rambler roadster

13 aug 2015
driving experience the unibody driveline
original ideas a pillar rear suspension
parts donor rear deck lid big nut axle
original car cockpit rough-out engine
completion (sic) pedals assembly front suspension
  chassis cleaning brakes
  chassis front cockpit
  chassis rear electrics
  front fenders  
  door indent  


alas, my poor 1963 Rambler American hardtop. though a joy to drive, was fairly rough when i got it. i think it's safe to say that it will be the last unreconstituted "old car" i kept on the road. i got it as a hulk, got it running, rebuilt the drive train, fixed up the rest, and drove it. it ran great, but rattled, squeaked, leaked, and generally the chassis drove me crazy. in 2012 i drove it on the Bonneville Salt Flats -- big mistake, don't do that with any car you care about, the salt eats everything and rapidly and you cannot rinse it all out -- then some jerk shot the windows out when it was parked on the street in december 2013. by that time this roadster was under way, and i'd already decided the '63 would be the driveline donor, so i took it off the road at that time.

it now sits under the ramada on jack stands, looking very sad.

the original

the 1961 rambler american i got from the other joe. it drove here, barely, but needed a new everything. once parked, it was clearly not going to drive again without massive intervention. at 53 years old every single thing was worn past usefulness. i oiled down the cylinders and decommissioned it.


currently my definition of "done" is simply registered, insured, on the road, and driven 100 miles without failing or parts falling off. this car is a situation, a process, not precisely a thing.

but these are at least representative of a functioning car and represent intent and some idea of completion, even if i change them later.


all production cars are ephemeral products, function and construction devoted to profitable sales. the inherent and obvious transportation angle long since resolved (from a marketing point of view). cars are designed to last long enough (just) to not harm the manufacturers' reputation. all of this is visible to those of us who take cars apart. my ramblers were reasonably well made (at least through the late 1960's), but this nash-designed car turned out to be exceptionally well made. shockingly so. as wonderful as this was to stumble on, it is part of why the Nash Motor Company wasn't doing well -- expensive to make, old (pre-war) technology, poor sales. none of that is of concern to me, now.

after the revelation of the Golden Rule the design came together quickly. the major design hack was to turn a four passenger sedan into a two seat roadster; that meant repurposing the rear seat space in a way that made sense, design-wise. know- or like-it-or-not, most westerners are ersatz car-design experts; car fan or not, you can easily see ugly (imagine: Pontiac Aztec). this sets a high standard.

while i was cogitating on how to pull off this task, i decided to chop out and replace the firewall, as part of the "remove ugly" phase (and to practice welding).

cowl, firewall

a pillar

the a-pillar is the vertical post that holds up the roof in front. it was a major detail in the roadster conversion. it was one of the few areas of the unibody that Nash made [internally] ugly, a side effect of continuous styling revision; "dressing up the old girl" a later AMC designer would call this process. since i decided to forgo the usual roll bar i added a vertical/diagonal brace to the a-pillar, in a way that mimiced the original vent window, visually integrating the brace into the design.


the rear deck lid was the single most significant restyling element in the car. again, the Golden Rule discovery dominated here; it singularly solved the design problem. the lid is the hood of the car, shortened. a subframe was built to at once stiffen the chassis and provide a mount for the lid, typing the b-pillars together, creating a stiff and light box around the passenger compartment. (side-impact safety was not a factor in cars of this era.)

it is here that i became (somewhat) enlightened to the abilities of cold-rolled sheet steel. sheet steel is plastic, the verb. the lid required construction of metal structure that conformed to a specific curve -- the Golden Rule -- in three planes. i made it so with a cheap ball peen hammer and a piece of plywood. (OK, and a plasma cutter and MIG welder.) this is "ten foot" body work -- by choice and by limitation. but it is functional. it is stiff and light.

cockpit rough-out

with the lid done, defining the back half of the cockpit, it was time to rough out the rest of it. in the original car, the instrument panel occupied the space below the windshield, and hid all the ugliness of the clutch, brake and parking brake pedals, electrics, heater, wipers, instruments, etc. since all of that was scheduled for removal everythign needed to be revisited.

the placement the steering wheel was a given, at the end of the steering shaft out of the old steering box. everything else was an open question. it was rapidly obvious that the pedals and attendant infrastructure would go in the floor, where this stuff would be for the first 30 years of automobiling, and where most hot rods have them. i further decided to make my own. this took place, incrementally, from this point (december 2013) through completion (october 2014).

the peculiar tripod that holds the steering shaft up is peculiar for a reason -- the tripod provides the required critical extreme stiffness, but the placement of tubing keeps the vertical space very short -- the instrument panel attaches to the cowl just below the glass. the support tripod cannot be further down the firewall because right at the base of the steering wheel is the head-end bearing and support. the adjustable sleeve holds the stock Nash bearing assembly. not obvious is that at the firewall is a nylon bushing that locates the steering shaft when you lean on the steering wheel; given the combination of shaft lengths the "leverage" at the nylon bushing is adequate to keep everything very stiff.

pedal assembly

though i meant this website to be an overview, by necessity this section, the pedals, is am overwhelmingly important functional area. get this stuff wrong and the car is unusable. this mundane-seeming stuff is the very core of the human interface; done right, they become invisible, unnoticed. get it wrong, and it could be literally fatal or at least, uncomfortable and/or inconvenient. i spent a lot of time getting this right. two iterations were made, the first was scrapped. unsurprisingly, what i ended up with looks a lot like commercially available race car stuff. however, no one makes anything to fit Ramblers. and i didn't want a race car, or to look like one.

chassis cleaning

the largest most difficult single task in this project was cleaning, especially the unibody. talk about understatement -- cleaning just the bottom of the chassis took three full weeks of eight-hour days. scraping, chipping, grinding, sanding. easily 100 pounds of undercoating, sound-proofing, rust and dirt were removed. the sound-proofing/undercoat seemed to consist of asphalt and clay; cheap and heavy. the purpose of said material is evident when you whack this car with your hand; it now rings like a drum. light weight has it's own costs.

the entire chassis was scraped to bare metal or good primer and painted with Eastwood Chassis Black. this is the source of the "shiny side down" tag line, and befits the industrial aesthetic. utterly unsuspected from above, the working parts of this car look like aereospace gear down below. this is no accident. few will ever see it, even fewer will bother to look. this is not of concern to me,

chassis front

the front quarters of the car contain the engine and front suspension. rather mundane and uninteresting in spite of the critical function, nonetheless a lot of time gets spent here, building and maintaining.

an oversight in this website is the elimination of the opening doors -- roadsters do not have doors, passenger cars do. it's the roof of a car that provides box-like stiffness, necessary for safetly and reliable operation; most "chop jobs" ignore this and the car sags in the middle. i added triangulated DOM tubing sub-framing between the a- and b-pillars, which ties in with the lid support subframe and the skinned firewall to make for (what i hope is) a very still box containing the passengers. it's certainly safer than factory, where one layer of 18 gauge sheet steel, the windup window, and a cardboard panel was all that was between you and the truck about to T-bone you.

rear quarter

at this point in the project i was confident of completion. everything to this point was contingent on actually executing the fundamental ideas behind the roadster: firewall, a-pillar, doors, sedan-to-roadster in-fill. those are the big chunks. now it was time to move on to important but functionally-secondary aesthetic changes. here, reshaping the wheel cutouts.

automobile styling from the 19th century through the 1970's was dominated by the "classic" -- the flowing fender, round in front, tapering to a point behind, a teardrop shape. explicit through the 1930's and 1940's, it remained a style element post-war, clung on through the slab-side movement of the 1960's, and only with the (welcome) assault of asian imports, did the "classic" teardrop line disappear. GOOD RIDDANCE.

in slab-sided nash ramblers, the classic line took the form of teardrop shaped wheel openings. that's what really makes a post-war car look fundamentally old! and at a minimum, it was a functionless decoration. there is a difference, sometimes blatant, sometimes subtle, between decoration of an otherwise necessary shape, that is either acceptable or outright desireable. then there's pointless decoration; nash wasn't so bad about this, but the wheel cutouts screamed design conservatism at it's worst.

i attack the hardest parts of projects first, in general, and specifially in this case, the rear quarters were it. though the car appears to be flat on the side, nothing in any car is actually flat flat. everything is curved. my first one (passenger rear) is fairly awful. the rest are OK as i learned technique.


the valance is the thing that you scrape on those concrete stops in parking lots. it's a dent collector. in cars of this vintage they mainly hide behind the bumpers -- which were clearly deleted by me, here. i ended up making this part completely. only portions of the curve at each corner are "original".

less obvious, but critical to the redesign, is that the front lower edge of each front fender is pulled in nearly four inches. a major "feature" ("wart") of the post-war Nash Airflyte design that rambler inherited is this feature where the tall skinny wheels are tucked well inside the fenders, necessary for the "streamline" but later derided as the "inverted bathtub" look. talk about old-fashioned... by pulling the fender leading edges in, exposing the tiniest corner of the tires, the upside-down bathtub looks is hopefully dispelled.

the same was done to the rear quarters, to a lesser extent. it's the front "bathtub" that dominates.

front fenders

at this point i had this process down. these went fast -- about a day each -- and i did a bit of metal shrinking on one.

i completely messed up the placement of the rear quarter wheel cutouts -- i could now claim that it was all intentional, but it wasn't. what began as serious dismay turned into a feature when i learned to see it properly... i did the rears first as stated, but without the axle, wheel and tire in place, going by measurement. the wheel openings are almost perfectly circle with the tiniest bit of increased radius towards the back -- yes, the classic! reduced to microscopic proportion... i made two mistakes: too large a diameter, and placed the new cutout off-center (it was meant to be perfectly concentric with the axle). i discovered this when i installed the axle months later. it took me a couple days to get over the rather severe disappointment -- until i realized that by placing the too-large radius up and back, it created a "nautilus" effect, invoking a bit of motion. so i duplicated the error on the front and lo! perfection.

door armrest indent


driveline and suspension

it took approximately 9, 10 months to get to this point; the rest was easy, though there was lots of trivia to deal with. the driveline -- engine, transmission, axle, suspension, brakes, etc -- was lifted almost as-is from the donor, my 1963 Rambler American, which i had built to a decent state of tune. you can read about that work here.

however some changes were made, some substantial. for reasons of light weight and performance the awful, worn-out and expensive to replace rear leaf spring setup was replaced with air springs. after going through a lot of designs (it's safe to say mosthot rod rear suspensions are terrible) i ended up with a single leaf for axle location, an air spring for support and suspension, and a torque link to deal with spring wrap-up due to braking torque (certainly not due to non-existent engine torque).

there is no "aftermarket" for this nash/rambler chassis. zero, none. not one part. but nash's clean, frugal little design was a pleasure to work with. i kept with the monocoque design as much as possible. time will tell if it's adequately sized, but the torque link is incorporated into the simple spring perch, and torque at the far end of a parallel pair of 1/8" levers anchors in the unibody above, near the b-pillars where it is especially stiff. i handily lopped 50 lbs! of unsprung weight per side. there is provision for a panhard rod, that i'll get to "some day" but an informal check of lateral stiffness was surprisingly good.

rear suspension

the rear suspension is at the moment a heavily modified single leaf, small convoluted air spring, and a torque link each side, sitting the car on the original live axle. given the terrible front:rear weight distribution, i needed to remove as much unspring weight as possible, and it had to be adjustable since i couldn't know how much mass had been removed until it was on the road.

i did a lot of research on rear suspension design, and rejected most "hot rod" suspensions as woefully inadequate -- many of them don't tolerate body roll (eg. one wheel up and down) and nearly all are designed for straight-line drag racing, in which i have zero interest (i prefer turning corners). given my choices and budget i stripped the stock leaf spring stack to a single leaf and augmented it with the air spring. the torque link limits spring wrap (due to braking, not so much from the non-existant engine torque). this scheme removed 50 lbs of unspring weight (nearly the entire leaf spring stack).

this has worked out fairly well in practice, though i'm running only 10 PSI in the air springs and overall it's too stiff. my medium-term plan is to ditch the leaf entirely, make a parallel linkage to the rear with enough compliance in the rod ends to allow for body roll without binding.

the steel limit straps have since been replaced with vinyl coated braided cable. the straps twanged and scraped. cable is lighter too, and will be replaced with nylon straps when i make the parallel link setup.)

all four corner air springs are adjusted with the simplest tool for the job: a double-action bicycle air pump. it weighs nothing, trivially simple, the air plumbing for each spring is about 24" of tubing and a chassis-mount Schrader valve. drop-dead simple and reliable.

there are very few replacement parts available for this car, never mind performance parts. i was unable to find even replacement front spring eye bushings that fit. i found some nice urethane bushings with insert, but they were too long; it turns out that a grinding wheel works great to remove material from them -- it doesn't melt, it just rolls up. nice!

big nut axle

the infamous (to rambler folk) "big nut" axle is roundly condemned by all. i'm like, meh. so what, just buy the right wrenches, $50 from Amazon. though it is true you need to have it up on a lift to wield the comically large wrenches to tighten the Big Nut. i do not know why it did not occur to me years ago to simply chop an access port in the floor to wrench from above. as is often the case, persistence wins out.


i spent a lot of time researching and building this motor some years back when it was in my 1963 Rambler American hardtop. it is a nash/rambler 195.6ci inline six, based upon the old Nash "Flying Scot" L-head engine. it is not well loved by anyone; it's heavy, sluggish, hard to find parts for, and has some stupid engineering problems. additionally not one signel performance part was made for this engine. that just all sounded like a challenge.

the work long since done, i plucked it forth from the donorand stuck it in. i degreased and repainted it my new color (Ford Red), and replumbed the PCV system and simplified/cleaned up things from my few years of experience with it in the other car. this part of the project was fairly straightfoward. the new clutch linkage worked out well; the factory bellcrank system flipped backwards interfaced with my floor mounted pedal system's pullrod. worked first time. the throttle linkage remains a PITA, as they always are; a deeply critical part of the feel of a car, get that wrong, and i'll never be happy. its' improving, now, is all i can say.

the thermostat housings on these engine are always rotted. there must be some source of electrolytic current that eats them. remind me to put a sacrificial bar of zinc in the radiator. even new ones rot. so i soaked mine in rust remover for a week then entirely coated it inside and outside with JB Weld. no part of the metal is exposed to coolant, now. i hope that doesn't mean it eats the block or head.

the '61 chassis originally had an automatic transmission, my donor is providing the manual transmission. surprise! the position of the engine is different in manual transmission vs. automatic transmission cars. the rear crossmember is diffferent, as it usually is, but the front engine mounts are shifted to one side by an inch -- with a manual transmission the engine is centered in the chassis, but with an automatic, it's shifted to the right, quite a bit. mysterious! i found this out of course the hard way. easily fixed since i had the right part.

in these pics you can better see two of the modifications i'd made and documented earlier; addition of a full-flow oil filtration system, and equalization of the center two exhaust ports. i'd also worked up a fix for this engine's most serious engineering flaw, a lack of coolant flow during warm-up that aggravated cylinder head sealing problems; i reduced that fix down to a strategically placed "leak" drilled into the thermostat and eliminated the plumbing.

front suspension

more or less transferred directly from the American, with new shocks, the major change here was ditching the nasty old wire springs for the air springs. even though this is the heavy end of the car, and this engine is quite heavy for it's displacement, at some 600 lbs, i still had to use the smallest air springs made, and they run at an absurdly low 30 psi (typically 60 - 80, maximim 100 psi). where hotrodders with fat cars with massive engines are worrying about getting enough spring in there, my little car is standing on it's tip-toes with barely a puff of air from a bicycle tire pump. which is in fact what is used to tune the chassis.

rather than use a chunk of pipe at the top of the air spring, as is customary, i made a lightweight perch from .125" hot-rolled stock. light and stiff once again.

the Goodyear air springs were a perfect fit -- every dimension was correct. even maximum extention matched what the Rambler stock suspension provided with the bumper on the upper A-arm. it's like Grandpa Nash wanted air springs in it.


this is a drum brake car. a high performance drum brake car. the front brakes are 9x2.5's borrowed from a 1976 Gremlin that did not need them any more; then they were substantially improved. the 1963 Rambler American donor had a nice disc brake setup but the modified drum system worked just as well and was substantially lighter. another factor was that since the master cylinder is in the floor, it is at or below the fluid level of the wheel cylinders/calipers, which would require check-valve tuning for the discs; drum brake systems have built-in check valves and make this a no-brainer. without waiting to see if i needed, it, i worked out a brake bleeder for the front brake line, from a brass tee and a "repair bleeder". total cost about 10 bucks. works great. all in all the drums are a win. the current fashion in performance cars seems to be everything possible as big as possible even when it's foolishly expensive and counter-productive. these drums are demonstrably more than adequate -- i did a hard vintage rally on them, Tepesquet Road for those SoCal ralliers, plus a few years of daily Los Angeles to Irvine in all weather. drop-dead great.

you can see the staggered pattern of holes in the drum below, as described in the drilled-drum page above.



this was the least stable part of the entire project. the irony is that i certainly have more expertise and experience with electric and electronic stuff than metal. the real reason for all the thrashing i did is that i know reasonably well how and why things fail, and i was juggling aesthetics, simplicity and reliability, with cost and ease of construction and maintenance.

the short story is that i ended up with a fairly ordinary instrument panel stuffed with Stewart Warner electrical gauges, plus a computer to manage the lighting, especially the rather unexpected complexity of nose and tail lights.

i fabricated three complete and completely different computer instrumentation systems, including multiprocessor intercommunication, information display, redundancy, self-checking, and more; got them all completely running and mostly debugged -- and abandoned each for Various Reasons. as the car came together and closer to on-the-road status, the homebrew electrics were the one thing i was afraid of failing on the road and leaving me stranded. it wasn't my abilities to design in reliability that bothered me (arrogant or not), it was the complexity and inherent delicacy of commercial-grade components in "hobbyist" stuff like Arduino controllers. those of us with more than modest experience in electronics knows that interconnections are a major source of problems; every wire, solder joint, and especially connectors, are inherently prone to fail. and home-brew electronics means a lot of non-printed-circuit construction. a core concern for this overall car project was drop-dead reliability, fail-safe construction, and field-repairability (and secondarily, field-recoverability, eg. kludge and limp home). for example the Ford EDIS-6 ignition requires an add-on computer (the Megajolt Junior Lite), made with commercial stuff, and in a non-automotive case, but EDIS has a very reliable "limp home" mode where without a functioning computer, it runs at a fixed 10 degrees BTDC timing.

at the moment engine management is traditional; ignition and starter driven by the key switch, Stewart-Warner gauges driven by traditional sensors. all other functions -- cooling fan management, fuel pump, lighting, turn and brake lights -- are software driven. as failsafe, all computer managed functions can be manually overridden with a jumper wire at the side of the road, if necessary to limp home, except turn signals and brake lights; being essentially "single filament" lamps (LED arrays) these require softwre drivers.

speaking of taillights, the first pass was a string of Adafruit's Neopixels, but they weren't bright enough for daylight use; i finally installed 13 one-watt red LEDs with series resistors and a big fat MOSFET and control them from the firewall-mounted computer. i may eventually replace those with 1-watt RGBs and put all the multiplexing logic in the tail light, and add a brakelight failsafe input (ground=ON). however, the inner rear wheel well is not a friendly place for electronics.

you can see the abandoned electrics in the dead ends section below.

i'm using Arduino Mega 2560's in the computer(s). they're cheap, have enough I/O pins i don't have to think about it, and i needed the code and RAM space not available in the smaller Atmel processors. the software is based upon a cooperative multi-tasking model, rigorously built with state machines and timers. it is a very reliable and simple model for structuring software that squeezes maximum performance from cheap hardware and is inherently stable (eg. heap size is predictable and is checked in realtime).

the code is visible/downloadable below, though if you really care about the latest version contact me, as it is certain the one here will be soon superceded.

the brake light switch, and the instrument panel switches, are 1960's aerospace surplus. i bought them from Apex Electronics in Sun Valley, sealed in air-tight foil packets, made in 1967. i should be wary of falling in love with objects like these; two of them have failed so far. sawing one open revealed immaculate, aerospace-quality innards, so i suspect light oxidation is the suspect, made vulnerable by the very low currents used for computer interface (1 - 5 milliamps). i carry spares... and i will add a "heavy" (20mA) pullup resistor so that the switches carry sufficient "sealing current" to overcome the thin oxide film.

dead ends