Formula Hybrid


Formula Hybrid

The KiWi 1

KiWi 1 2nd Place General Motors Award for Best Hybrid System Engineering

KiWi 1 2nd Place Best Engineered Hybrid Systems Design

One of the most challenging, time consuming and fun projects Ive been able to work on was building the Yale Hybrid Formula car. We placed 2nd in design at the 2010 Formula Hybrid competition. Not bad for a team of 4 and $17k! I’d like to describe here how that was accomplished.

My Background

I’ve been a gear-head all my life, mostly outside of the classroom. I spent many summers on a farm in the lake and forest Masuria area of Poland, where my time was spent repairing farm equipment from tractors to combines, motorcycles to barns all while waking up at 6am to milk the cows and moving thousands of square straw bales onto a trailer only to throw them off at the barn during harvest. Think of Ronald Regan, his way of relaxing was intense physical labor on his ranch, which describes my method of relaxation exactly.

At Yale I finally had the time (well not really, but I managed somehow) as a graduate student to work on a project that would require the merging of my computer science, electrical and mechanical engineering skills to build a hybrid vehicle from scratch. Building such a vehicle is a practical challenge, where without previous experience with machinery, complex systems and knowing how they fail and break (I’ve seen lots on the farm, machine shop at RPI and driving my old motorcycles which always had a surprise far from home) its pretty difficult to build something reliable reliable. Its baffling but pretty common, even in consumer products and especially in prototype graduate projects, to see a lack of understanding of a complete system and ending up with wires right next to a heat source, a lack of lock nuts in high vibration areas, and insufficient factor of safety due to not foreseeing the combination of heat expansion, torque and ‘someone stepping on it’

Another common misconception when working on a hardware problem, for those that haven’t, is the amount of time needed to complete a project and the presumed simplicity of a component. Everything is simple when it is not understood, and calling, fo ex., the pedal assembly, knuckles or brake system simple and easy to manufacture stems from having minimal design, machining and exploitation experience. You can BS a written document but there is no way you can BS hardware, its either designed and built to quality or garbage, and making a quality part requires countless hours of design, foreseeing possible scenarios, redesign and prototyping. Hence even though the pedal assembly might seem simple, the amount of moving parts and critical dimensions for bearings or bushings makes it a complex part that must function flawlessly even when pounded by the driver under sudden braking. Machining those critical dimensions to prevent binding is a time consuming task. Not to mention knowledge of different fits: clearance, interference; whether to use unsealed, double rubber or metal shield, roller or ball bearings; aluminum, steel; the number of fasteners holding the piece together and the countless milling, lath and welding operations needed depending on the design. Which is why I gladly allocated the $ for a set from Tilton instead of wasting time on something someone has designed and perfected already.

Alternator Drive Assembly

Alternator Drive Assembly

The alternator mount for a powermaster alternator

The alternator mount for a Powermaster alternator

Ninja 250 Flywheel on crankshaft

Ninja 250 Flywheel on crankshaft

The alternator drive assembled and Alternator with mount

The alternator drive assembled and Alternator with mount

Yale 2008 – Torquere et Velocitas

In March of 2008 I first met the Yale Formula Hybrid team, which had already built and designed a car for that years competition, a test mule allowing for testing of parallel or series drive trains. Given that the Formula Hybrid competition was only about 6 weeks away, and the car was built, the most i could do was to asses the situation and help the small team of 3, in bursts up to 6, complete the project. The main missing component was a way to charge the car while driving. Since the design was a parallel hybrid and the electric drive was a series wound 72V DC motor I couldn’t use it to charge, so the quickest and most effective way I determined would be to drive an alternator using the Kawasaki Ninja 250 motor.

Some other suggested ideas were to somehow attach the alternator to the final drive. This had a significant disadvantage of only charging while the car was moving, and at a decent speed. Ive seen alternators attached to the crankshaft on a different bike, the Junak and it seemed that will be fairly easy to do with the ninja also. The process involved modifying the alternator cover, drilling and tapping two holes in the flywheel, machining a drive shaft and cover insert to hold a bearing and seal. Let me describe each piece:

  • Alternator Drive Shaft Threaded on one end to replace the bolt holding the flywheel and tapped on the other to retain the alternator pulley. Given that the alternator was a 100Amp 12V unit, it required a decent amount of torque to drive it at full capacity.power: 100A*12V=1.2kW
    torque: T=1.2kW*33000/(2*\pi*2500RPM)=4.58ft*lbs 2500RPM is where the alternator almost reached full charging current. Given that the pulley ratio could of been up to 2:1, the crankshaft would be spinning slower then the alternator we have 4.58ft*lbs*2=9.17ft*lbs of torque at the drive shaft. Since the shaft was made of a steel similar to SAE 2330 (unhardened) and the treads were threaded with a die (instead of rolled) I wasn’t exactly confident with the strength of the threads and the last thing I wanted was for them to strip and the flywheel to loosen and slam into the alternator stator or do some other major damage. To reduce this risk notice the two holes and shoulder bolts that screw into the Ninja flywheel, preventing the shaft from over torquing. Everything was assembled using red locktite.The Ninja engine red-lines at 14k RPM which is above the specs of the alternator, for the few miles and short bursts up to that RPM range the alternator had no choice but to hold its breath and not self destruct, there were no resources for any other fancy solution.If your wondering how the shaft was tightened, the flared end was milled down to allow tightening with a wrench.

    The shaft was mostly machined on a lath, with the flared end milled on a vertical mill. The threads needed to be cut true, harder to do then most realize! I placed the shaft in a collet in a vertical mill, the die in a centered 3jaw chuck and cranked the chucks handle to rotate it and cut the threads as true as possible.

  • The Alternator Cover Insert held the bearing, spring seal and had a grove for the retaining ring (per original plans). However, I didnt use any retaining rings but instead left a lip on the outer side of the insert which didnt allow the seal to fall out. The shaft had a step which prevented the bearing from moving closer to the flywheel and thus put together the bearing and spring seal were locked in position. The bearing here served to prevent shaft bending under load from the belt tension, and the seal obviously kept oil from leaking.
  • The Ninja 250r Alternator Cover was modified to fit the cover insert which held the bearing and seal. The threaded plastic cap on the cover was removed and the hole was machined on a CNC vertical mill to allow for a press fit of the cover insert.
  • The Ninja 250r Flywheel was modified to enable the screwing in of two shoulder bolts to prevent the alternator drive shaft from over torquing. As you can imagine drilling and tapping the hardened flywheel was a very painful process with end-mills and drill bits snagging on the steel and loosing teeth. Tapping was very slow, no more then a quarter turn before backing back out while using plenty of tapping fluid. Its important to note that these holes could not be placed at just any angle, they had to match with the holes drilled in the alternator drive shaft. While the flywheel was still on the engine I torqued the shaft on and marked the locations of the holes on the flywheel.
  • Alternator Mount The alternator had to be electrically isolated from the low voltage system or it would trip the ground fault detector (a SAE rule requirement). Thus the mount was made of fiberglass. A pretty simple part that allowed for adjusting the belt tension and slots for the belt. Its shape was mostly determined by the location where it was to be mounted on the car.

The alternator was meant to charge a single 12V battery, so its output was directed to a DC-DC converter which charged the 72V battery pack on the car. A less then optimal setup as you have mechanical losses from the belt slipping, extra weight of the alternator, a less then efficient DC-DC conversion and plenty of heat generated. That’s one reason I wanted charging to be done via the drive motor on the design I was slowly conceptualizing.

Yale Formula Hybrid - The KiWi1 engine box

The KiWi 1 (KiWi One)

After the end of the 2008 academic year I was left for the summer alone with the project. To summarize the year: I, Kamil Wasilewski, motivated a past member, Henry Misas, who raised all our funding for the coming year, and together we found a few more people that were key to the teams success: Alicia Fernandez and Jonathan Biagiotti. We placed 2nd in design at the 2010 Formula Hybrid competition. Not bad for a team of 4 and $17k! I’d like to describe here how that was accomplished. There were a number of other members we trained, and had attend meetings, but a small tightly knit group for this project was very effective during the design phase but a bit undermanned when it came to producing the completely modeled car. There was a significant amount of fabrication where appropriately trained hands would of helped greatly, on the other hand communication was excellent and little time was wasted conveying ideas.

The Yale Daily News ran an article about the team, unfortunately after I had already left Yale: Yale team racing to glory

KiWi 1 Specs
Gas Engine
Electric Motor
Final drive

SAE Formula Hybrid Rules 2009

SAE Formula Hybrid Rules 2009

Define the problem

With such a challenging and complex task such as building a hybrid vehicle the problem, goals and limitations should be described and visible for all to see in order to stay focused and create a matching solution, instead of designing the drive train for gobs of torque when the rest of the vehicle is designed to light and nimble. There were three main design drivers: the 2009 SAE rules, physical limitations and the teams strategy for the car. Its important to note that this is a real problem, not an academic one, you have to be realistic in your goals and wants.

The SAE Rules: To compete the car must meet the SAE safety and design requirements. To minimize the possibility of not complying with the rules is to just memorize the rulebook… Well its easy to forget specific requirements and truthfully not everyone will read carefully sometimes misinterpreting a given rule. To mitigate this problem I created a rule summary on a 24?x15? sheet which was much easier to read than the 87 page rule book provided by SAE. In this way I knew each team member knew the rules, they were also posted everywhere for quick reference while designing.

Physical Limitations: We had 6 12V Odyssey PC625 @18Ah dry cell batteries at 6kg each. We had insufficient resources to use lithium battery pack, besides the lead dry-cells could take a lot of abuse and misuse.

Time was another major limitation. Given the amount of time and resources at hand they should be allocated smartly, we avoided reinventing the wheel where possible. Besides we were not creating new technology but using proven components. Hence like most of life the problem ended up being an interface problem: How do we interface the motor and engine with the wheels and the driver with the engine and motor?

Team Strategy:Given the table below describing the points allotted for each event of the competition you notice the endurance event is worth considerably more then the other dynamic events. This should be a major clue that the car should be designed for reliability, ie well built, and efficient, ie lightweight. This goes against many first gut feelings to slap on the largest most powerful electric motor and battery pack possible, unless your main goal was to win the acceleration event which is worth considerably less then the autocross where a nimble car and the endurance where an efficient one will have the advantage.

Static Events
Presentation 100
Engineering Design 200
Dynamic Events
Acceleration – Electric 75
Acceleration – Unrestricted 75
Autocross 150
Efficiency & Endurance 400
Total Points 1000

Hence our main goals were:

  • Durable and well built car. All machined parts had to undergo QC, to be consistent I was the sole QC. I set a tolerance of .005? and .5deg, anything outside of the tolerance range had to be remade, this ensured a high, and defined, standard among team members and well fitting parts.
  • Light weight. Staying light gives many advantages, the lighter you are the lighter you may become. In other words, using light components such as a low mass motor and engine the supporting structure can be lighter. Less power is needed to accelerate the vehicle, hence raising efficiency.

In short we applied the KISS principle.

The competition provided us with a scope and to see where we can have and advantage. One clear area was quality and reliability. Where lots of teams fail is the machining and assembly quality which leads to many problems at the race. No part was placed on the KiWi 1 that did not meet the set quality standards mentioned above, and looking at the car this high standard shows from the beauty of the TIG welds to the precise clearance or interference fits of fairly complicated sub systems and the overall clean well fitting look. One reason the high standard was easy to maintain was our tiny team and one main designer.

To the right were the hard numbers the competition provided when I began designing the car. The events were won by two different teams, one had a massive motor while the other was light and had a much smaller motor. As mentioned, I didn’t see much sense in burning rubber for the acceleration event if it was to mean doing poorly in the endurance.

The 2008 Competition
Acceleration Endurance
winner: 75m in 4.994s
Giving an acceleration of 6.014m/s2
Velocity at 75m of 30m/s
0-100km/h time of 4.619s
They weigh 360kg.
W = (360*9.8)75m/4.9s
W = 54000Watts
2008 winner: 22km in 33:45.396
First 11km in 16:41.834
Second 11km in 16:59.562
Average velocity: 10.86m/s

Divide and Conquer

How do you even start conceptualizing, never mind actually building a complex system where every component is dependent on the design of surrounding components? Just think about it for a minute, most reactions vastly underestimate the complexity, can you imagine every washer, bolt, ball bearing and rod end that make up the front right suspension, suggest a material and thickness for the A-arms given the expected mass and performance of the vehicle, then assemble the knuckle, hub, bearing, disk brake and caliper, oh and I forgot about the steering arm and tie rod? Once you have the components there’s also the task of designing the geometry of the front suspension: Kingpin inclination, Ackermann, toe, anti-dive, bump steer, camber, roll center, instant center. That only leaves the shock, belcrank, connecting rod and spring rate. That’s only the front right suspension! What about the brakes, frame, drive train, electrical and underestimated cockpit controls? And you cant design those separately because the frame depends on the suspension which depends on where the engine is which depends on the driving dynamics you want to achieve which depends on the suspension… Are you getting a taste for how complicated and how involved designing and building a car like this is? An absolute anal retentive attention to detail is a must, you are designing to the washer, whens the last time you actually noticed the flat and spring washers and where they go on a piece of machinery?

To start you have to set goals as I did in “Define the Problem” above. The next big decision is what will hold the entire car together a honeycomb/carbon fiber or space frame design? In our case carbon fiber was out of scope. Hence a metal frame, I already hinted at a space frame design which is strong and lightweight and was used for quite some time in F1. What material: aluminum, carbon steel, 4130? A TIG welder was at our disposal hence 4130 was a good choice, much easier to weld than aluminum and stiffer than carbon steel. Sure aluminum would be lighter but the cost and time needed outweighs those benefits.

A major challenge is where to start, which sub component to tackle first? The proper way to start is with the wheels and suspension. However, that makes it hard to design the cockpit and drive train if a concept for those doesn’t exist. So i started with the frame, which made it challenging to design the suspension as the attachment points were guessed at which limited the amount of freedom I had during suspension design, but it was much easier to design the rest of the car. Note, this was a series not parallel process, ie I designed and built the frame then brakes then drive train and suspension so when one component was finished all others were built to fit.


Guessed Brake System Variables
Total car weight 500kg (actual was much lower)
Front/Rear weight distribution 45/55
Wheel Base (m) 1.651
CG Height (m) .508
Tire Size 21?
Tire Coefficient of friction 1.2
Brake Pad Coefficient of friction .34
Force of foot on pedal 75lbs

Define the problem: SAE rules require a braking system that locks all four wheels and a master cylinder for the front and another for the rear brakes. The obvious choice is a hydraulic disk brake system, but which one? To determine what can be used a few variables need to be defined. I could only guess at what those will be since nothing was designed yet, hence my estimates were on the very safe side based on previous year designs by other teams. None of these were set in stone and changed significantly, but the combination of changes left an acceptable net change in the final results.

Skipping ahead I chose a motorcycle rotor and caliper, steel braided brake lines and fittings, a Tilton master cylinder and pedal assembly. All those are attached to the car in a fail safe way. The brakes are life critical on the car, this system can not fail, which is one reason for having 2 master cylinders: if one part of the system fails the other will still provide some braking force. However having two master cylinders means the braking force you apply at the pedal is split between the master cylinders (front and back calipers). This is where a brake bias bar comes in handy as you need more brake force on the front since when you start braking there is a significant weight transfer to the front as the rear lifts.

So what we have is a 4 wheeled car weighing up to 500kg with sticky tires that needs to be brought to speed and when the driver slams the brakes all four wheels have to lock. The secret to do that is to find the force that the tires exert on the ground and make sure the calipers exert a greater force on the rotors. You squeeze the calipers harder by raising the PSI in the brake lines which depends on the master cylinder and brake pedal. Unless you have superman driving the car you need to remember that the foot strength of the driver is limited. I measured a this force by placing a scale against a wall, myself on the floor leaned against a heavy cabinet and pressed to find a the force I can exert, its rather low and on the safe side.

Selected Component Characteristics
Caliper piston diameter 32mm
Pistons per caliper 4
Pedal ratio 5.5
Front brake Bias 0.65(adjustable)
Rotor diameter 220mm

Reduce the unknowns: With the given weights and coefficients we still have to choose the rotors, calipers, master cylinder, brake lines and brake pedal. The rim size defines the size of the disc we can use remembering that the caliper also has to fit over the disc in the rim. The variables hint at a motorcycle brake system, killing a few birds with one stone. Most sport bikes are in the range of 200kg wet, with a driver and passenger that’s over 350kg so 500kg over four wheels is very acceptable. Sport motorcycle brakes are already designed to be light and powerful in a small package. The parts needed are the disk and the caliper as the master cylinder needs to be custom selected based on performance expectations. A trip to a local dealer with a caliper and tape measure is how I found our set. The front rotor on bikes is larger then the rear, too large in fact to fit into our rim, but the front caliper has more and larger pistons then the rear. A hybrid of the two would fit into our hybrid. Meaning the clearance between the top of the caliper and rotor will be smaller to compensate for the reduced radius of the rear rotor, we want the entire area of the pads to be in contact with the rotor. The winning combination was a front caliper from a ’05 Yamaha R6 and a rear rotor from a ’05 Kawasaki zx14 which has an OD 23cm(9.05?) Bolt Diameter 12cm and 4 bolt pattern.

To actuate the brakes a master cylinder and pedal assembly need to be selected. I wrote in the beginning that designing a pedal assembly from scratch is not worth the effort, hence a ready made assembly from Tilton, a 3 pedal 600-series floor mount fit the bill.

Solve: After the above selection all that’s left to pick are the master cylinders and brake lines, as well as some fittings and washers. The selection of a master cylinder is a five step process. First we need to find the weight transfer under braking. Then we find the force of the tires given their coefficient of friction. Next we do a simple torque calculation to find the force needed to exert on the rotor by the caliper to exceed the force of the tires. After we find the force the caliper must provide we can find the pressure in the brake lines needed to obtain that force. Finally we find the ratio of master cylinder diameter to caliper piston diameter that will give us that pressure.

Weight Transfer Under Braking
W_f (kg) Front weight
W_r (kg) Weight rear
d Static weight distribution on front tires
W (kg) Total car weight
g_b Braking acceleration in GE’s (.8, 1..)
h (m) CG height
w (m) Car wheelbase
Force of tires
F_{Ftire} (N) Friction force of front tires
F_{Rtire} (N) Friction force of rear tires
\mu Coeficient of friction between tire and ground
g (m/s^2) Force of gravity
Tire and Rotor Torque
\tau (Nm) Torque
r_{tire} (m) Tire radius
F_{tire} (N) Force of tires exerted on the ground
r_{rotor} (m) Rotor effective radius
F_{rotor} (N) Force of caliper exerted on the rotor
Brake line pressure
P (PSI) Brake line pressure
A_{caliper} (m^2) Caliper piston area
\mu Coefficient of friction of brake pads
Master cylinder diameter
F_{foot} (lbs) Force applied by the driver on the brake pedal
A_{masterCyl} (in^2) Area of master cylinder
R_{pedal} The pedal ratio
R_{frontBias} The front brake bias
    1. Weight Transfer Under Braking

Under braking the weight of the car shifts to the front changing the weight distribution and hence the normal force of the tires and the ground. We need to calculate this weight transfer to later use these weights to figure out how much frictional force the front vs back tires generate. This is a moment calculation. The weight on the front wheel under 1g of braking
W_f = d*W+g_b*W*(h/w)
The rear wheel weight will be:
W_r = (1-d)*W-g_b*W*(h/w)
Check: The sum of W_f and W_r under 1g of g_b should be equal to W.


    1. Force of tires

Given the weight of the front and back wheels under braking we calculate the friction force of the tires in Newtons. Since the formula tires are rather sticky we can assume a high coefficient of friction. Ive seen other calculations that omit this step and jump straight to the torque calculation, they are obviously assuming a \mu of 1 and doing the calculation in ft lbs instead of Nm so they have no need to multiply by g=9.8m/s^2:
F_{Ftire} = \mu*W_f*g
F_{Rtire} = \mu*W_r*g
where \mu is the coefficient of Friction of the tire.


    1. Tire and Rotor Torque

The force of the tires and the ground creates a torque about the axle, as does the force created by the rotor and caliper. To lock the wheels we need to generate a larger force at the caliper than the tire; and since the tire radius is larger then the rotor, we need to generate even more force at the caliper to compensate. To imagine why, think of a spinning bike tire and how easy it is to stop by grabbing the tire and how much harder it is to do so touching the spokes near the center.
Torque is found by
\tau = r*F
Setting the above formula equal to itself and doing some algebra we can find the force needed at the rotor:
r_{tire}*F_{tire} = r_{rotor}*F_{rotor}
Now solve for the force needed at the caliper to match the force generated at the tire:
F_{rotor} = \frac{r_{tire}*F_{tire}}{r_{rotor}}
Note: The rotor radius (r_{rotor}) needs to be the effective radius not the actual rotor radius. The effective radius is the radius of the rotor to the center of the brake pads. So if your rotor radius is 4.5? and your brake pads are 1.5? in height then the effective radius will be 4.5in-1.5in/2 = 3.75in. A way to think of it is the force is being applied at the center of the brake pads, at a smaller radius on the rotor than the outside edge.


    1. Brake line pressure

Next Calculate the Line PSI needed to stop the car, in other words you have to exert a force on the brake rotors greater than the tires exert on the ground. Since the calipers will be exerting this force on the rotors we multiply the area of the caliper by its coefficient of friction, essentially reducing the area since \mu is less then 1
since we calculated the force needed at the rotor to lock the wheels (F_{rotor}) in the torque section, we can solve for the pressure needed in the brake lines:
The above formula uses a conversion factor 145.04*10^{-6} to obtain the answer in PSI. Without the conversion we would obtain the pressure in Pascals (N/m^2) as the friction force is in newtons and the caliper area in m^2.


    1. Master cylinder diameter

From above we have the pressure needed in the brake line to generate a force on the rotor that will match the force of the tires and the ground. So we need to figure out the line pressure that we generate when the driver slams his foot on the brake pedal. The brake pedal multiplies the drivers foot force by the “Pedal Ration” (R_{pedal}). This force is split between the two master cylinders in the system by the bias bar (if installed) where usually the front master cylinder gets a higher percentage of the foot force applied. Multiplying R_{pedal}*F_{foot}*R_{frontBias} gives us the pounds of force generated, to find the pressure (PSI) we divde that by the area (in^2) of the master cylinder. Master cylinders from Tilton come in fixed sizes (5/8?, 7/10?, 3/4? 13/16?, 7/8?, 15/16?, 1?, 1 1/8?), so we plug the sizes into this formula:
P_f < \frac{R_{pedal}*F_{foot}*R_{frontBias}}{A_{masterCyl}}
Now select a master cylinder size that will give a PSI on the caliper greater than the PSI needed to lock the wheel as found in “Brake line pressure”. The pressure generated should be larger then the pressure needed as the system is not 100% efficient.

Equation Summary and Example
In summary the equations to calculate the PSI and master cylinder needed to lock the tires are:

Finding a Master Cylinder Summary
Front Tires Rear Tires Description
W_f = d*W+g_b*W*(h/w) W_r = (1-d)*W-g_b*W*(h/w) Weight transfer under braking
F_{Ftire} = \mu*W_f*g F_{Rtire} = \mu*W_r*g Friction force of tires
F_{Frotor} = \frac{r_{Ftire}*F_{Ftire}}{r_{Frotor}} F_{Rrotor} = \frac{r_{Rtire}*F_{Rtire}}{r_{Rrotor}} Torque calculation to find force needed on rotor
P_f=\frac{F_{Frotor}}{\mu*A_{Fcaliper}} * 145.04*10^{-6} P_r=\frac{F_{Rrotor}}{\mu*A_{Rcaliper}} * 145.04*10^{-6} Pressure needed in brake lines to generate force on rotor
P_f < \frac{R_{pedal}*F_{foot}*R_{frontBias}}{A_{masterCyl}} P_f < \frac{R_{pedal}*F_{foot}*(1-R_{frontBias})}{A_{masterCyl}} Find master cylinder to give a higher pressure

I always find it immensely clarifying when i see a real world example so here is one:

Finding the Front Master Cylinder Example
W_f = .45*500kg-1*500kg*(.508m/1.651m)
W_f = 378.84 kg
The front axle becomes heavier under braking
F_{Ftire}=1.2*378.84 kg*9.8m/s^2
F_{Ftire}=4455.23 N
Since the tires are sticky we multiply by \mu to get the frictional force of the tires
F_{Frotor} = \frac{.2667m/2 *4455.23 N}{.220m/2}
F_{Frotor} = 10801.91 N
Given the force of the tires we can find the force needed on the rotor doing a torque calculation. The rotor radius (r_{rotor}) should be the effective radius, or the radius of the rotor measured to the center of the brake pads not the outer edge of the rotor.
P_f=\frac{10801.91 N}{.34*.0257m^2}*145.04*10^{-6}
P_f = 716.19 PSI
Use the sum of the area of both front calipers.
P_f < \frac{5.5*75lbs*.65}{(\frac{5}{8}in/2)^2*PI}
716.19 PSI < 873.95 PSI
Notice the area of the master cylinder is in in^2. The system is not 100% efficient so the pressure generated should be larger than needed.

Putting it together: With all the calculations completed and components selected the entire system still has to be mounted, and mounted properly as any failure can mean serious injury or worse, so we are currently only at the halfway point.

Rotor and mount

Rotor and mount – Notice the mount lips which seat the rotor concentric to the mount as the bolts should not carry all the shear force.

Caliper mounted to knuckle

Caliper mounted to knuckle – The stainless steel caliper positioning ring also carries most of the loads under braking.

Tilton pedal assembly and mounting plate

Tilton pedal assembly and mounting plate – The plate’s irregular shaped cutouts were based on the pedal mounting base. Tilton provided the pedal assembly CAD model which made aligning the mounting holes a breeze.

Drilling of grade 8 bolts for safety wire

Drilling of grade 8 bolts for safety wire, a painful but sensible process. Putting safety wire through the drilled holes and attaching it to a fixed point prevents the bolt from completely unscrewing in case the locktite applied to the bolt failed.

—Rotor Mount The rotor mount was designed with the drive train. It took some time to clarify how exactly the wheel and hub assembly will look like in our case. A custom hub was out of the question due to the high cost (we were quoted a minimum of $1k per part) of machining a spline for the CV shaft, so the hub and rotor mount could not be one piece. After finalizing a hub the rotor mount could be designed. A critical dimension of this part was the thickness which positioned the rotor properly with respect to the caliper. Initially I designed the part out of 420 stainless steel as it would be similar to the surrounding parts, as a fair amount of heat is generated while braking, and actually attempted to machine it. After something around 20 hours, 4 carbide end mills, a few drill bits and countless headaches I had a beautiful rotor mount with holes that had countersunks offset from center as the plate moved slightly during machining on the CNC vertical mill when one of the end mills became blunt. So I redid the Finite Element Analysis using 7075-t6 Aluminum and within 4 hours had all four rotor mounts machined. Sometimes its just not sensible to do the sensible thing.

The rotor was attached to the mount using the original bolts used to mount it to a Yamaha. The entire car was therefor a slight mix of metric and SAE bolts due to some of the OEM mixed and matched parts. Drilling and tapping metric holes was the only setback as many machine shops don’t have metric tools so we had to purchase our own. The rotor bolts were shouldered, hence the countersinks in the rotor mount, so when the brakes are applied the shear force on the bolt is not on the thread. The lip by each of the four mounting holes for the rotor is a copy from a motorcycle hub which also has a similar lip. It removes some of the stress under braking from the bolts.

—Caliper Mount The caliper needs to be stationary and so it is mounted to the knuckle. I have heard of debates frowning upon positioning the calipers above the axle due to this raising the CG height, I had not accounted for the argument before designing the mount, and it would of not changed my initial idea, as i find that many such type of arguments are nonsensical. If its to mean that designing the knuckle will take 10hours more just to lower 2lbs of caliper 3? is ridiculous, mount the fuel tank lower and shave instead. That said, I did mount the calipers below the axle. Its important to design the mount so that the caliper is “pushed” into the knuckle instead of pulled out leaving its bolts in tension under braking. Having the caliper push against the knuckle is much more robust then relying on the caliper bolt threads to support the tension force. Hence for a wheel moving counterclockwise, looking at it from the inside, the caliper should be either at the 12 to 3 o’clock position or 6 to 9 o’clock.

The positioning of the caliper relative to the rotor and the clearances between them and the knuckle are critical and offer little room for error. The caliper was measured and remeasured while modeling it in SolidWorks as well as the rotor. In SW I was able to position the parts to provide enough clearance for when the parts expand under braking and align the caliper so that the rotor is placed symmetric to the brake pads.

—Pedal Assembly Mount While designing the mount for the pedal assembly I had a goal of providing a simple way of adjusting the pedal position relative to the drivers foot. The pedals could of been mounted to the thin sheet metal floor, but to mount pedals that take the pounding of a frightened drivers foot is suicide. A few concepts came to mind, including one where the adjustability would be similar to a telescoping pole with a spring loaded pin or something similar to that found on bench presses (picture on right). The simplest was however to use a bolt. Come to think of it the weight equipment type of fitting would of worked here, by replacing the bolts with wire lock pins, but it wasn’t easy to fit those in plus they can leave a fair amount of play making the pedal assembly feel sloppy.

The mounting tab (x4) was welded to the frame. Since this should be quickly adjustable having nuts that fall someplace hard to find when adjusting the pedals a rod was tapped and welded to mounting tab so the only step needed was to loosen the bolt. Many motorcycle manufactures use this fixed nut technique to avoid mechanics cursing at them for every small nut being lost or hard to hold in position while inserting the bolt.

It should be obvious that Finite Element Analysis in SolidWorks was performed on this part and the tabs to optimize the strength to weight ratio. The tabs used were the same holding the a-arms to the car.

—Lines, bolts, fittings etc To route the fluid from the master cylinders to the calipers I chose flexible steel braided lines. These are much easier to work with than stiff lines that need to be flared to attach fittings. Flexible lines allow for moving of the calipers and cylinders without worrying that the lines will break, which is unavoidable on a car which is constantly being worked on. In addition, having pedals that can be positioned in a position suitable for each driver the flexible lines avoided the possibility of eventually snapping from constant repositioning. Between every banjo fitting or similar copper washers were used to prevent leaks.

All mounting bolts associated with the brake system were at least SAE grade 8, metric 10.9. The caliper and rotor mounting bolts were the original as supplied by the motorcycle manufacturer. All these bolts were were cross drilled at the head to allow for safety wire to pass through the bolt head. This was a painstaking process as the bolts are hardened, but a sensible requirement of SAE preventing the bolts from unscrewing. To drill the bolts I bought a number of cobalt steel drill bits which are more durable then HSS drill bits but cheaper then carbide bits, and I knew that some of the bits would break due to human error the carbide price tag was not worth it. Using a center drill to start the process and plenty of oil while drilling each bolt is an absolute must. During assembly red loctite was applied liberally to all fasteners. I only use red loctite as i find i can always loosen the bolts later even though red loctite is said to be almost permanent.

Coming soon…


The frame design and construction was the lengthiest part of building the car. Design of the front part of the frame was an iterative process over the course of 2.5 months with many 12 hour sessions, while the rear approximately 4 weeks as the engine box constituted most of the back and is not included in this discussion. Construction of the frame: jigging, tube fab, cleaning and welding required well over 150 man hours. Everything in front of the rear roll hoop was designed first, while the rear was designed after the drive train was decided upon. Designing the frame to fit a driver required an actual mock-up of the design to be absolutely sure someone can fit. Stiffness is the most important goal to aim for when designing a frame as this prevents the suspension from using the frame as an additional spring making handling feel sloppy. Given a blank sheet of paper the number of places one can place tubes to create a frame are endless, which configuration gives the stiffest frame that also fits a driver? To answer the question I performed an iterative process where after the placement of a few frame members performed finite element analysis and compared the new frames stiffness with previous versions, basically a manual simulated annealing and genetic algorithm. But some rules and knowledge have to be applied first to drive a sensible design. In this section Ill describe why a space frame design was chosen, the material used, the design and FEA process and the build process of jigging, clamping and welding. Watch the video to the right to see how the front half of the frame evolved from a sketch to the final CAD design. The FEA images included every so often were major design drivers, after a few key tubes were drawn much of the design was iteratively improved for torsional stiffness. The rear half of the frame was designed around the engine and motor mounting structure which is a stressed member.

-Chassis Design

Material Yeild Strength Density
Aluminum 43,000 osi .0027g/mm^3
Mild Steel 35,000 psi .0078g/mm^3
4130-T6 CrMo 63,000 psi .0077g/mm^3

Obviously its necessary to comply with SAE rules, and complying to them mostly limits the design to look like a formula car. There is still plenty of room for creativity and challenges when designing the frame. Material selection sets a design path: monocoque or space frame maybe ladder. A carbon fiber-aluminum honeycomb design is light but expensive so the material selection was reduced to aluminum or steel, the truth is Ive dealt with structures like this so the decision was made before it even came up, 4130 CrMo TIG welded. To the right you can see why its a sensible decision. Why not aluminum? Well, its a pain to weld, costs more, needs to be age hardened and the roll hoops still have to be steel.

Aside from the material of the tube its also important to know the fabrication method of the tube. Tube comes in two popular varieties: DOM (drawn over mandrel) AKA seamless/race car and welded AKA rolled. When buying tube from a yard you’ll find that even the person selling you the tube wont know if hes selling seamless or welded tube because some mistake seamless (DOM) tube for any tube without a seam that you can feel. DOM tube is made by piercing a cylinder and forming a tube without any seam as the tube is one uniform piece, and so it is seamless because the tube was not welded anywhere. On a DOM tube you cannot spot any discoloration line along the tube, it is perfectly uniform. Welded or rolled tube is formed by rolling sheet metal into a tube and welding the seam. If it is also drawn the weld will be very smooth but you will still be able to tell the discolored seam running along the tube, on a undrawn tube you will be also able to feel the seam. For a space frame pick DOM tube, tube that is made from a cylinder not welded because it is predictably stable under load and will match FEA results, whereas welded tube is weaker along the seam. You will find DOM round tube but welded square tube.

Making the frame as small as possible but still able to fit a driver requires some live mock-up. PVC pipe is cheap and easy to shape making it a good prototype material. Hot glue makes for quick joining of the tubes. After designing part of the frame in CAD an afternoon PVC mock-up of the frame allowed to identify which dimensions needed to be enlarged or shrank. Drawing a person in CAD and placing them in the frame is not as informative as actually being able to slam the brakes and realize that your knee will hit the top frame member.

Thanks to some teams with drivers smaller than a 3rd grader SAE imposed a minimum cockpit size which gives a decent cockpit size. However the pedals, steering column and tie rods, seat, controls and information displays still have to go someplace. So I designated a minimum boundary, as provided by SAE and our own measurements, for the driver and designed around it.

A tetrahedron resists forces in 3D space

A tetrahedron resists forces in 3D space

Here are the steps I took to reduce the problem:

  • Defined a construction style: Space frame
  • Chose a material: 4130 CrMo
  • Chose a wheelbase and width
  • Enclosed the frame in a boundary box
  • Drew the minimum cockpit size after a live mockup

Initially I looked for inspiration from the Ariel Atom and reverse engineered and scaled it’s frame. However the bowed main members of the frame were essentially pre-buckled and making the frame stiff and light proved a challenge. Another major drawback was finding a machine shop with a CNC roll bender to manufacture the huge diameter tubes and that would sponsor the team, most shops have only a rotary draw bender. Additionally jigging the tubes for welding with the budget and chemistry lab welding table we had was going to guarantee days of annoyance. Finally the design of the engine bay would need to be completed even before an engine/motor/drive train was picked and designed. Since the most important subsystem is the drive train I wanted it to shape the frame design not the reverse. However, using the Atom as a base design did drive most of the shape as the most exterior points of the frame intersected the main curved members which were left for construction.

As mentioned the frames goal is to resist torsion applied by the suspension A-arms. To visualize how the frame is twisted, picture what happens going into a right hand turn while sitting in the cockpit. The car accelerates outward and the left tire exerts a force opposing the leftward motion which is reacted in the lower left A-arm by compressing it, while the upper left A-arm is under tension. On the right side the reverse is true and the upper A-arm is under compression while the lower under tension. This creates a torque on the frame, thus the frame design needs to transfer these loads to reduce the torsional effect of the suspension arms. Forces can be transferred from node to node by arranging frame members to carry the load in compression or tension, never in bending. A triangle in a 2D structure is the strongest structural shape, for a 3D frame the strongest structure is a tetrahedron which resists loads applied on all planes by loading it members in either tension or compression. Extending this concept to the frame, all loads, especially those transferred by the a-arms, should act on the apex of a tetrahedron.

When drawing lines in CAD its almost natural that you drag the ends of adjacent lines to meet at a point, this is important as this point becomes a node which transfers loads to other tubes. If the lines do not meet at a point, parts of the lines will be in bending. When welding its easier to miter the tubes so that a joint does not include more then two tubes, but thanks to SolidWorks Trim/Extend feature mitering a few tubes to meet at a point is simple.

-3D Sketch and Weldments
Drawing using the 3D sketch option takes some getting used to. You start with lines and convert them to weldments using tube profiles you created, which you then Trim/Extend to make weldable joints. I used construction geometry liberally to ease the sketch process by drawing boundaries and key parts such as the wheels, cockpit minimum area, height and ground. What happens next, or how the points are joined so that everything holds together stiffly is part knowledge into frame construction techniques, part intuition into how forces are transferred, part FEA and part art. I obtained inspiration from existing formula car chassis, Apache helicopters, bridges, the Arial Atom and available literature.
SolidWorks comes with a few tube profiles so we need to create our own for the type of tube we are using. Select “Open..” and view all file types, navigate to “C:\Program Files\SolidWorks\data\weldment profiles\ansi inch\pipe\” and open one of the profiles. Modify it to your needs and “Save As..” a new profile. This profile will now be available in the “Structural Member” “Selection” menu.

Torsional stiffness of KiWi 1 frame iterations

Torsional stiffness of KiWi 1 frame iterations for the front half

The figure to the right shows some of the design iterations and the torsional stiffness of each. Notice how the designs using the Ariel Atom’s main curved tubes are significantly less stiff, pushing the final design to a space frame type. the rather high torsional stiffness is due to only the front area of the frame being tested here.

Finite element analysis, while easy to perform with COSMOS, is only informative if the simulation is setup to simulate the real world as closely as possible. The challenge thus lies in applying the correct forces of similar magnitude and direction in the correct locations on the part being tested. Failing to plan the simulation will provide useless results. I found good literature on the subject of performing correct FEA simulations, including for formula space frames. Anecdotal information on just how stiff a frame should be from driver tests set a goal to reach for of ~1600ft-lbs/deg.

Frame FEA
Torsional Stiffness goal ~1600ft-lbs/deg
Forces applied 1.5g braking
1.5g lateral acceleration
3.5g bump
COSMOS analysis Beam Mesh: Static, buckling
Frequency Analysis: Engine mounts
Weight < 75lbs

As part of a real world check I rapid prototyped two frame versions in ABS plastic, a scaled Ariel Atom design and the final frame design. It was interesting to see the models scaled down and to twist the frames to see if there is a difference in the torsional stiffness. Not surprisingly there was a big difference in how stiff the two frames were, with the Atom style being much easier to twist. While not a scientific method its enlightening to actually feel the frame and see how it reacts to different forces. I find its a quick, effective method to decide over two different frame designs allowing to show and tell results to others. Rapid prototyping a frame design requires that all structural members are changed to solids instead of tubes, forgetting this will make for a very weak prototype but most likely wont even be printed correctly as the wall thickness will be under the resolution of the printer.

SolidWorks weldment for one of the more complicated joints on the KiWi 1 frame

SolidWorks weldment for one of the more complicated joints on the KiWi 1 frame

-Save as life size..

Exciting as the CAD design phase might be its time for a reality check, lets try to see if the design is buildable and prepare it for fabrication. Determining if the design is buildable requires welding and jigging experience to identify nodes which might be difficult to jig and weld by hand. Acute angles at tube intersections cause difficulties in welding the tube even with a small cup size and extended tungsten electrode because its difficult to reach deep inside the angle with the electrode. SolidWorks 2009 weldment trim/extend feature miters intersecting tubes as in the figure to the right eliminating the need to do this by hand. Some intersections require a few headstands and compromises to get SolidWorks to miter the tubes properly.

The order in which tubes are mitered is important to avoid unweldable joints

The order in which tubes are mitered is important to avoid unweldable joints

With the weldment tool it’s still important to order the miters in a way to avoid a dining philosophers problem, where one tube has to be welded before another. The image to the right depicts the order in which the tubes should be mitered. In this configuration we can jig and weld tube 1-2, then tube 4-3 and finally tube 4-5 needs to only be clamped as it should fit snugly into position. If we switched the miter order to 5, 4, 2, 1, 3 we would have to jig all three tubes at the same time for welding as tube 5-4 would need support at end 4 so tube 4-3 needs to be placed in position which requires tube 1-2 in position as joint 2 needs to be under joint 3, creating another problem where joint 2 cannot be fully welded as its partially covered by joint 3. While seemingly obvious, its easy to overlook if the frame is not reviewed a few times before fabricating the tubes.

Save each tube by selecting Insert into New Part

Save each tube by selecting Insert into New Part

Once a tube is mitered at both ends you select the tube in the features tree and unroll the “Trim/Extend” features created. Right clicking on the “Trim/Extend” select “Insert into New Part..” which will save the mitered tube as a new part for further processing. I find it odd that you have to unroll the “Trim/Extend” features and insert them into a new part instead of the tube, as essentially each tube is repeated a few times in the sub-trees.

Unrolling of frame members

Unrolling of frame members

So far we’ve drawn the frame, analyzed it, mitered all the tubes, made sure the frame is buildable and saved each tube into its own part. Now we have to transform the mitered tubes into metal. One way of doing this is via unrolling of the frame members, creating a sketch, printing and gluing the sketch onto a tube and grinding away at a grinder to create the miter. To unroll the tubes we either slice the tube with a thin feature or create a tube profile that is already sliced very thinly (the later might cause problems when using the Trim/Extend feature). If creating a sliced tube profile, replace every weldment with the new profile and open each tube in a separate part. Unwrap the tube using Sheet Metal–>Bends and set the Bend Parameters: “Fixed Edge” = the inside edge created from the sliced extrusion, “Bend Radius” = OD of the tube, “K-Factor” = 1. Now create a sketch of the tube and print if you have a plotter that can print any length tube. For a normal printer you can right click on the drawing and select Drawing Views–>Break to show only the ends of the tube which you can glue to the ends of the actual tube making sure the sketch centerlines match. Its best when you have a grinding wheel that has the edges worn down to form a half circle at the end instead of the sharp corners of a new wheel, it will be easier to avoid cutting too much in an area. Take away plenty of material at first and then carefully reach the lines on the tube so that the entire end is smooth without any peaks or troughs as the former will misallign the tube and the latter will make welding difficult. A common mistake is to start slowly, become impatient and miss the line, go fast then slow down as, the material before the line has no impact on the tube fit its only the last few “hundredths”. Now repeat this about 120 times for each member of the frame.

Nest Step Engineering tube profiling services

Nest Step Engineering tube profiling services

Hand grinding is a very tedious process, prepare to waste lots of tubes before obtaining satisfactory results. A much easier way is to have the tubes mitered using a CNC tube profiler. I sent the tube files along with an order of tubes to Next Step Engineering and within a few days received a 3D frame puzzle ready to be jigged and welded. The plasma cutter used to cut the tubes does leave a small nipple at the end of each tube which is easily ground off. Construction accuracy and speed are magnitudes greater then when grinding by hand. The number of jigs needed is far less as the tubes just fit, its easy to take for granted just how much work a tube profiler saves and how good the results are. TIG welding the frame requires the tubes to fit well all around a joint without gaps and preparing the tubes in this way guarantees gap free joints which are a pleasure to weld. Unlike hand grinding by spending hours to get a tube just right on one end to realize that its off center on the other.

Jigs used during KiWi 1 frame welding

Jigs used during KiWi 1 frame welding

-Getting jiggy with it
Proper welding requires proper jigging, this is even more important when TIG welding which is much less forgiving with gaps than MIG welding. While I don’t recommend it, if MIG welding you can get away with a person holding a tube by hand while another tacks it into place. That is not possible while TIG welding because of the wide gaps that form between touching faces without proper clamping, aside from the fact that you can forget about positioning accuracy. The integrity of the frame and its ability to handle loads as simulated using FEA largely relies on the precision with which it is manufactured. Tubes should intersect at a point to transfer loads effectively, welding should not overly heat the surrounding material and positioning should be accurate so as not to interfere with suspension geometry. All these requirements can be met with proper jigging and clamping.

Once the frame was designed and the tubes sent to be mitered, I designed the jigs that would be needed to position the tubes in place for welding. For jigging material I used 6061 aluminum from a cheap source as it’s quick to machine. Looking at the frame, design jigs to fit between frame members so that they are positioned correctly for welding, remembering that there needs to be some way to clamp the tubes and jigs. The front bulkhead and bottom frame tubes are ones from which all dimensions are later referenced from, so these should be positioned as precisely as possible. To the right are some of the jigs used while welding the frame, some of which were reused in different locations. Notice the rectangular cutout in some of the jigs, they were placed over a 6?x6?x4? parallel which also acted as a jig in order to save aluminum.

Its best to design the jigs right in SolidWorks on the frame and then open them as separate parts. For speed and precision purposes I CNC vertical milled all the jigs. Using MasterCam X which comes with a plugin for SolidWorks, you export the part into MasterCam X and prepare it to be machined. If you have ever used MasterCam you will know its absolutely annoying and its sometimes quicker to redraw the part using the CNC machine’s display. I used both and later reserved MasterCam for parts that were too tedious to draw on the CNC machine as the work involved in setting the tool paths was less than drawing it on the CNC. The CNC I used was a Bridgeport with an Acu-Rite MILLPWR control system.

Tools for jigging
6?x6?x4? parallel
Set of 1/8? and 1/4? parallels
Various sized C and Bar clamps
6? and 24? caliper
Height Gauge
Digital protractor
36? ruler
Tape measure
6061 Aluminum for jigs

In the table to the right I listed the tools needed to properly setup a frame for welding. C Clamps and bar clamps get used a lot, at one part of the frame at least 15 clamps were used to hold together the jigs and tubes ready to be welded. Less clamps could be used if you have enough funds for aluminum to make large jigs. The set of 1/8? and 1/4? parallels were used for precisely spacing jigs and tubes, Ill also admit they were used in ways that parallels should not as they were clamped around tubes etc. A height gauge simplifies measuring in the vertical and placing it on large parallels gives enough height to measure any part of the frame. Obviously there are angles in the frame and a digital protractor makes it easier to measure those, its only hard to find a decent one for a reasonable price. Most suffer from being too big and were designed for woodwork not 1? tubes in a tight space frame configuration. Id love to hear about one that looks like a small foldable 90deg square with a flat digital readout on the body instead of the large housings that hold the electronics of most readers and get in the way of measuring, and the ability to read angles on both sides of its joint.

Proper welding requires propper clamping

With all the tools and jigs ready the frame tubes were setup and welded in the order they were mitered in SolidWorks, so the frame was virtually assembled (looked over) in SolidWorks before any tube was cut. The tubes should fit the jigs not the other way around, don’t change the jigs if a tube doesn’t fit. Double check measurements and modify the tube as there might be some extra material, especially if the tube was to have a feathered edge.

The base of the frame, the bottom and front bulkhead which were made from square tubing became points of reference for all other measurements on the frame. Great care was taken to precisely position the tubes, they were tacked and welded very slowly to minimize any heat distortion. Having these as reference points made it easier to precisely place all other frame members.

Jigs ensure tubes are correctly positioned and aligned, but they do not ensure that the tubes will stay so if not properly tacked and welded. Welding the tubes expands and contracts them, causing stress which moves the tubes, thus applying this heat in a disproportionate amount to one side will cause distortion to that side. The next section describes a proper welding setup and procedure to avoid such distortions by tack welding first and then running beads in a manner to cancel the distortion.


4130 CrMo TIG Welding Setup, .065?and .095? tube
Product Type McMaster part number
Tungsten Electrode 2% Thoriated Tungsten
1/16? and 3/32? diameter
8000A96, 8000A97
Inert Gas 100% Argon
Tungsten sharpening Silicon Carbide wheel 4422A35
Filler Wire ER70S-2 1/16? diameter
Cleaning alcohol, sand paper, wire brush
Welder Setup Water cooled,

80% of the work while welding is not welding at all, it’s the preparation to weld: jigging, clamping, measuring, cleaning. Above I described how to jig and setup the tubes properly. I should mention that before doing so the tubes need to be cleaned to remove the oxide coating or mill scale and any filings, burrs, oil, dirt. I recommend a good deburring tool to clean the burs inside of tubes. An angle grinder equipped with a Scotch-Brite™ Clean & Strip Disc makes cleaning the oxide coating super fast, but be careful to not apply too much pressure as it will grind away at the tube also. After the mechanical cleaning processes use alcohol or acetone to clean any oils and impurities, you should clean all around the area to be welded inside and outside of the tube and on the outside just prior to welding. During welding keep a small wire brush available, obtainable at any welding supply store, to clean a partially welded joint before completing the weld, such as after tacking.

TIG welding takes at least 40 hours of practice before you can start to weld tubes, spend 2 weeks constantly practicing for 4 hours a day before welding joints that can have an impact on someones safety. 40 hours is a minimum, I have trained others (I’ve been a machine shop TA for 4 years) in MIG and TIG, while some catch on quicker or are just more dexterous having less then 40 hours of experience shows in burnt through tubes, huge heat radiation, constant tungsten resharpening and an inability to pulse weld. However, those 40 hours include only the skill required to move the torch, apply the filler and depress the pedal. It does not include the other 80% of welding which requires jigging, clamping, cleaning and picking of appropriate filler and tungsten electrode for the material being welded. The table to the right lists the materials needed to weld 4130 CrMo tube of .065? and .095? wall thickness. Notice the filler rod is not 4130, as using such a rod is recommended when heat treating the structure later, otherwise the joints are less ductile and may crack.

Welding table – A flat table is essential to a true frame. Use a large level or straight edge to determine if the table is flat and not bowed in the middle as this unevenness will transfer to the frame being welded. While we didn’t have a welding table or the funds to purchase or make one, we had a chemical resistant solid black surface from a chemistry table, it was flat and big enough to handle the front part of the frame. Spotless, is the adjective that should describe the table at all times.

Tungsten sharpening – Tungsten electrodes should not be sharpened with an Aluminum Oxide grinding wheel as it implants impurities into the electrode, use a silicon carbide wheel instead. Hold the electrode perpendicular to the axis of the grinding wheel so that when grinding you grind along the tungsten electrode not across. Keep your hand above the wheel not below because if the electrode catches on the grinding wheel it will go straight into your hand. Turn the electrode so that it is sharpened to a long sharp point. Resharpen whenever the electrode touches the material being welded as it becomes contaminated and will weld poorly.

Protective equipment – Thin gloves made from a soft leather allow for finer control of the torch, which is a must for tight spots. Waking up in January with a sunburn can be surprising and common when forgetting to wear long sleeves while welding. The welding helmet causes the most controversy: auto-darkening or standard? All noobs will want an auto-darkening helmet even though it provides poor visibility while actually welding, your eyes are flashed before the helmet can react to the spark while welding thus causing headaches and poor vision, especially the ones under $200. Then again most noobs wont actually spend a lot of time welding but sharpening electrodes, so you may use an auto-darkening helmet until realizing that it is much worse then a $30 standard helmet. Using a standard helmet with great visibility and no flash requires practice as you have to flick your neck to slide the helmet over your face while keeping your hands in position, then depress the pedal to start the spark and quickly reposition your hands if they have moved. When deciding on standard helmet choose one with a large protective glass instead of ones with just a slit for the eyes, you will have much better visibility.

KiWi 1 first half of the frame

KiWi 1 first half of the frame

Tack welding – Welding introduces much stress into the frame from the constant expansion and contraction of the welded material, follow these guidelines to minimize the stress buildup and to not distort the frame. After jigging the first step is to tack weld everything into place. Use a diagonal pattern, so if welding a square, tack weld the upper right corner, then the lower left, then the upper left and finally the lower right, repeat this to get about 3 tacks per edge before starting to run beads. This ensures the tubes will not move from the stress caused by the heating and cooling of the material. Start welding a bead at a joint and use the same welding order as during tacking. Don’t weld the entire joint in one shot. Most likely it will be impossible to weld the tube all the way around without stopping to reposition, at that point let the joint cool while welding another joint diagonally away. Do not begin welding a joint before it has cooled as this will cause cracking in the recently welded joint.

Common welding problems – Most welding problems occur due to impurities. Clean the tubes with a wire brush and alcohol, resharpen the electrode and try again. Make sure that the gas is flowing and the amperage is just right, about 1amp for every .001? of an inch, so weld .065? tube with 65amps, with experience you can weld with more amperage and faster. Check if you are using the proper electrode for the material, if it hasn’t been sharpened on an Aluminum Oxide wheel and if the filler rod is suitable for the material welded. Never use any wire brush that has been used on aluminum, it will embed impurities into the metal being welded and cause difficulties, always use a separate brush for aluminum.

Engine/Motor Box

laser scanner, hair spray

Drive Train


Bump steer
King pin inclination (KPI)
Roll center

CNC mill

Interference/clearance fits
Liquid nitrogen

KiWi 1 HV Diagram

KiWi 1 HV Diagram

HV Electrical System

KiWi 1 Hybrid Control and LV architecture

KiWi 1 Hybrid Control and LV architecture

LV Electrical System

Drive by Wire

No HV conduit needed to the drive pedal as opto couplers were used at the controll box to seperate the HV and LV systems.

Yale 2008 – Torquere et Velocitas