones. Then
9:11to clamp the top half of the casing to
9:12the bottom half of the diff casing, I
9:14use these 30 mm M3s. To connect the 3D
9:18printed drive shaft to the hardened
9:20steel drive pinion in the front
9:21differential, I used a threaded insert
9:24that allows a set screw to be tightened
9:25against that pin. The opposite end of
9:27that drive shaft has that cross feature
9:29that interfaces with the cups on the
9:31center diff. And once that's slid into
9:33place, it can be locked in that position
9:36by tightening down the screws on the
9:37bottom of the chassis plate that have
9:39slotted holes and allow for some micro
9:41adjustment. At the rear of the chassis,
9:43the drive shaft is much shorter, but it
9:45uses the exact same mechanical design as
9:47the longer shaft. Once the rear diff is
9:49slid into place and tightened down, the
9:52drivetrain is more or less complete. You
9:53can see that the whole chassis plate has
9:55flex to it, and that flex creates an
9:57opening and closing gap between the
9:59shaft and the diff casing. To fix this
10:01issue and to just generally stiffen up
10:03the whole chassis, I use this structural
10:05beam to tie together the top of the
10:07front and rear diff. It has little
10:09alignment pin features that help guide
10:11it into position, and once this is
10:13torqued down, it's pretty remarkable how
10:15stiff the chassis feels. When I really
10:17give it my best effort to flex it, I
10:19can't even get that gap to open up. So,
10:21the drive shaft stays perfectly aligned.
10:23All eight control arms are attached with
10:25these 3-mm e-clip pins, 50-mm and 25-mm
10:29in length. In this case, there's just
10:31really no substituting with a 3D printed
10:33part. You need the strength of a steel
10:35pin, and you'll see exactly why that is
10:37in a moment. And if we were to use a
10:38screw and a nut, it would slowly come
10:40undone over time. So, as far as I can
10:42tell, these are the best possible
10:43solution. The knuckles have a bunch of
10:45small finicky bits, but the build guide
10:47has really detailed and clear
10:48instructions that show you the exact
10:50stack up in assembly. And for the
10:51off-the-shelf parts like the 55-mm axle
10:54and the e-clips, there are links
10:55embedded into the guide to help you
10:57order them. The front knuckles are just
10:59like the rear knuckles, except there's
11:00an extra axis of freedom that allows the
11:02wheels to rotate while steering. So,
11:04there's a few additional parts in this
11:06assembly. The servo is liquidy smooth,
11:08and the drive shafts are sending power
11:09to both diffs. So, I'm ready to put on
11:11the first pair of shocks that I tried
11:12out. And after mounting the wheels and
11:14tires, I was ready to take it out for
11:16its first test drive.
11:18I need to pause quickly here and humbly
11:19request that if you like this channel,
11:21please like and subscribe. You're not
11:22going to want to miss getting alerted
11:24for the next video, where we're going to
11:25take you through a full industrial
11:26design process of designing and building
11:28a car from scratch. Please help us to
11:30not have to get real jobs. Okay, back to
11:32the video. And shout out to the true
11:34fans who know why that front right tire
11:36is purple. The first thing I noticed is
11:37that the springs weren't quite stiff
11:39enough. So, when I cornered hard, I
11:41could hear the chassis plate scraping on
11:42the ground.
11:43And the rebound was a bit too sluggish,
11:45so I was losing traction really easily.
11:50I did about 15 minutes of back and forth
11:52laps, running the chassis through all
11:53the paces, and trying to observe what
11:55was happening to the suspension while
11:57cornering, braking, and accelerating.
11:59Now, granted, mounting my iPhone to the
12:01top gave it a really high center of
12:03gravity and made the handling much
12:05worse, but still it was clear that the
12:07suspension needed some recalibration.
12:09Now, I like to practice a type of
12:11engineering called crash it and then fix
12:13what broke when it crashed. So, at the
12:15end of the session, I decided to run it
12:16into one of the logs and see what broke
12:18first.
12:23Now, you'll recall I mentioned we really
12:25needed to have metal e-clip pins here
12:27instead of 3D printed ones, and this is
12:29the main reason why. Anytime there's any
12:31sort of collision with this chassis,
12:32it's going to send a lot of that force
12:34straight into that pivot point of the
12:35lower control arm. But, here's where a
12:373D printable RC car starts to really
12:39outshine a retail one. Once you have the
12:41files, you can just reprint as many
12:43times as you crash it. And the chassis
12:45is designed to be extremely modular, so
12:47it's only like a 10-minute fix to swap
12:49out the control arm. Since the control
12:50arm was the part that failed in the
12:52crash, I changed the print settings to
12:54be three wall loops and 85% infill
12:57instead of 15, and this thing is so much
12:59more solid now. I made a number of other
13:01design edits as well, like redesigning
13:03the motor mount to be slightly stiffer,
13:04so instead of using that cleat, having a
13:06third attachment point with a screw.
13:08Also, I felt like the gear ratio was a
13:10little bit short, and nothing exploded
13:12when I ran it into that log, which means
13:13we're definitely not going fast enough.
13:15So, I decided to up the drive pinion
13:17from 12-tooth to 15-tooth. And here I'm
13:19tightening down that third screw on the
13:21new motor mount. Final update was
13:23redesigning the suspension to be a
13:25little bit more calibrated to the weight
13:27of this chassis. The shocks I used were
13:29a little bit too soft in terms of spring
13:31stiffness and you could see it
13:32especially when braking and cornering
13:33how the front end would dive and be very
13:35sluggish to rebound.
13:37So, given that I designed this part to
13:39be modular, I just had to change the
13:40geometry slightly, reprint, and then
13:43source a new set of shocks with the
13:44right specs.
13:46The old drop test proved that these
13:48shocks were a much better match.
13:49Immediately, I could tell I liked the
13:51ratio better with the new 15-tooth drive
13:53pinion. I set up this side view camera
13:55mount to try and get a better picture of
13:56what the drive shafts were doing, but it
13:58ended up just being a really cool shot
14:00of the rear suspension at work. I took
14:01it out to some dirt trails to really
14:03test the new suspension and it's crazy
14:05how much better this thing handles with
14:07a properly calibrated set of shocks.
14:09Again, the high center of gravity of my
14:10iPhone mounted to the top was throwing
14:12off the handling a bit, so I was
14:14constantly swapping back and forth with
14:15the camera mount.
14:18On roads, the thing is so buttery smooth
14:20and so fast with the new ratio. I felt
14:23like an F1 pilot and I was actually a
14:24little worried about killing my iPhone
14:25if I crash. So, I was trying to drive
14:27very conservatively when I had my phone
14:29mounted to it.
14:37You can see the front end dive
14:38characteristics are a lot friendlier
14:40with this version. When I slam on the
14:41brakes or corner, it rebounds nice and
14:43smoothly. Another reason I opted for
14:45off-the-shelf links here is so that you
14:47can dial in the toe of the front end in
14:49real time to get perfect handling. So,
14:51if you want to drive one of these
14:52things, you should go over to my website
14:53curvelab.com and download the files. And
14:56the best part is it's kind of like you
14:57have a lifetime subscription to this
14:59chassis because if you crash it, you can
15:01just reprint and replace that part. And
15:03we have a super exciting new project
15:05being released very soon on this
15:06channel. So, if you want to stay
15:08up-to-date, please hit the like and
15:09subscribe button and stay tuned for the
15:11next video where we're going to take you
15:12through a full industrial design process
15:14of designing and building a car from
15:16scratch. Thanks for watching. Until next
15:18time.

About this transcript

This page contains the full transcript of From Catastrophic Failure To Print-Ready RC Chassis by curv lab, generated from the public captions YouTube serves with the video. The transcript has 3,419 words across 489 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.