Jim Bartuska , who had the data, graciously allowed me to look at the wiring diagrams, and logic of the control system of the 1958 Bendix Electrojector—something I have long been curious about; and Gloria Moon had a file of internal Chrysler engineering test reports relating to this effort. Various documents were then selected out, and are a part of what follows. We have put the most relevant in PDF now.

The following info and quick evaluation is from an electrical engineer’s personal retrospective view, in May 2018, of the actual prints. The detailed prints are Jim Bartuska’s personal property and not intended to be disseminated, but one can see certain things by examining them.

First, the system is 100% analog, designed around very early PNP germanium transistors, some of which appear to be made by Bendix in house, and some are Raytheon sourced. I would expect both were derived from early 50’s military R &D. PNP circuitry or logic is inverted in a sense from vacuum tubes and later more common NPN transistors, such that the power supply must be negative with the positive side grounded, or “pretended to be grounded” (aka, low) to make it look “conventional”. This lends lots of confusion to most EE’s today, (including me) used to + at the top and conventional current flow, flowing down to (-) as everything is “upside down” although logically/functionally the same result. Kind of like having to read a book from right to left, or bottom to top, for non-engineers. PNP current goes in at a transistor’s emitter, while NPN or vacuum tubes it comes out, so all the signals “go the other way”. So the logic of the circuitry can be frustratingly obscure to grasp easily or intuitively at first. My limitation.

Primarily, the circuitry is generating a variable xx millisecond wide pulse for each injector squirt, at each intake stroke; the duration of that squirt is directly controlled by various “calibrated” or “curved response” sensors, which is then amplified and so moves the injector valve to open, via a magnetic coil. So, one squirt per power stroke, at the intake. Timing the squirt start is by points, just like a spark point set.

That said, these same Bendix transistors made the first “hybrid” car radios possible, in their bolt down metal cans (TO-3), and were used in those hybrid radios, which also used special new 12 Volt operable miniature tubes, instead of prior 100’s of volts on radio tubes; only the final audio frequency power amplifier stage used the one or two PNP germanium (not later silicon) transistors. The audio amplifier was also “upside down” internally, compared to the external radio supply of +12. This mostly eliminated the vibrator powered 250 volt tube radios on 6 volts, common up to 1955 or so, although some 12V vibrator radios were built, too. 12V was required, for getting rid of the vibrator, so the introduction of Bendix transistors tracked the change from 6 to 12v and may have been one of the main drivers of that large industry change about 1956.

The first big takeaway is that this FI system is 100% open loop. With no oxygen sensor, or feedback, it has no idea where it is really running; instead, it is relying on hopefully stable preset calibrations, and what must be strange and highly compromised analog curves built into all the sensors, such as manifold vacuum, throttle position, and temperature. With all the variations in temperature under hood greatly impacting the early transistors, and varying supply input voltages, (I do not see any internal voltage regulators at all) it must have been all over the place on AF ratio. In one related “test” document provided by Gloria Moon, it discusses how carburetors could be held by Chrysler specification to about 2% AF variation – surprisingly good, (!) to me, (Carter, metering rods, etc. — WCFB was a highly evolved thing), while for this EFI Chrysler had to accept 8%-- or else none of them would pass. So claims of improved performance probably are for sure wishful thinking, at the very best. Wrong A/F ratio equals bad gas mileage and/or poor performance, plug fouling, low HP, stalling, you name it, unless someone is hovering over it on the dyno with the “Golden Screwdriver” peaking the HP readings at one best RPM.

The transistors had very low gain, known as ‘beta’ and yet had to run at high currents, to run the injector coils, so a big box of hot resistors was involved in biasing the power transistors, probably switching several amps of current (the external resistor box). The pulse width was intended to be about the same as now, roughly 1-10 msec, but the injectors from Bendix were way too slow, often taking 5-6 msec to close after the signal was removed. So, with that issue, one cannot really control anything…. Chrysler spent a lot of time on this aspect, improved it greatly, but never got to reproducible items that were fast enough. So injectors had to be hand selected or – ideally matched as a group, as each one was driven with the same pulse width, even if mechanically it might be slow or fast, immediately causing rich or lean cylinders. And then, from there, it drifted…. Change one, the AF ratio is probably off on that cylinder.

One of the worse problems was leakage from the injectors, which might then fill a cylinder with gasoline, causing hydraulic lock and bent parts, but also oil dilution with that gasoline, resulting in wiping out cams, lifters and ring seal. That means new engine needed. Various crude things were attempted to “fix her up” with band aids, such as tripling the pulses when starting (no accelerator pump, no real choke), an acceleration sensor that arbitrarily richened the mixture in a step, when floored, etc. etc.

A particularly troublesome part of the design approach (to me) was a mechanical “distributor” that sent the 12V pulses, coming alternately from each of the two duplicated control channels in the smarts box , to one of each of its related 4 injectors. This involved a pair of traveling mechanical rotary contacts touching terminals in succession. Not “near them”, like a spark distributor, (it jumps a space there, very high volts=easy) but actually touching them and carrying 12v at an amp or so. This is lucky to last 1000 miles, in my opinion. Similar to Model T Ford ignition, with 4 buzz boxes and traveling roller.

Despite all this, the concept was 100% correct, timed millisecond squirts, and varying the squirt duration to best match engine conditions. How it works today. But it took a digital computer and an oxygen sensor to let it learn the correct precision A/F curves as you drove, and to store and or revamp them; and a long difficult engineering trek to come up with fast injectors that not only work quickly – but they will dependably stay closed, not leak. Today each one has its own transistor driver and calibrates itself, for the most part.

In recent years it has been found that carburetors usually give more HP than FI, too. This is because atomization/ evaporation of gasoline far upstream of the inlet valve gives more time for that evaporation to cool the incoming air, adding more oxygen due to higher density. This was true in 58, too, although the 300D carb manifold was not very good, pressed down due to low hood, and of single plane design.

Another of Gloria’s documents describes how the service department essentially said “We have had it with the cost of this”, causing cancellation, and the statement from them that “The only cure for all the difficulty is to remove it” and instructions how to do that…removed units were supposed to be sent back to Chrysler. About 80 were on the initial Bendix order, which was apparently not completed. Add in engine failures and all warranty labor…the service department was probably run as a profit center, this was a brutal $ problem for them. And very upset customers.

Finally anecdotes that neon signs etc. impacted it cannot be true, as the transistors of 1958 were very slow and would show almost no response to radio signals, or electrical noise. Inside, it is working at very low audio frequency rates; and sparking going on in the ignition cap right next to it had to be a 1000x larger interference problem, if there was one. I doubt there was.

I have great admiration for those who tried to do this – along with admiration for a wildly optimistic, golden voice Bendix salesman who “got the sale“, but it was just WAY too early in the game. The decision to remove it was absolutely correct.

With even greater respect for those who tried,

John K Grady, PE