01 may 2023
Few automotive ideas are as reviled as the "feedback" carburetors of the 1980s (the "Malaise Era"). Though they worked fine when new, as their control systems aged operation degraded into maddeningly difficult to diagnose terrible performance.
The problem was the control boxes, not the carburetor, which was generally an improved last-years-model modified to electrically modulate fuel/air mixing and a bunch of passive changes to reduce evaporative loss. The control boxes however were a mashup of quirky undiagnosable hoses, check valves, switches and dumb sensors and software cobbled together by the carb manufacturers, who clearly knew their product was a dead end, and probably not too enthusiastic about spending money on them.
Electronic fuel injection (EFI) has been the superior choice for 40, 50 years. Early EFI systems were more or less "electronic carburetor replacements", throttle-body EFI bolted onto carbureted engines, used a dedicated controller, and did a fairly good job. General Motors' TBI sytems were as reliable as a carburetor (which after 60, 70 years of development were quite reliable), yet "stand alone" enough to be adapted to other engines.
Around 2005 I put a 1980's-vintage GM TBI system on my Rambler wagon and it was great. After tuning it was a reliable 20 mpg in a car that typically got 18 mpg which wasn't bad for a 1963 car.
GM's TBI was inherently reliable, and parts were available at any parts store, meaning if one failed on the road, as unlikely as it was, it was repairable.
Modern 21st century EFI is no longer "stand alone", it is deeply integrated into the engine and car electronics, and so no longer a candidate for retrofit. And 1980's TBI systems are now antiques.
There is a robust amateur EFI subculture around Megasquirt, very well-developed software plus controller hardware that uses OEM components. Unfortunately for me my engine doesn't have an intake manifold, or any means to install injectors; and throttle-body hardware has been abandoned as a dead-end.
In the last few years (writing in 2023) commercial aftermarket EFI has become more-affordable. In 2021/2022 I installed a Holley Sniper Autolite 1100 EFI system on my little Rambler wagon. It is a throttle-body type system, designed for bolt-on adaptation. That part worked great. I got it to work fine running as well as the Carter YF it replaced -- but it proved to be unreliable. The throttle body suddenly stopped delivering fuel, I sent it back to Holley for repair, they fixed it for free, told me it was an internal connector issue, reinstalled it continued to work just great.
It proved to me that it was not a device I could rely on on long trips. An on-road failure with a two-week factory repair turnaround time is not acceptable.
Carburetors suck in many ways, but their failure modes, as in "sudden" car-stopping leave-you-stranded failures, are extremely limited -- floats sink, needle and seat stick. They definitely degrade with time, leaking, throttle shaft wear, requiring constant adjustment, fiddling with the choke and idle -- but with any maintenance they do not leave you stranded on the road.
The the Sniper came off, got sold on eBay, the 1970's AMC Gremlin Carter YF went back on. I had it dialed in to perfection; warmed up and driving, at low altitudes, it performed as well as the Sniper. And I couldn't trust the Sniper system enough to benefit from its advantages with altitude and changing conditions.
I'd long wondered about what "feedback" carburetors did and how they worked; no one knew, everyone who had one yanked it off, did the "Nutter bypass" or otherwise disabled the solenoid/stepper mechanism, etc. I had searched for years for documentation on theory of operation and never found a word.
One day a newly-worded search term found the pot of gold -- a Carter/Weber document on the 1980's YFA Carter YF/YFA Service Manual #3608B which towards the end of the document had exactly what I had been looking for, which is electrical specs for the solenoid and what it did (12V, pulse-width modulation, 10 Hz). I can tell you also that it draws about 200 milliamps, minimum pulse open is about 5 mS so it has a good dynamic range.
The only serious difference between the feedback YFA and non-feedback types is the solenoid hanging out on the side. On "altitude adjustment" YFAs there's a simple cover plate over the two holes in the top casting. Plugged, it's just a carburetor.
The solenoid provides an electrically-controlled connection ("valve") between the two ports/holes in the casting. One leads to the air intake, the other into a small passageway in the emulsion tube.
The solenoid here came on the used carburetor I bought from eBay; it failed within a week, clogged from years of use. I disassembled, cleaned, and reinstalled it and it's been fine over the last few months. I carry the spare NOS solenoid I bought, also from eBay, should this ond one fail on the road. But it's a good design, and the rubber seal was soft and undamaged, the only issue was oily grit in the solenoid plunger cavity.
Please refer to the feedback carburetor controller document for details of operation, and how the controller maintains closed-loop control of the air/fuel mixture. The controller is completely automatic, needs no interaction of any kind while driving, though it does provide some "electronic dashboard" insights into carb and engine operation.
Alas, it is not a simple bolt-on fit on the old Nash/Rambler six. It is probably an easy bolt-on on the "modern" AMC six (199, 232, 258). Though essentially the same carburetor internally, the YFA is shorter to fit under lower hood lines, and to accomodate larger throttle bores for newer engines, the base pattern was widened (the biggest impediment to adapting it). The idle mixture screw is in an inconvenient location, and the wider body hits the valve cover.
For the early Rambler engine I fabricated a two-piece aluminum adapter that attaches to the trough cover, shifts the carburetor outboard a half inch and blends the offset and increased bore. This improves access to the idle mixture screw, though I had to fabricate one from an allen-head screw since the one that came-with was bent (it was surprisingly easy to make).
Hopefully you won't have to do through this nonsense.
It's still a carburetor; you need to tune it. The controller can only lean out the mixture, not enrichen it. the general strategy is to tune for "sea level", the controller will lean it out as necessary.
I will document tuning strategy eventually, more speedily if anyone asks.
I have an in-dash Autometer wideband air/fuel ratio meter in my car, and I've been installing one in all of my carbureted cars in the last decade or so. They're so useful I can't imagine tuning a carburetor (in my case, rarely the original) without one.
The controller requires a wideband A/F meter. The Autometer, at least, has an extra wire on the back that outputs a voltage proportional to A/F ratio and is tagged as "for a datalogger". My controller requires this signal.
Here are my tuning jets and metering rods notes in some detail.
Historically, carbureted cars were tuned to err on the rich side, not lean. All engines run smoothly (...) when too-rich. Not well, but tractably. Too-lean cuses flat spots, backfires, stalls, etc. This is in tension with best economy acheived at very lean mixtures (slightly over 16:1). Carburetors are simply not that precise; and even tuned closely, are susceptible to change with altitude, temperature, etc.
With feedback control the carb can still be set to the rich side, then leaned out under automatic control. This is what makes this system so nice as retrofit for old cars: the drop-dead-reliability of a carburetor, and software control of the mixture at all times.
(This does not address NOx (nitrogen oxides) and other issues caused by lean running; in the table below NOx peaks at peak-economy: this is what EGR (exhaust-gas recirculation) addressed; it lowered combustion temperatures that messed with atmospheric nitrogen. This might make an interesting side project...)
Here is the Carter
YF/YFA Service Manual #3608B, with explanation of how the various
circuits work (theory of operation) including the only description of the
operation of the feedback version of this carb I've ever seen
(briefly: 10 Hz PWM, ON leans the mixture).