I'm a new member of the AAC community and a neophyte to all things circuitous. I know just a little, and am in awe of the expertise here.

I'm investigating the practicality of building an analog EFI system for use on 30-80 HP aircraft engines (4 strokes--V-twins to air cooled VWs).
I ran across the circuit below that was designed for this use, but it's not clear if it was ever flown (more explanation/detail is at this archived copy of the rotaryengine. net site: https://web.archive.org/web/20190913133244/http://www.rotaryeng.net/simple-cheap-555.html ). There are some inconsistencies within the schematics.
I'm sure this approach can never achieve the accuracy and "hands-off" operation of a true digital EFI system (with many sensors, a fuel lookup table, maybe closed loop operation with lambda sensor, etc). But an "old school" approach free of software/firmware is relatively easy to understand, can't "crash," and has the >potential< to be very reliable. Most pilots were trained on aircraft with carburetors and manual leaning is not seen as objectionable.
My (first!) questions are below the schematics.
Below: Board layout and main schematic:


Below: Sub-circuit for the trigger system, the injector driver (which differs a bit from the main schematic above--the main circuit at top includes an LM78LO5 linear regulator and different capacitor values around it), and the leaning circuit.


As I understand it, things work like this: The injection process is triggered when an external timing probe/tooth passes the trigger sensor (the document described using an antilock braking system (ABS) sensor for this). The trigger circuit converts the "full sine" signal to a 12v-0v-12V spike needed by the TLC555, and it is fed into Pin 2. The duration of the injection event (the pulsewidth) is set by the leaning circuit. The leaning circuit uses input from a manifold air pressure (MAP) sensor and a potentiometer knob in the cockpit, through a TLC279 opamp, to produce the voltage input signal to Pin 5. The output from Pin 3 energizes the injector for a pulsewidth dictated by the leaning circuit, and the return voltage through Pin 4 resets the TLC555 to repeat the cycle when another trigger is received through Pin 2. In this 4-stroke engine, there will be two injection events (triggered by the crankshaft) for each time the intake valve opens for an induction event. The fuel will just wait in the intake runner and get taken in when the air moves.

I have some questions, and would appreciate any assistance in aiding my understanding:
1) That leaning circuit. Let's assume our manifold air pressure (MAP) sensor is inside an intake runner that feeds just one cylinder. On a 4-stroke engine the crankshaft rotates twice per engine cycle, so the trigger sensor will call for two injection events. The MAP reading when the triggering events occur will feed the leaning circuit and determine the pulse rate and how much fuel gets injected. The pressure inside the intake runner can be expected to be the same as the outside air pressure for about 75% of the time (throttle at least partially open, intake valve closed, so pressure in the runner equals outside pressure). For the other 25% of the time, the pressure may change (decrease) rapidly as the piston pulls in air against a (partially closed?) throttle valve. At most, our MAP sensor (and, thus, our injection duration) will be affected by the pressure drop due to the intake event only once for every two injection events. The other injection event will be "leaned" by the MAP sensor only to the degree that the outside air pressure is decreased (due to altitude, etc). A longer and shorter pulse per cycle is no problem, but I'd need to figure out some way to scale the resultant input(s) so that the total fuel injected per cycle is correct. Or, is that what the leaning knob/pot in the present circuit does?

2) Semi-related to Question 1: If we are talking about an airplane with a fixed-pitch propeller, then we know the required fuel for each induction event will differ a lot. This is very different than the situation with a car (with its transmission and gearing). The load imposed by the propeller varies as a function of the >cube< of the RPM. Also, the efficiency of the engine changes with RPM. As an example, at max power and 3600 RPM, our little engine might need a total of .1 ML of fuel per induction event (two crankshaft revolutions, so total of two squirts), but at 1/2 power (2800 RPM), we'll need only .05ml per induction event. If we can get enough adjustment through the MAP-derived adjustment of pulsewidth, maybe that will suffice. But, I can see that it might be useful to somehow adjust/scale the input from the MAP sensor to provide a better fit to the actual fuel requirements after we test things a bit. Is there a way to do that?

3) The overall suitability of the circuit: Paul Lamar did a good write-up on his development of this idea/circuit, but I'm not sure it has ever been used in an airplane. Obviously, this needs to be reliable. Planes can have a lot of electromagnetic noise (magneto ignition systems, etc) and sometimes the electrical power isn't very "clean." Opinions on whether that NTE67 FET will be up to handling the backvoltage from the collapse of the injector coil? Suggestions for "hardening" this circuit and ancillary sensors/etc are welcome. Or, maybe it is just fine as it is.

Thanks very much for any assistance. I hope to learn a lot and not be a pest.

Mark

 

Welcome to AAC!
I'm surprised and concerned that you're looking for advice from a bunch of anonymous nerds on a forum, considering the safety implications of this project. We try to help, but if you rely on advice from this site it is entirely at your own risk. This site and its members accept no liability.
With that said, ......

Or, is that what the leaning knob/pot in the present circuit does?

Yes.

Is there a way to do that?

The leaning circuit shown does scale the input from the MAP sensor, by an amount dependent on the potentiometer setting.

Opinions on whether that NTE67 FET will be up to handling the backvoltage from the collapse of the injector coil?

I'm doubtful. I don't know if the specific injector mentioned incorporates any BEMF protection, or if the inherent body diode of the FET is adequate for the job. According to its datasheet the NTE67 current rating (3A continuous @ 100C) seems rather feeble. I'd go for a much beefier FET for handling modern injectors. A dedicated injector driver IC might be a better option, since it would provide peak-then-hold control of injector current.
It's not clear from the circuit how the 15V supply to the electronics is derived. That would certainly need to be hardened, to protect from load dumps and transients generally.

 

you should get the full specs of the device you attempt to drive 1-st
something ? https://www.google.com/search?q=mazda+electronic+rx+injector+wiring+diagram+reaction+delay+datasheet
returns ? https://www.service-engine.com.ua/webroot/pdf/for MAZDA 2.pdf ?

 

Thanks for the reply, Alec.

Welcome to AAC!
I'm surprised and concerned that you're looking for advice from a bunch of anonymous nerds on a forum, considering the safety implications of this project. We try to help, but if you rely on advice from this site it is entirely at your own risk. This site and its members accept no liability.

Yes, I should have made that more clear. All suggestions will be appreciated, but I'm not depending on anything said as necessarily applicable or even right.

The leaning circuit shown does scale the input from the MAP sensor, by an amount dependent on the potentiometer setting.

I'll dig into the datasheet for the TLC279 and hope to understand it better.

I'm doubtful. I don't know if the specific injector mentioned incorporates any BEMF protection, or if the inherent body diode of the FET is adequate for the job. According to its datasheet the NTE67 current rating (3A continuous @ 100C) seems rather feeble. I'd go for a much beefier FET for handling modern injectors. A dedicated injector driver IC might be a better option, since it would provide peak-then-hold control of injector current.
It's not clear from the circuit how the 15V supply to the electronics is derived. That would certainly need to be hardened, to protect from load dumps and transients generally.

The power would come from the approx 14VDC produced by the engine's alternator and through the charging circuit. This can be a little rough, especially when yhe starter is engaged/disengaged.

I see that AAC has a whole section devoted to circuit protection, I'll need to spend some time there. The TLC555 datasheet also provides some guidance and even some sample circuits for protection of the IO pins.

I guess my next step is to get a protoboard and the components and try things out. On that score--I see a lot of very inexpensive offshore LCD oscilloscopes for sale. I'd think something like this would be handy for seeing the signals from the trigger and the PW to the injector. If there are any opinions on suitable ones, or whether these very inexpensive O-scopes would work, please feel free to chime in.

Thanks again,

Mark

 

Won't you have to get FAA (or other officialdom, depending which jurisdiction you come under) approval for use of a home-built EFI system?
Will you be able to get insurance cover?

 

Won't you have to get FAA (or other officialdom, depending which jurisdiction you come under) approval for use of a home-built EFI system?
Will you be able to get insurance cover?

You are quite right to ask the question. Not to go too far afield, here's the situation in a nutshell. I'm in the US and most familiar with the environment here, but many other countries have similar regulatory regimes (though the US tends to be a bit less restrictive than most).
Certified Aircraft: This would be the typical small Cessna, Piper, etc. The FAA is quite strict about modifications to these aircraft, and who may perform them. There are a few EFI systems available for these aircraft, but they are very expensive and not very common in the lower HP ranges. The major engine manufacturers for certified aircraft still primarily sell carburetors on their new engines of less than 180 HP. The cost of gaining FAA approval for an EFI system is astronomical.
Experimental-Amateur Built Aircraft ("E-AB" aka "Experimental" aircraft). This is the bin my projects occupy. Typically contructed in homes by non-professionals from kits or just plans. The FAA allows the designer/builder quite a lot of freedom in construction methods, engine choices, etc. The plans and the aircraft get some FAA inspection. There is a (generally 40 hour) restricted flying period when passengers can't be carried and the flight location is limited, during which testi ng must be completed. After that, the rules are not very onerous: the individual who did the building can make changes, do all maintenance and inspections, etc. For a major change (and adding an EFI system would be one), another restricted flying period would need to be completed. This works surprisingly well.
Ultralights (aka "Part 103" aircraft): In the US, these are totally unregulated. To save themselves trouble, for regulatory purposes the FAA just doesn't classify them as "aircraft" at all. They can weigh no more than 254 lbs empty, can carry only the pilot, and have restrictions on stall speed, max speed, and a few other things. No pilot certificate or other training is officially required (obviously, anyone with any sense gets some training before taking flight). I don't think this category of aircraft exists in many places outside the US. To add to the confusion, the aircraft known as "ultralights" in much of the rest of the world are more similar to (and regulated in a similar way) to US "Experimental" aircraft.

Insurance: If we have a Certified Aircraft, the insurer will require that everything on it be in accordance with the rules--so no DIY EFI. For an Experimental Aircraft, each insurance company has its own underwriting criteria. In some cases they won't cover a plane/liability until after the restricted flying period has been completed.

I own an E-AB aircraft with a carburetor and an engine derived from a Type 1 VW. It has about 600 flight hours now. I'd probably >add< this EFI system to my existing carb-based system at first, to gain experience with it. My other interest is in use of small industrial engines for flight, and maybe I'd fit a primary and backup EFI system on that--after a lot of bench testing.

Sorry for the detour . . .

 

you should get the full specs of the device you attempt to drive 1-st
something ? https://www.google.com/search?q=mazda+electronic+rx+injector+wiring+diagram+reaction+delay+datasheet
returns ? https://www.service-engine.com.ua/webroot/pdf/for MAZDA 2.pdf ?

ci139,
Thanks for the links. The schematic I posted showed the Mazda injector primarily because the origina; circuit was to be used for a Mazda rotary engine in an aircraft. I'll probably fit an injector with a smaller capacity since I'll only need about 1/2 (or less) of the HP of that Mazda engine. Some of the smaller motorcycle injectors should work well. I'll need to check out their documentation and will also try to have a look at how existing aftermarket EFIs handle the voltage surge when the injector closes.
Mark

 

Sorry for the detour . . .

Not at all. Being UK-based I've had no occasion to study FAA requirements. Thanks for the clear explanation.

 

. . . the fuel demand (but also the ratio in the fed in mixture) depends on more than just the desired "RPM"

. . . air speed (to fw/bw wind speed) , ascent/descent angle , flight altitude , geo position , RH , you name it . . .

 

. . . the fuel demand (but also the ratio in the fed in mixture) depends on more than just the desired "RPM"

. . . air speed (to fw/bw wind speed) , ascent/descent angle , flight altitude , geo position , RH , you name it . . .

Yes, there are a lot of things that determine the fuel requirement. But in many ways an airplane with a fixed-pitch propeller is simpler than an auto or many other applications. Unlike a car in neutral, or even first gear, that might be able to make 6000 RPM with a very small throttle opening, there is a fairly tight relationship between the RPM of the propeller and the amount of power (and, therefor, fuel) needed to get it to turn at that speed. Airplane EFI systems often don't even have throttle position sensors (used in car EFI systems to allow for fuel enrichment for acceleration)--the throttle movements in an airplane are more smooth, and if the mixture is kept slightly "rich" the engine will accelerate right along with the throttle movement.
The MAP sensor in the schematic in the OP is there to help to provide information needed to compensate for differences in the density of the air being inducted (less density= less air mass = less fuel required for the same CCs of air). The air density can vary for many reasons, but two are most significant: Ambient air density (due primarily to altitude) and throttle position (when the throttle is fully open, the air density during induction will remain almost the same as the ambient air density. But when the throttle is partially closed, the air pressure is reduced by the suction from the descending piston drawing the air in while a lesser quantity of air can get past the throttle valve.) I understand the principle, but I am concerned about getting the scaling right. I suppose that's what the manual potentiometer is for, but I suspect there's a way to get closer to the right mixture with less pilot attention.
This analog approach is never going to be as hands-off as a digital (programmed) EFI with a fuel lookup table, etc, but I I'd like to see if it is possible to get "satisfactory" fuel-air mixtures with the right "rules" executed through analog circuits.
Thanks for the observations.
Mark

 

This is what I use for EFI sensor testing, it a cheap LCD CRO & it can freeze frame the display. It does what I want it to do for about $30. The obvious failing is there is some noise in the unit which I just ignore.

 

You could look at how speeduino.com is developing their open source fuel injection. They seem to think that stp62ns04z is a mosfet that should drive most fuel injectors.
How are you measuring air fuel mixture? EGT and CHT? A lot of interesting information about fueling aircraft piston engines is available at AVweb Piston Engine Columns by John R. Deakin .
A wide band oxygen sensor might be a better sensor but I do not know how they respond to changes in atmospheric pressure.
Since most aircraft engines are tuned for relatively low RPM (in comparison to automobiles), the volumetric efficiency curve should be relatively flat along the RPM and MAP axis. A simple voltage from a MAP sensor would then be a good approximation of fuel requirements for a specific manifold pressure and RPM. Trimming that value with a pot should give you a consistent fuel ratio as long as the MAP and RPM remain constant. The fuel ratio for a specific trim pot setting should only change mildly for small changes in MAP or RPM. If either change, you should be able to find a new trim by watching how EGT and CHT responds to the change.
You could just feed the 555 output to the circuit listed as 'Injector 1' in the Speeduino circuit diagram:

 

Ed,
Thanks for the input. The Speeduino folks are doing a lot of good work, and I'm exploring the idea of going that route (using the NO2c board) instead of the 555 analog path described in this thread. Using the same stp62ns04z MOSFET they are using to drive the injector may make lot of sense. Also, I noticed that the explanatory document that contained the schematic in my OP mentioned use of a "TO22 ST" mosfet for this purpose, so I'll have to look at some spec sheets.

A wide band oxygen sensor might be a better sensor but I do not know how they respond to changes in atmospheric pressure.

There are a few reasons I'd like to avoid using a wide band O2 sensor in this project, if I can. First--fuel containing tetraethyl lead is still the most widely available aviation fuel for spark ignition engines in the US. The lead is tough on O2 sensors, it kills them sooner or later. Many folks can get unleaded gasoline for their airplanes, but it isn't always possible. Also, for this application we'll want to be running a "best power" mixture by default rather than "best economy"mixture, and "best power" is fairly far from the stoichiometric mixture that a wideband O2 sensor is optimized to detect.
Despite all that, I'd thought of the idea of using >just< a wideband O2 sensor as the feedback mechanism, and just use the 555 circuit to adjust mixture based on that. It would make things pretty simple, in theory. In practice, though, if the mixture ever goes too rich (e.g. rapid throttle closure) and the engine starts to run poorly and "miss," then a bunch of unburned O2 gets into the exhaust manifold due to the missed combustion events. A wideband O2 sensor will pick this up ("extra O2 in the exhaust") and the system would respond by sending >more< fuel to the intake manifold, so negative feedback loop. I suppose this is one reason "real" EFI systems use Lambda for fine tuning and not for the primary setting of mixture.

How are you measuring air fuel mixture? EGT and CHT?

When I get this built and start working out the bugs, I'll probably use a wideband O2 sensor and display for fine tuning during development. Once operational, I plan to use EGT to manually adjust mixture in cruise (as we do with a carburetor).

Since most aircraft engines are tuned for relatively low RPM (in comparison to automobiles), the volumetric efficiency curve should be relatively flat along the RPM and MAP axis. A simple voltage from a MAP sensor would then be a good approximation of fuel requirements for a specific manifold pressure and RPM. Trimming that value with a pot should give you a consistent fuel ratio as long as the MAP and RPM remain constant. The fuel ratio for a specific trim pot setting should only change mildly for small changes in MAP or RPM. If either change, you should be able to find a new trim by watching how EGT and CHT responds to the change.

I think this is the logic in the circuit as given in the OP. I'm wondering how well that MAP input will work in practice with this analog approach. At first glance, MAP would seem to be a good way to capture almost all the info we need about the pressure (and therefore the density) of the air charge entering the cylinder. The MAP is a function of the ambient air pressure (changes considerably with altitude) and our throttle opening. Super! But if our MAP sensor is in a runner feeding just one cylinder, then 75% of the time there's no induction happening and that MAP sensor "sees" only the ambient pressure. This is easily addressed in a digital EFI system (smoothing, "use lowest MAP reading over the last period" etc), but it's tougher to address using an analog approach, especially if we don't have a camshaft sensor to tell when the intake valve opens. The MAP sensor gets read only when an injection event is triggered (probably once per rev=twice per induction event). At this point, I'm considering that >maybe< a throttle position sensor input (so, an "N-alpha" type EFI model) together with an ambient barometric pressure input might be better. More opamps . . .

Thanks again!
Mark

 

You can make a physical average of manifold pressure. If you do not want to combine pressure from several cylinders, you can still run a hose to a canister with the MAP sensor in it. One hose for each cylinder to be averaged and one canister for each separate average you want to sense. Worst case, you could install a restrictor in the hose. (hose/restrictor is a resistor and the canister is a capacitor - adjust to suit) N-alpha would be mentally complicated to operate without a full blown EFI and it is usually used for engines where weak vacuum doesn't really represent engine load.

I forgot about lead. It's been so many years since I had to think about it (check out Deakin's lead article). Narrowband oxygen sensors are optimized for stoichiometric, wideband should easily cover both lean economy (18:1 or more) and rich power (10:1 or more) ranges of operation. However misfires could be a problem and I would not use a wideband for direct/unrestrained feedback. Your brain and a knob should be in the loop for any transitions between static settings for a simple 555 based EFI like this.

Simplicity of 555 is fascinating but Speeduino could be downgraded/modified to just a MAP/RPM fuel table plus a target AFR trim pot input for really simple engine control. (more complex EFI but simple control)

 

You can make a physical average of manifold pressure. If you do not want to combine pressure from several cylinders, you can still run a hose to a canister with the MAP sensor in it. One hose for each cylinder to be averaged and one canister for each separate average you want to sense. Worst case, you could install a restrictor in the hose. (hose/restrictor is a resistor and the canister is a capacitor - adjust to suit).

Maybe I'm missing something (very possible). As I understand it, the purpose of the MAP sensor in a speed density system is to give us the density of the air that is entering the cylinder. We know its volume (in rough terms-- it's the swept displacement of the cylinder), and the MAP tells us the density--from that we can figure the mass, and know how much fuel needs to injected. Now, in my example V-twin engine, both induction runners are entirely independent (they have to be, or the uneven induction and exhaust rhythm significantly reduces VE of one cylinder. ( As cheap cost conscious as Briggs & Stratton is, they still buy twin carbs for their twin cylinder engines for this reason). With the throttle body even partially open, the air in each runner will be at (effectively) ambient pressure most of the time regardless of the throttle setting until the intake valve opens. But the only thing that matters at all in determining the needed fuel quantity is the MAP during the induction stroke. With this analog setup, the pulsewidth is determined only by the MAP instantaneously read just a moment before the injector is triggered. If I'm injecting on every revolution, I'm going to get that other-than-ambient MAP reading, at best, every other "spurt." An average MAP (derived electronically or mechanically) still isn't an accurate reading of the (lower) density of the actual charge (for the 25% of the time that the intake valve is open and producing suction against the TB opening) . Maybe the true MAP during induction isn't critical, and a (mechanically) averaged value can be used in some way to determine the charge density (and throttle opening). Maybe some amplifier circuit could be used to find the difference between the average MAP and the ambient pressure, multiply that average difference by about 4, add that back to the ambient pressure, and use the result as a stand-in for the MAP during the induction phase. It sounds hard to me.

N-alpha would be mentally complicated to operate without a full blown EFI and it is usually used for engines where weak vacuum doesn't really represent engine load.

Going along with my "thoughts" above, a relative advantage of N-alpha in this case is that, unlike MAP, the throttle position sensor (TPS) remains constant and available over the engine cycle. Whenever I need to inject fuel, I can get an accurate reading of throttle opening.
I'll drone on more, and supply some numbers, later. For now, I'll just offer that, with a fixed-pitch prop, there's a fairly tight relationship between HP (and fuel) required and RPM.

Simplicity of 555 is fascinating but Speeduino could be downgraded/modified to just a MAP/RPM fuel table plus a target AFR trim pot input for really simple engine control. (more complex EFI but simple control)

That's true. As ignorant as I am of analog circuits, I'm just as clueless in programming with 'C'. The Speeduino project is pretty big, and the software has grown to include control of variable valve timing, etc, etc. I just want very simple fuel injection, and am wary of the volume of code and the nearly unanticipatable consequences of every type of sensor failure when combined with that code. If Speeduino could be slimmed down so that it would be feasible for me to even assess the risks and address them, then that would probably be the route to a more accurate EFI system and one requiring less pilot action in flight. But, I've known folks who explored Speeduino and were treated to repeated unexplained crashes of the firmware. The analog approach has its own (big) challenges, but program crashes aren't among them. But, I'm not writing off Speeduino, or perhaps using it as a jumping off point for a simpler digital approach.

Thanks for your patience.

 

The variation of the air pressure value in the intake runner will be dependant on its volume relative to the cylinder volume but also throttle opening. When the throttle is nearly closed, the value should hold between pulses and when the throttle is wide open, it will also be relatively stable. At half throttle, the variations will be related to altitude. In all cases the pulsations will be consistent and will add some non-linearity that will need compensation in addition to the natural non-linearity of the engines volumetric efficiency.

When your engine is running above or below its torque peak, the fuel required will be less than the MAP would indicate. Also when the throttle is more towards the middle, the fuel required will be less than the average MAP would indicate and that error will decrease with altitude.

What that means is that you should only need to adjust the mixture trim pot whenever altitude or RPM changes but probably not by much. I didn't think you would ever get the kind of accuracy where you could mark the trim pot with air fuel ratios that you could set regardless of flight status. The best you could expect is to use the same trim pot settings for the same flight status plus x% more on the trim pot will give you y% richer on the fuel mixture.

On the speeduino, you might not have to program anything, just fill in settings or defaults. There are coolant and air temperature sensors and settings you would want to include. Throttle position is only used for acceleration which you could ignore. The ignition settings would just not get filled. You could use the existing flex fuel sensor input as your trim pot, it will give you a continuous blend from a lean fuel table to a rich fuel table. However I don't know if speeduino has incorporated barometric altitude compensation although a lot of boards have a second MAP for that function.

 

The variation of the air pressure value in the intake runner will be dependant on its volume relative to the cylinder volume but also throttle opening. When the throttle is nearly closed, the value should hold between pulses and when the throttle is wide open, it will also be relatively stable. At half throttle, the variations will be related to altitude. In all cases the pulsations will be consistent and will add some non-linearity that will need compensation in addition to the natural non-linearity of the engines volumetric efficiency.

Ed, thanks. I'll need to look at some measurements to better understand MAP variation over time in a single-cylinder runner. At 3600 RPM, the 4-stroke engine cycle takes about 33 milliseconds and induction "surge" ("suction") will take about 8 milliseconds, leaving about 25 milliseconds for the manifold to bleed back to ambient pressure through the throttle. The Freescale MAP sensor's response time is 1 millisecond. I'd just need to figure out if, in this analog approach, a MAP sensor can do what we need.

Another approach for consideration: Here's a graph from Lycoming that shows RPM vs HP and specific fuel consumption (all for sea level standard day conditions). This is for their O-360 engine with a fixed pitch prop, but we’ll assume roughly the same attributes apply to other air-cooled pushrod engines with fixed-pitch propellers. The “full throttle power” and “full throttle spec. fuel consumption” curves don’t apply to us, since the prop is mounted firmly to the crankshaft and can only make full power at a single RPM (in this case, 2700 RPM).

With a fixed-pitch propeller, the HP we’ll produce varies approximately as the cube of the RPM. Power (and fuel required) falls off quickly with reduced RPM: At 78% RPM we’re making less than half power. Also, with a prop our brake specific fuel consumption (BSFC) (and, consequently, our EFI’s injection requirement per HP) is lowest at about 70% power (2400 RPM) and BSFC goes up about 20% at both highest and lower power levels.

Now, let’s assume the same principles hold true for a small 30 HP V-twin industrial engine. The max RPM is higher (3600), so I’ve just stretched the Lycoming figures out proportionately. For purposes of this analog EFI system, the changes/trends in fuel required (injector pulsewidth) per revolution is what we care about. Here’s how that looks (right column):

As we would expect, since our % power increases more rapidly than RPM, our fuel dose per RPM has to increase. Here’s fuel per rev vs HP:

This required increase in fuel (injector pulsewidth) with power appears to be close to linear. That’s good. If throttle position is directly related to HP, then maybe a simple potentiometer to vary the voltage to Pin 5 of the 555 will do most of what we need to keep the mixture reasonably right at all power levels. Throttle positon and power don't have to be linear for this to work--we can offset the potentiometer with a linkage to the throttle cable to achieve a close-to-right curve.

This approach to fuel dosing could never be adequate in a car, but in a plane with a fixed-pitch prop, maybe it's not too terrible?

Other adjustments needed:
- Barometric pressure (power at all throttle settings varies directly with barometric pressure)
- Manual mixture control (because we know this primitive analog EFI approach won’t be very precise).

So, I think I'd need to find a way to integrate three inputs into the input we provide to Pin 5 on the 555 chip (which sets the injector pulsewidth).

1) Either throttle position (as desicrbed above) or MAP (as described previously)
2) Barometric pressure (the pulsewidth should decrease directly with barometric pressure). This may be less important if MAP can be used)
3) A manual mixture control knob.
Given the leaning circuit (bottom panel, below), any suggestions/hints/snarky comments regarding including/integrating these three inputs to the 555 Pin 5 voltage signal would be appreciated. Thanks,

Mark

 

Given constant RPM, constant throttle position will not equal constant power or constant fuel requirement. As altitude changes that fuel requirement for the same throttle will change greatly.

On the other hand, given constant RPM, constant MAP will be much closer to constant fuel requirement. As altitude changes, the volumetric efficiency of the engine will only change due to back pressure on the exhaust. Meanwhile the average MAP relative to instantaneous MAP error will also change. I believe both of those changes will be relatively small. Even as RPM changes, fuel required per revolution will not change much if MAP remains constant (BSFC .47 to .57 at the extreme).

Overall, part throttle at sea level should deliver the same power and the same fuel requirement as wide open throttle at an altitude that would measure the same MAP minus changes from exhaust back pressure (more fuel at higher altitude due to higher volumetric efficiency / lower BSFC) and average vs instantaneous MAP (also more fuel at higher altitude due to smaller error).

In my mind, your best option is collecting vacuum hoses to one common canister with a MAP sensor and testing to see how close calculated fuel tracks average MAP. Consistency and linearity is what is important not accuracy.

I am an engine enthusiast, I am mildly interested in flight but I am not a pilot. I forget all the stages an engine goes through from takeoff to cruising altitude. I would recommend looking at John Deakin's articles on air fuel mixtures for aircraft engines. They are not directly about EFI but they do go into detail about air fuel ratios in all stages of flight and how they relate to MAP and RPM. Although I seem to recall he is biased towards variable pitch props which probably changes things a bit.

 

Given constant RPM, constant throttle position will not equal constant power or constant fuel requirement. As altitude changes that fuel requirement for the same throttle will change greatly.

Yes, if we've got a naturally aspirated engine, as we gain altitude the reduced air density -> less O2 mass in each "fill -> less fuel required (and less power produced). That's why as we climb we'll either need to manually lean the mixture with a knob in the cockpit (as pilots presently do with a carb-equipped engine), or depend on the MAP sensor to do this, or do it with a separate barometric sensor (if instead of MAP we use throttle position to modify our injector pulsewidth to account for changes in power/fuel per revolution).

Even as RPM changes, fuel required per revolution will not change much if MAP remains constant (BSFC .47 to .57 at the extreme).

Yes, but with a fixed pitch prop, the RPM can't change unless the MAP changes (barring a steep dive, etc). To make the prop turn faster requires more power, and to make more power requires more air, so a more open throttle and a manifold pressure that is higher (until, at WOT, it is darn close to ambient pressure). Also, it's very possible I'm overlooking something important.

Interestingly, as we climb and less O2 is available and our power output declines, the RPM tends to stay very steady. It's a bit of a head-scratcher, but is due to the fact that the prop has less resistance as the air thins, and this occurs at about the same rate as the engine's power reduction.

In my mind, your best option is collecting vacuum hoses to one common canister with a MAP sensor and testing to see how close calculated fuel tracks average MAP. Consistency and linearity is what is important not accuracy.

Thanks, this is good advice. I think the guys on the Speeduino forum have discussed the MAP pulsing/averaging issue, I may learn a lot there. With small-ish hoses/orifices from the runners and a large-ish chamber for the MAP sensor, I'm sure some sort of average MAP can be achieved, then possibly used to modify the injector pulsewidth. It won't be the "effective" manifold pressure (i.e. the pressure/density of the air as it is being inducted), but maybe the average MAP can be used as a proxy.

I still suspect there's something fundamentally faulty with the way that depicted leaning circuit functions. The manual leaning knob, through the op amp, scales the input received by the 555 from the MAP sensor. I've got to think through that, but it seems to me that it would be better to have more direct control of the voltage (and the resultant injector pulsewidth) rather than just tweaking multiples of the MAP sensor voltage.
1) If the MAP sensor fails, having direct manual control of the mixture would be good.
2) If the MAP sensor input is small and I'm adding a lot of "gain" to get the mixture right, then the mixture might change dramatically with only tiny changes in the MAP (slight throttle adjustments, etc). That's not optimum. Similarly, the amount of mixture change produced by a given twist of the mixture knob should remain constant and not be a function of the present MAP (and its subsequent changes).

I'm not sure how to modify the "into Pin 5" circuit so the mixture knob adds/subtracts from the MAP voltage signal, rather than using it as a multiplier. I think what is needed is to retain the existing POT and use it in tuning/setup of the EFI (so the MAP sensor's input has the desired effect on the pulsewidth under expected conditions) and what we need in the cockpit is a separate POT that adds/subtracts voltage to Pin 5.

Different subject: I've been thinking over your suggestion that maybe a simplified Speeduino might be a better approach. It might be. It would be simpler to design in many ways, probably more precise, and more flexible. I'd need to understand the code (and what is underneath it) very well to assure no overlooked problems. Or, mount two of them. But, because of the fairly simple relationships here and the small number of variables, I can't help thinking an analog approach can work, and has advantages of its own. I'm still on the fence, I guess.

Again, thanks for the thoughts.

 

I'm not sure how to modify the "into Pin 5" circuit so the mixture knob adds/subtracts from the MAP voltage signal, rather than using it as a multiplier. I think what is needed is to retain the existing POT and use it in tuning/setup of the EFI (so the MAP sensor's input has the desired effect on the pulsewidth under expected conditions) and what we need in the cockpit is a separate POT that adds/subtracts voltage to Pin 5.

Ahh, it looks like what I need is a "Summing Amplifier." I found the description here.


-- V1 would be the voltage from the leaning circuit as given in the OP (Voltage from the MAP sensor, as scaled through the op amp in that circuit by a separate potentiometer during tuning. This allows us to get the "best fit" rate of change to the injector pulsewidth across expected MAP readings.
-- V2 would would be output from a potentiometer on the instrument panel, a manual mixture control knob to directly change the voltage input to the 555's "Pin 5" and thus the injector pulsewidth. I'll have to read more to figure out if/how I can put in a negative voltage if required, or if I can only use this to increase the voltage.