7433-S Robotics Smart Machine: CNC Parts That Outlast the Kit
The 7433-S robotics smart machine is a Thames & Kosmos HoverBots-style kit with 227 pieces, two-wheel self-balancing, ultrasonic obstacle sensing and Bluetooth block coding. This page is for teachers, makerspace leads and hobbyists who want to replace the parts that fail first. Read it and you can decide which pieces to machine, in what material, and what to measure before you send a drawing.
What Actually Breaks on This Kit
Start with the failure points, not with the whole robot.
Where the Plastic Gives Out First
Most 7433-S kits fail in the same three places. The gear teeth on the drive train round off after a few dozen hours of running, the wheel hubs crack where the axle press-fits into the plastic, and the sensor bracket flexes enough that the ultrasonic reading drifts. None of these are electronics problems. They are tolerance and material problems, and they are the ones a machine shop can solve.
The balance behavior makes it worse. A self-balancing two-wheel robot corrects its tilt many times per second, so every gear mesh and every hub gets loaded in both directions at high cycle counts. Molded plastic handles that load for a while, then develops backlash. Once backlash appears, the control loop has to work harder, the robot wobbles more, and the wear accelerates.
A CNC-machined replacement breaks that loop. Aluminum or POM gears hold their tooth profile, hubs stay round, and brackets stay stiff. The control loop then sees a mechanical system that behaves the way the firmware expects.
- 1Drive gearsRounded teeth from repeated reversing load.
- 2Wheel hubsCracks at the axle press-fit.
- 3Sensor bracketsFlex changes the ultrasonic angle.
- 4Chassis platesScrew holes strip after repeated rebuilds.
How the Drive Train and Balance Loop Work
Two DC gearmotors drive the wheels directly, one per side. A gyroscope reports angular rate and an accelerometer reports tilt; the firmware fuses both and adjusts motor current to keep the robot upright. In the stock kit, motors are rated around 0.5 Nm of torque and the control loop samples near 100 Hz. That sample rate is fast enough that mechanical slop shows up as visible oscillation.
If you replace gears, keep the ratio identical. Changing the tooth count changes how much torque reaches the wheel for a given motor command, and the balance gains in the app are tuned around the original ratio. A machined gear with the same tooth count but a tighter bore is a drop-in improvement. A gear with a different ratio is a re-tuning project.
For the wheel hub, the critical dimension is the bore that receives the axle. A press fit that is too loose spins on the shaft; too tight and the hub splits during assembly. On a milled hub we hold that bore to ±0.005 mm, which lets you press the axle by hand with an arbor press and no heat.
Sensor geometry matters more than people expect. The ultrasonic module has a cone-shaped detection field. Tilt it 5° off axis and the effective range against a flat wall changes noticeably. A rigid machined bracket keeps the module square to the chassis, so the distance readings your code receives stay consistent across builds.
Where Precision Parts Change the Programming Experience
Block-based coding through the Bluetooth app teaches the same ideas a PID loop uses in industry: proportional response, integral correction, derivative damping. Students tune three numbers and watch the robot's behavior change. When the mechanics are sloppy, those three numbers cannot fix the wobble, and students conclude the code is wrong.
That is the real argument for machined parts in a classroom. A stiff chassis and a low-backlash gear train make the tuning knobs behave predictably. Turn up the proportional term and the robot stiffens; turn up the derivative term and the oscillation damps out. Students learn the concept instead of fighting the hardware.
Bluetooth 4.0 links run at short range and the app reconnects often. If your enclosure blocks the antenna or the module sits in a metal box, range drops. Keep the radio module on the outside of any aluminum shell, or leave a plastic window in front of it. We machine those windows as pockets with a 1.5 mm minimum wall rather than cutting a hole, which keeps the shell rigid.
Material Choices for Upgrade Parts
Match the material to the load, not to the look.
| Part | Recommended material | Why | Watch out for |
|---|---|---|---|
| Drive gears | POM (acetal) or 7075 aluminum | Low friction, holds tooth profile | Aluminum gears need lubrication |
| Wheel hubs | 6061-T6 aluminum | Stiff, presses cleanly, light | Anodize before pressing the axle |
| Chassis plates | 6061-T6 or 5052 aluminum | Good stiffness per gram | Thin plates chatter when milled |
| Sensor brackets | 6061-T6 aluminum | Rigid, keeps the module square | Leave clearance for the cable |
| Bumper guards | POM or PC | Takes impact without denting | PC scratches, POM does not |
| Arm links | TC4 (Ti-6Al-4V) titanium | High strength, low weight | Costs far more than aluminum |
Machining These Parts: What the Shop Needs From You
You do not need a full CAD model to start. A measured sketch with the critical dimensions marked is usually enough for a quote. Photograph the original part next to a ruler, note the tooth count and pitch on any gear, and mark which fits are press fits versus running fits. Those three pieces of information decide the process.
Gears are the one part where the drawing has to be precise. Tooth form, pitch diameter and bore all interact. If you can measure the original with a caliper and a pin gauge, send those numbers. If not, send the gear and we will reverse-engineer the profile.
For flat plates and brackets, 3-axis milling covers almost everything. Plates under 500 × 500 mm are routine. Parts with features on five faces, such as a hub with a bore, a flange and cross-drilled holes, go on one of our 16 simultaneous 5-axis machining centers and come off in a single setup, which keeps the bore and the flange concentric.
Prototypes run from one piece upward. There is no minimum order quantity, so a single replacement hub or a full set of gears for a classroom of thirty is the same workflow. Quotation and a free DFM review come back within 12 hours, and production can start within 24 hours of approval. Parts ship in 3–5 days.
Finishing is usually simple for robotics parts. Clear anodizing on aluminum adds wear resistance without changing dimensions much. Hardcoat anodizing is worth it on sliding surfaces. Keep laser marking off any mating face; the marked area sits slightly proud and can affect a press fit.
- 1SendSketch, photos and the critical dimensions.
- 2MarkWhich fits are press fits and which run free.
- 3NoteTooth count and pitch on any gear.
When Machined Parts Are the Wrong Answer
Not every kit problem deserves a machined fix. If a part broke because a student forced it during assembly, a stronger part will just move the failure to the next weakest point. Fix the assembly instruction instead.
If you are still deciding whether this kit is right for your program, buy the kit first and run it for a semester. Note what breaks. Then machine only those parts. Ordering a full aluminum robot before you know the failure pattern usually wastes money on parts that were never going to fail.
Weight is the other limit. Replacing every plastic piece with 6061 aluminum can add enough mass that the motors struggle to correct tilt quickly. Keep the upgrade targeted: gears, hubs and brackets. Leave the large covers and decorative panels in plastic.
Finally, if the goal is a one-off demonstration rather than a reusable classroom set, the stock kit is fine. Machined parts pay off when the same robot has to survive many students and many rebuild cycles.
Questions Engineers Ask Before Ordering
Can you machine a replacement gear without a CAD file?
Yes, in most cases. Send the original gear or clear photographs with caliper measurements of the outer diameter, bore and tooth count. We measure the tooth profile and build the model.
If the gear is worn, the wear shows in the measurements. Tell us where the tooth contact pattern sits so we can distinguish original geometry from damage.
What tolerance do you hold on press-fit bores?
We work to ±0.005 mm on critical bores and shafts. That is tight enough for a hand-pressed axle in an aluminum hub without heat or adhesive.
Tell us the fit you want in words as well as numbers. A slip fit and a press fit need different allowances, and the drawing does not always say which you intended.
Will aluminum parts make the robot too heavy to balance?
It depends on which parts. Gears, hubs and sensor brackets add little mass and remove the slop that hurts balance.
Replacing large plastic covers with aluminum is where weight becomes a problem. Keep total added mass modest and the balance loop will not notice.
How fast can I get a small batch for a classroom?
Quotation and DFM feedback come back within 12 hours. Production can begin within 24 hours of approval, and parts ship in 3–5 days.
There is no minimum order quantity, so one piece and thirty pieces follow the same process.
Do you sign an NDA for a school project or a product idea?
Yes. Uploads are handled as confidential and an NDA is available on request before you send files.
This matters if the upgrade is part of a product you plan to sell rather than a classroom set.
Which materials should I avoid for gears?
Avoid soft, gummy plastics that cold-flow under load. They deform at the tooth contact and backlash returns quickly.
Unfilled nylon absorbs moisture and changes dimension. POM holds dimension better in a classroom where humidity swings.
Send the Part That Keeps Breaking
Sketch, photo or CAD file. You get a quote and a free DFM review within 12 hours, and every part is inspected before it ships.
12-hour quote±0.005 mm tolerance3–5 day shipping
