Honda / Acura K20a K24a Engine Forum banner

1 - 4 of 4 Posts

24 Posts
Discussion Starter #1 (Edited)
Hi guys,

It's time to make a build thread for this car. I'll start out by giving you guys a run-down of the engine plans and build philosophy and the progress thus far. I'll give you guys some pictures for now, but may hold off taking pictures of much of the work until I can get the car out of the garage, since I think I'll get better pictures and it will be much more convenient. I should be able to help if you guys ask to see something in particular.

The idea for the car is to use it on the awesome roads we have just northwest of Madison, in the driftless region of Wisconsin, and at blackhawk farms raceway in South Beloit, IL. These roads can be somewhat bumpy, and have corning speeds that range from 40-~100mph mph. I think the EP3 it is one of the least expensive platforms to use for the level of performance it can achieve. The EP3 has a very mixed reputation for its tuneability compared to the older models of hondas, however my research makes me think most of the disadvantages associated with it (and the DC5) are from people lowering and/or stiffening the suspension too much. The car has several price and performance advantages versus older hondas that I think give the EP3 the edge on the price to performance ratio, plus, I inherited the car from my parents so the car was basically free. I want to show that anyone can built certain cars, like the EP3, that can achieve very good track performance at a comparatively excellent price using a carefully selected combination of parts, internet research, and time.

Decisive advantages versus older hondas:

Greater chassis rigidity: 95% better torsional, and 22% better bending rigidity than em1.
- Of course, roll cages can be used to increase rigidity, but doing so can be very expensive i.e. $1000 for a weld in roll cage, plus the work to remove everything, repaint the welded areas, modify the dash, etc. Unless it were really nice, adding a roll cage to an older car may make it only about as stiff as the EP3, and increase to the weight to not a ton less as well. For this reason, I want to keep performance at a limit that corresponds to the chassis rigidity without a cage. I'll reduce chassis loads by by keeping weight low, suspension plush, and tires not too gnarly (~140 treadwear).

Low Cost:
The EP3 does not require many of the modification that the older hondas or DC5's do for certain upgrades. For example, the 02-03 EP3 uses the same proportioning valve as the JDM DC5 for the ABS, so DC5 brembos are a direct bolt on (ep3 requires more force at the pedal due to smaller booster). The 02-03 EP3 OEM front sway bar is the same 25.4mm diameter as the JDM EP3 and DC5 type R's, so only a 21mm (05-06 RSX Type-S) or 22mm (JDM DC5 or EP3) rear sway bar is needed to replicate the sway bar setup. The overhang weight and length is the smallest of any FWD Hondas. The Ep3 comes with EPS (electronic power steering) that doesn't seep power from the accessory drive belt, and has a tight ratio compared to other OEM racks. Seat bases, engine mounts, brake lines, exhausts, suspension bushings, chassis braces, etc. designed for DC5 can be used for EP3 (with few exceptions), making parts easier to find and therefore more likely to be cheaper.

Suspension Design:
The unique feature of the EP3 and DC5 that the older civics and integras lack is that deflection is used to improve geometry. The front compliance bushing is located on the front of the arm, resulting in toe-in under load. That toe in, while cornering, reduces the ackerman, causing the front contact patches to become more parallel, the caster angle of the outside front wheel to increase, and the camber to increase as well, potentially resulting in more available forward grip. The rear compliance bushing is also located on the front of the arm, and also results in toe-in under cornering load, which is helpful for rear traction and stability and is therefore used very successfully on the honda prelude (the 1st gen. prelude had the fastest 60mph slalom time in the world at the time of it's release - faster than the Ferrari F40), and many of the worlds fastest cars i.e. Porsche GT2 RS, Lamborghini Aventador SVJ, etc. The suspension must therefore remain in a very specific range of it's suspension stroke to retain the proper deflection characteristics under load, which explains why the Type-S, Type-S A-Spec, and Type- R models all have progressive spring rates, and why this car behaves so poorly when lowered.

OK, moving on to the build:

Current mods:

K20a3 Bc Stage 3 cams
Insulated Airbox w/ Cold Air Snorkel
PLM Header with Custom 3" Flange
3" Custom Exhaust
Custom Oil Catch Can
Innovative Engine Mounts (85a side, 75a front/rear)
Sunroof Delete
Sound Deadening Removed
A/C Removed
Rear Seats, Carpets, Interior Panels (Except door cars and Dash) Removed
Lightweight Battery + Relocate
R. Wiper Delete
OMP Front Tower Bar
OEM F. Sway Bar (25.4mm)
22mm JDM DC5 R. Sway Bar (22mm)
RSX A-spec suspension (new top hats, bearings)
Hardrace front LCA bushing
Energy suspension rear LCA bushing(80a)
Hardrace Front RLCA bushing
Powerflex Rear Shock Eye Bushing (95a)
Superpro Knuckle Bushings (80a)
Prothane Rear URLCA Bushing (85a)
Hardrace Pillow Ball Rear Inner Bushing
Whiteline R. Sway Bushings (80a)
05-06 RSX Type-R R. Tower Gussetts
Custom R. Tower bar
OMP Steering Wheel
Cobra Suzuka Pro-Fit Seats (Fixed Passenger, Sliding Driver)
5 Point Harnesses (OEM Belts + Airbag Delete)
F/R 06 Type-S 5-lug Swap
JDM DC5 Brembos + EBC Yellowstuff pads
4x AP2 Front Wheels (17x7 +55) w/ 22mm spacers (+2cm Track Width f/r)
Michelin Pilot Super Sport 215/45/R17
S2000 299.6mm F. Rotors
Carbon Hood

Right now it only makes 180whp and 145tq, I think because I used a NACA duct intake that had and ID <3" (less than advertised, arrived right before dyno'ing). I think a simple velocity stack on the end of the snorkel would make 5-10hp more, but I'll wait until the new engine to revise it. Its actually really fun and pretty fast on winding roads already.

The durometer of the bushings is optimized for proper deflection... I chose a hardened rubber bushing for the front LCA bushing so that it would maintain it's original deflection characteristics while in use with 140 treadwear tires (235/45 Advan AD08r's are here on standby with new wheels). I chose a urethane bushing from the rear bushing so the suspension would move more freely. A urethane bushing was used at every point where the articulation of the pivot is coaxial to the bushing's inner sleeve. All of the urethane bushings in the system were bored to match the inner sleeve diameter more closely, lubricated with Shin-Etsu Silicon grease and fitted with zerk ports to keep them moving as freely as possible. I think this setup could move even more freely than the pillow-balls that are used instead of bushings on most racecars, and with zero play. I also experimented with various tie-rod ends and end links to find which move most freely. OEM tie-rod ends are BY FAR the smoothest and most free. OEM end links are better than knock-offs, but I may end up going with a pillow-ball setup unless the OEM end links free up with use. Moog was the least free, and became pitted very quickly in my experience.

After watching some youtube videos of EP3 and DC5 rear suspension deflection...

I concluded that the knuckle bushings are used primarily for shock loads (i.e. bumps, potholes, etc), therefore I chose a urethane bushing that was the same or softer than the remainder of the bushings and about the same, or a tiny bit harder, than the compliance bushing. Urethane has a different rate of elasticity so a urethane bushing that is the same hardness as hardrace rubber when static, but stiffer under compression. The front RLCA bushing is by far the biggest bushing in the system and is the biggest factor in rear deflection, therefore I chose to use a hardened rubber bushing to keep the deflection characteristics the same as OEM when using stickier tires, just like in the front. When the rear deflects, the upper control arm bushings and rear inner bushings must also deflect in the process - the flat metal plate that connects the knuckle to the rear inner bushing is actually as thin as possible in the middle [and thicker where it attaches to the knuckle and inner rear bushing] so that it can bend to aid in front RLCA bushing deflection. Therefore I chose 80a urethane upper control arm bushings, and a pillow ball for the rear inner bushing. The shock eye bushing is 95a (almost solid). I had to use different manufacturers for each location depending on their durometer and availability (prothane upper inner, superpro upper outer, hardrace lower inner, etc.). In the videos you can see that quite a bit of defection occurs are the rear inner bushing, generating toe out. Since the OEM compliance bushing has relief cuts, it deflects more than the rear inner bushing, but as the system wears, I think more and more deflection occurs at the inner rear bushing, and less and less at the front due to the rear inner bushing wearing at a faster rate, even more so if your cornering loads the rear wheel frequently, just because there's so much less rubber than in the compliance bushing. Furthermore, it's likely the guys in these videos had urethane bushing kits, which include the compliance bushing. All available rear urethane compliance bushings are >90a, therefore there is a 100% chance your suspension deflection will result in toe-out instead of toe-in! Don't use hard compliance bushings (or use harder bushings everywhere else)! Same goes for in the front - even a front compliance bushing with the same durometer as the rear front arm bushing will behave like it's stiffer, since while cornering the bushing is already compressed from the arm articulating, unlike the rear which rotates coaxially. It's also possible that their aftermarket LCA came with a soft bushing - I called several companies after watching these videos and learned that most or all company's arms come with bushings 70a-75a whereas most or all urethane bushing kits use 80a - Prothane, and Energy Suspension use all 80a (with harder compliance bushings), superpro is 75a-80a, Powerflex uses 95a ("street" and "race" powerflex bushings are THE SAME for the EP3/DC5)! The guy in the first vid has skunk2 LCAs, which have 70a bushings, and you can see that's where most of the deflection is occurring.

The factory Type-R setup is close to perfect - usually times go up when the suspension is altered in JDM specimens. Likewise, the NSX-R was close to perfect - the fastest light-tuned specimens use the stock setup, or merely lower the car by an inch or less, even when using slicks. The design of OEM suspensions for this chassis put the strut bearings inside the spring whereas all available coilovers use a smaller diameter spring that requires the bearing to be positioned on top of the spring, which reduces suspension stroke. Furthermore the 05-06 versions of the car uses inversely wound springs, so even using an aftermarket damper that uses stock springs would require custom spring seats. All available suspension kits for this chassis are plagued by these issues. Therefore I think the only way to improve on Type-R suspension in a way that retains it's versatility on bumpy surfaces is a very custom setup with progressive springs that keep the full suspension stroke, or to extend the strut tower somehow. My research indicates the 05-06 A-Spec suspension for the RSX is the same as the 05-06 Type-R, with spring rates adjusted for LHD. Therefore I am, for now, using A-Spec suspension. At $600 new, it is another major cost advantage of using this chassis.

I'll be positioning the seats closer to the middle and farther back compared to stock. I'll use a steering extension to let me sit as far back as possible without needing pedal modification. I'll determine the driver's seat height based on visibility and feel, and will mount it on rails, because my brother will be using the car as well. I want the rotation of the car's body to correspond to the center of rotation of the driver's body to better transmit feedback. This is a similar approach to what can be observed on the BTCC Ep3 (car in my sig) and in most racecars at high levels of competition (GT, JGTC, etc. - lower isn't always better). The passenger seat will be mounted directly on the base and as far to the middle and rearward as possible.

I've cut the front bumper in front of the tires and installed a custom splitter that I made out of plywood and painted using sailboat hull paint, which extends to the back of the subframe, similar to what we see on the BTCC car, and Rikli Motorsport EP3, which are the fastest EP3's that I know of by a somewhat significant margin.
Rikli Motorsport EP3
I'll post pictures of how I made and attached the cowls and installed the splitter ASAP. This will work in combination with a 60" APR single-foil rear wing, mounted to the hatch using 2 plates of 7075-T6 7/32" thick, similar to the BTCC car. For now I'll be using 04-05 EP3 side skirts, but plan to use a flat underbody with integrated f. quarter panels in the future to make airflow more consistent and to vent air that comes through the f. bumper cowls from behind the front wheels. I'll post pics of the wing, skirts, and installation process/design as soon as possible. The rear bumper is cut about .5" above the level of the bumper bear for aero as well, similar to the Rikli Motorsport car.
Rikli Motorsport Cut Bumper

OK, Now Onto The Engine:

I'm using a k24a1 complete engine as a base, swapping the pistons, rods, oil pump, cams, etc. I chose the a1 because it's cheap, and because the head flows better or as good as any of the other k series heads. My goal is for torque to peak over 205 lb/ft, and to have at least 260 whp, with an emphasis on lightweight components and minimal parasitic loss. I think a powerband like that will make the car even faster for my use than a 290 or 300whp car with 190ft/lb. The BTCC car only made 260whp, but a whopping 200 lb/ft toque of of the 2.0l K series!

I'll be doing a VTEC-killer setup with '08 TSX cams and RBC manifold. Half the reason for choosing TX cams is because they're cheap. The second half is because I remember seeing a dyno of someone with similar pistons, etc. making 210 toque and around 260 whp with those cams - much more than most, or everyone making greater than 300 whp. VTEC killer allows me to keep the a1 head, reducing the cost, and recides reciprocating weight by ~900g by using single-lobe exhaust rockers on all 8 vtec lobes. Plus, I save an additional ~1lb by deleting the VTEC solenoid!


I'll be using supertech exhaust valves, and +.5mm (35.5mm) Crower flat-face intake valves, with 3-angle job of course. I chose to do new exhaust valve just to refresh my valve stem clearances, and because the original valves were really dirty. I chose 35.5mm intake valves mostly for experimentation, and because I remember seeing several dyno graphs showing that it can make power. The Crower flat-faced type were the only option that I could find in stock in that size.

As you can see, I smoothed-out the surface of the combustion chamber, a bit before the valve work and decking, using a wire wheel bit on a dremel.

Here's my port work. I used machining dye to port match it to the hondata gasket mounted as high as possible on the mounting studs, in order to remove more material from the tops of the ports rather than the bottom in order to make the port shape a bit more vertical and transition smoothly to the 35.5mm valves. I tried to keep it close to congruent with OEM. Hopefully it works well lol. I used a rotary tool with a carbide bit, followed by a sanding bit on the intake, and only the sanding bit on the exhaust side, just to smooth it out.



Since the cams are only 12.35mm lift, and I have ap2 seats, retainers, and keepers laying around, I chose to use Ballade Sports springs for the S2000. They are designed for use with the AP2 stuff that I already had, and are actually quite stiff, almost as still as Supertech's HD1021 springs, which are by far the most common for cams with 14mm lift or more. I think these should be safe to at least 13mm, which is probably the most I'll need to get over 200 torque.

Here are the Ballade springs next to AP1 intake springs (ap1 has 12.65mm lift so I may actually just use AP1 springs until I need to switch cams, to make a couple hp and reduce valvetrain wear).

I drilled holes for oil squirters into the block using a drill press. I checked the flatness of the squirter-to-block interface using machining dye. I could fit a .008" feeler gauge under Number 3 so will require refacing the mating surface by 0.7 degrees. I chose to use oil squirters to increase the engine life, since track sessions may be 20 minutes or more at WOT, and to reduce temperatures so that I can run a tighter p-w clearance (also for increasing engine lifespan).


I used the crank and block codes to determine a yellow-green-green-green-yellow bearing combo as my starting point for measuring clearances, which I then checked multiple times using a mitutoyo .0001" bore gauge + micrometer, as well as plastigage. Interestingly, my plastigage measurements were about .0001" below my bore-gauge measurements. I chose to represent my data my as the following: .0019"(1), .0018"(2), .0020"(3), .0018"(4), .0019"(5). Using this data, I chose to go with the following final bearing order: TOP - blue, black, brown, black, blue BOTTOM - black, brown, green, black, brown. My new clearances are perfect at .0016", .0016", .0019", 0016", 0016". Using this bearing order also keeps the clearances symmetrical top and bottom. I see that many people run slightly larger than this (i.e. 0.002"), I chose the highest end of oem spec to allow very safe use of 30 weight oil. My research shows that it's possible to gain 2-10hp using certain brand oils over others, so this will give me some extra protection in case that oil turns out to be thin. I'm thinking about using Eneos 10w-30 oil. Scott at King Motorsports, who's worked with me before on the tune and exhaust, also told me they use "30 weight Red Line oil" on all their endurance engines.

I used BC lightweight sportsman rods, which weight about 458g each (36g over claimed). I chose these for the price, and because I think most rods are overbuilt for a setup like this. I'm starting to think these may be the weak link in the engine, just because everything else should be so durable - the oiling design is somewhat different than OEM and looks somewhat inferior. (LMK and I'll snap pics)


I sent the Supertech 88mm 12.5:1 pistons to Calico Coatings to have a skirt coating applied. I did this to reduce piston to wall clearance. Years ago, back when I bored the block, I had read that Supertech spec'd the pistons at .0040", then, after slapping issues, revised it to .00315", which is what the spec sheet said, and is what I used. Since then, I've noticed that the largest spec for this engine, by any manufacturer, is .0025", and that is for 2618 alloy (these are 4032) which expands about 14% more. JE and Mahle, who are the only other companies who make 4032 pistons for this engine spec .0020" and .0016", respectively. I called Supertech and asked them about this, I was told that down to about .0020", or even slightly less, is actually fine for NA! He did say less than that may be risky if the engine were to overheat or experience detonation (of course). Calico told me the thinnest they can do the coating is .0008", so we are shooting for .0008"-.0012", therefore the piston-to-wall clearance will be .0019"-.0023".

The Supertech pistons came with a wrist pin that has a .2" wall thickness. This is the same pin that's provided with all of their pistons for this engine, which makes me think they are overkill for a NA setup like this. I see that people upgrade to thicker-walled pins somewhat often, even on NA builds. The pin, especially with .2" walls, should be way stronger than the pistons or the rods, therefore I had custom pins made with a .15" wall. Precision Performance Products in NC modified one of their off-the-shelf pins to my specified length and chamfer; they cost only about $19 each and PPP was an absolute pleasure to work with! The new pins are a stronger material and weigh ~19g less per pin. Obviously this is better because it reduces reciprocating weight, but it also reduces the load on the connecting rods, sleeves, and piston skirts, therefore helping to preserve the piston skirt and bore, which are more prone to wear than an equivalent 2.0L K would be.



Calico Coatings will also be doing a DLC coating to the rings. Instead of using a traditional ring filer, I made a jig using random bits I had laying around, including R/C car bearings. The ring gap is positioned over a groove, in which I place a fine diamond file that is used to remove material for the gap. I place this setup under a binocular microscope that I happen to have laying around. This way I can be sure that the edges are perfectly square and parallel.


Stay Tuned

24 Posts
Discussion Starter #2 (Edited)
Weaver Auto Parts here in Madison has been awesome and inexpensive, taking care of business for all my machining needs. Their bigger machine shop in Sauk City, WI (25mi away) is equally badass. My friend Paul here in Madison does head work, decking, cleaning, etc., and Sauk City does the block and balancing work. The Sauk City shop offers a 5g "street" balance and a 2g "race" balance. They were totally game for doing a 0.1g" crankshaft balance!... with the flywheel, pressure plate, and harmonic balancer attached!! Yes, the entire rotating assembly is balanced to within 0.1g!!! They said the crank was almost perfect, so here you can see it only took a bit of grinding, no drilling. However, the Clutchnet pressure plate required 10g of material to be removed! Remember, Spoon's engines are balanced to within 0.2g and make >10hp over stock, with no other changes other than very precise clearances.




They needed to deepen the existing balance drilling of the flywheel just a bit. You guys will probably end up seeing the PRC pulley later on. They removed about 1g. I used the PRC pulley from a 06 Type-S because it is more than 1lb lighter than the original k24 CR-V pulley, plus is 5.4" in diameter instead of 6.125". This is an easier ratio for the accessory drive, which is only running the water pump and alternator anyway right now, gaining at least a few hp.

The flywheel is 8 1/4 lb.

Unsurprisingly, I'll be using a k20a2 oil pump. Here you can see where I ported it. Rumor has it that porting like this increases the threshold for oil cavitation, so the engine can rev a bit higher safely. Basically it just means the oil flows better and more easily. It was actually really, really easy and quick to do using a Dremel with a sphere-shaped grinding-stone bit. With the non-VTEC rocker conversion, OEM valve springs, slipperier oil, balanced assembly, easier accessory drive, and ported pump, I think my marginal gains should be in the ballpark of 8-18hp.

Here's a pic of the car right now. It's been sitting like this in the garage for several months.

Here you can see a quick snapshot of the bumper cowl. I still need to attach the airdam (between the splitter and the bumper), and make a gasket to take up the small gap between the cowl and splitter.

I drilled out the reinforcement beam and structure inside the dash, saving about 10lb! I'll probably removing the blower assembly from inside the dash, which weighs about 13lb. I'll then 3d print some adaptor cowls to attach computer fans to the defroster ducts in case the windshield ever fogs.

Stay Tuned

24 Posts
Discussion Starter #4 (Edited)
Several days ago I removed the Sparco Sprint Junior from the car that had been in it several years. It was a base-mounted design, manufactured in 2003. I wanted the seat positioned higher and tilted back; mounting the seat directly to the rail moved the bump in dash over the gauge cluster directly into the line of sight, and the driving felt better with it positioned higher as well. Movement is transferred to the body in a more rotational direction the closer to the floor the seat is positioned, because the rotational axis of the body roll of the car is located near the floor and middle of the car. The seat transfers more lateral movement when it is positioned higher and more total movement may occur, therefore offering at least as much driving feedback.

The old seat used a series of spacers to change the seat's angle. Eventually I realized that in order to achieve the optimal height and angle, I would need to make large wedge-shaped spacers that would put mover leverage onto the sliders since they were so parallel, possibly putting an unsafe twisting force into the seat-to-wedge interface. Furthermore, tilting the seat farther back more would cause the seat to interfere with my legs when operating the pedals. I thought a different base would be better as well because it didn't fit very well, making it much more difficult to install and remove, moving the seat into position being careful to avoid scratching the interface, and inserting the bolts solo.

I wanted to improve it, my friend wanted to buy my seat, and passengers frequently mentioned that the stock seat was insufficiently supportive. Side mounts can lean back much more than base mounts because of the bolt hole locations on the seat and the mounts, and can be positioned much higher because tall mounts are available that aren't parallel and therefore much stiffer and stronger and easy to raise more if need be.

I sold the Sparco to my friend after finding Cobra Suzuka Pro-Fits that appeared on Craigslist after several months of scanning. They are in excellent condition. These seats have taller thigh bolsters that give more feedback and support, although less under-leg support, allowing me to more easily straighten my legs down to the pedals. My dimensions are 6'3" 170lb 31" and I think 33" waist is a good maximum , although up to 36" may be able to ride uncomfortably. They weight about 18lb and cost $750 for the pair. They are also very breathable and I think will be cool, especially when it's hot.

Sparco and Planted mounts were the only convinient options, I found OMP on a european website at a similar price, even incuding international shipping. The final 2cm or so of every edge of the seat base surface curve downwards about 1cm below the plane of the seat base surface, which is a unique feature that must make it stiffer. The fit is excellent, and the seat base surface is very close to parallel to the ground when installed. Cost was $150 each shipped.

Ebay yielded a used tall OMP HC/733E and a discontinued Sparco tall mount. The side seatbelt holes in the seats are located very near the rear bolt hole, so interference was a factor when choosing. Most tall mounts have less material between the attachment points; since I thought the seat would likely be using some of the highest holes, I thought the extra material would make it much stiffer.

I test fitted the bases a couple days ago and the angle was way too far forward using both top holes. Lowering in the rear caused the mount to interfere with the plastic around the belt holes. With the bases installed, I sat in the seat and tilted it back to where I wanted it, then measured the distance between the ground and the front of the base while I was still sitting. I got out of the seat, and used the measurement to determine that it would be possible to achieve my position by drilling new holes and removing material to avoid belt hole interference.

Here are the mounts, hardware, and stencils used to guide the cutting edge. I wanted them to pass as original certified designs; they're stronger, after all. The raw edges were painted with a Duplicolor black touch-up marker to prevent rust.



Here we see the back angle and height difference.




I took the block to Paul today to have the oil squirter faces leveled. He said he'll be using a pilot into the hole and and a cutting surface that rotates coaxially and perpendicular to the pilot to remove just enough material to make them new oil squirter mating surfaces perfectly perpendicular to the holes for zero leakage.

Also, today someone in Europe accepted my ebay offer for an RSP intake manifold. I'll be port-matching, smoothing and enlarging the runners very slightly to match the growth factor of the 35.5 intake valves, plus will smooth the transition at the throttle body. 35.5mm valves increase the area of the valve faces by 55.376 mm per cylinder over stock; thats a 2.8% difference in area.

I talked to a new person at Calico coatings today; I hadn't received a response to my email that I sent the day I sent the pistons, received the invoice from Calico, and relayed the information I received from Supertech that day determining the final thickness spec. We are aiming for .0010"-.0012" of CT-3 / .00215-.00195" piston to wall clearance, limited by the machine's spray thickness parameters. They said their equipment only allows DLC coat to the outer face of the rings, and that that capability depends on whether they have the correct tool for this ring size.

Stay Tuned
1 - 4 of 4 Posts