### Help Support HomeBuiltAirplanes.com:

#### Vigilant1

As I said, How Long before the O2 Sensor would really go Bad on a 2 Stroke, 50hrs, 100hrs, 300hrs, 500hrs? At $20 for a New O2 Sensor if you shop around, I don't see it as a problem. With 4 strokes, there are people who treat the O2 sensor as an expendable item and change it when they change their plugs. Challenges/considerations as I see them: 1) Does this actually eliminate/significantly reduce the chances of failure? Many people report that the expected sensor life is quite variable and unpredictable. 2) The O2 sensor is a thing, and things fail. With some OEM EFI software, it is not easy to know what happens to engine performance when the sensor fails (fails high, fails low, fails null, and when it produces intermittent/erratic inputs). 3) Replacing the O2 sensor is "another thing.” Perhaps my experience is unique, or maybe these sensors and their mounts are made of Inconel or other expensive and durable material, but everything I attach to an exhaust system eventually becomes unitary with all other parts of the exhaust system, and attempts to remove them result in a crumbly, expensive mess and/or a trip to the ER for sutures. In an airplane, the O2 sensor provides little value compared to a well designed open-loop EFI system (or a manual closed-loop using EGT). An O2 sensor (or two) adds considerable costs/hassle. Why have it/them? Last edited: #### Vigilant1 ##### Well-Known Member Lifetime Supporter Valery, thanks for the information. ECUs - my subjective opinion .... 2. Second was Speeduino. Can hangup in any moment and once you will don't have time to reload. Few guys achieve to make better than original boards, but it become commercial projects. The people who know Speeduino well and who are involved with writing the code generally tell me that it is solid and that, while it is now quite complicated, you can just use the parts you need and it will work fine. I don't know enough about Speeduino to have a useful opinion, but my experience with other hardware and software has not been like this. Thanks for your opinion. 4. RusEfi. Use a lot of expensive components and sold without profit. They trying earn money consulting customer for 90USD per hour. In theory, selling the consultation/handholding separately seems very fair--the customers who need it can buy it, others don't. In practice, though, the company has an incentive to keep the customer dependant on more services, and the customer can't easily say "no" if scanty documentation, required system fixes/"upgrades" etc makes it impractical to do things himself. So, the devil is in the details. Most customers (no matter what they think) aren't buying EFI hardware. They really want to buy the capability that these EFI systems promise, and that requires hardware, software, and some support. As long as the customer and the vender are absolutely clear on what is being bought and what will eventually be required, I suppose a lot of different approaches can work. But things work best when everyone's interests are congruent. Last edited: #### blane.c ##### Well-Known Member HBA Supporter I like the O2 idea for tuning and then remove it. Once you or more importantly the computer knows the parameters it isn't necessary. So run some high test unleaded for programming and it should be fine on the shelf till the next time you need it. #### Armilite ##### Well-Known Member With 4 strokes, there are people who treat the O2 sensor as an expendable item and change it when they change their plugs. Challenges/considerations as I see them: 1) Does this actually eliminate/significantly reduce the chances of failure? Many people report that the expected sensor life is quite variable and unpredictable. 2) The O2 sensor is a thing, and things fail. With some OEM EFI software, it is not easy to know what happens to engine performance when the sensor fails (fails high, fails low, fails null, and when it produces intermittent/erratic inputs). 3) Replacing the O2 sensor is "another thing.” Perhaps my experience is unique, or maybe these sensors and their mounts are made of Inconel or other expensive and durable material, but everything I attach to an exhaust system eventually becomes unitary with all other parts of the exhaust system, and attempts to remove them result in a crumbly, expensive mess and/or a trip to the ER for sutures. In an airplane, the O2 sensor provides little value compared to a well designed open-loop EFI system (or a manual closed-loop using EGT). An O2 sensor (or two) adds considerable costs/hassle. Why have it/them? ===================================== It's my understanding the Ecotron EFI setup only uses the O2 Sensor to set it up, then removed. It's been I while since I looked at the Mega Squirt EFI stuff whether they leave it or take it off. Many Cars/Trucks/etc., that use EFI since 1989 are also Heavy Oil Burners and it takes a long time for an O2 Sensor to go bad. Per Kitplane magazine, the Avg Ultralight Flyer on flys 50 hrs a Year. So as I said, How Long before the O2 Sensor would really go Bad on a 2 Stroke, 50hrs, 100hrs, 300hrs, 500hrs? At$20 for (1) New O2 Sensor, if you shop around, I don't see it as a big problem.

Since most 2 Strokes are out in the open on most Planes, and you only have (1) O2 Sensor to remove or replace, I don't see that as a big problem either. You do have to Weld on a Bung for it. They use the Standard GM O2 Sensor.

#### TiPi

##### Well-Known Member
Log Member
The standard O2 sensor is useless for tuning an EFI system unless you want to run stochiometric, which is not what you want in an aircraft. The standard O2 sensor simply tells you if you are lean, stochiometric or rich but not what the actual A/F ratio is.
To tune any system (carburetor or EFI), you need the Wideband O2 sensor AND the processing/display unit.

#### blane.c

##### Well-Known Member
HBA Supporter
I think also it isn't the oil as much the issue with the sensor as it is the lead in aviation leaded fuel. Get rid of the leaded fuel get rid of the issue. Also less plug fouling as a bonus. You know with automotive and industrial engine type spark plugs.

#### Armilite

##### Well-Known Member
The standard O2 sensor is useless for tuning an EFI system unless you want to run stochiometric, which is not what you want in an aircraft. The standard O2 sensor simply tells you if you are lean, stochiometric or rich but not what the actual A/F ratio is.
To tune any system (carburetor or EFI), you need the Wideband O2 sensor AND the processing/display unit.
====================================================

Both Ecotrons and Mega Squirt gives you the Software needed and a USB Port to hook Cable to a Lap/Desk Top or Smart Phone/Tablet to Tuned it. I haven't looked at the other EFI listed.

#### Armilite

##### Well-Known Member
I think also it isn't the oil as much the issue with the sensor as it is the lead in aviation leaded fuel. Get rid of the leaded fuel get rid of the issue. Also less plug fouling as a bonus. You know with automotive and industrial engine type spark plugs.
======================

Since the vast majority of people running 2 Strokes are using Unleaded Pump 91+, then it shouldn't be a problem. If Oil Ratio used isn't a problem then you could use any Brand and Ratio you want. Even if you were to run 100LL, I bet it would be many Hours before the O2 Sensor would really need to be replaced and their low Cost is acceptable to just replace it at X amount of hours yet to be determined. EFI is being used on Sleds, Jet Skies, Bikes, Cars, Trucks, etc., today, so it can be used on Planes also.

I have an Ecotrons Small Engine kit, I just haven't gotten around to playing with it. It came with a 28mm Throttle Body setup for Gas and one 28mm setup for Alcohol. I want a Bigger 36mm Throttle Body.

#### Geraldc

##### Well-Known Member
Microsquirt will do all you need and is fully assembled.
With a megasquirt the quality is up to who built it.

#### stanislavz

##### Well-Known Member
HBA Supporter
I was working with megasquirt ~ 10 years ago or so - but even it is more complicated, than it have to be. For airplane - you are not going to use oxygen sensor or mass air flow sensor anyway. And you are left with throttle and rpm, which is called alpha-N (You need 1ms of opened fuel injector each cycle x tps openning. ). Was implemented in megasquirt itself. But - it is totally fine - we are still running propeller, with fairly narrow power band.

Going next - you are building kind of TBI fuel injector anyway.. Which may run even at fairly low fuel pressure like, 7 psi/ 0.5 Bar.. And to control it - single coil wound on magneto-like structure with 1ms injector opening cycle + throttle opening compensation.. And, yes, you do not even need an microelectronic for it..

Will post some schematics of first car ecu, builded with transistors only.

#### stanislavz

##### Well-Known Member
HBA Supporter
As promised : SW-EM Bosch D-Jet Notes

But it still have accelerator enrichment controll too. Remove it, make it from smd components with some potentiometes (after it to be replaced by solid ones) for pre-tuninig - and it is ok for aircraft.

#### stanislavz

##### Well-Known Member
HBA Supporter
More on Ecu math, being as simple as possible. 4 stroke engine, at each revolution, consums air-fuel mixture. Due to compressability of air fuel mixture - it is not 1:1 as in displacement. As rule of thumb it is 0.8-0.9 of its displacement. Easy example - engines with flat torque curve, like this side valved lf-26
Flat curve shows - for open throttle, we could have fixed amount of fuel, per revolution, or fixed opening time (X) of injector per revolution.

But - due to real environment - air will have different temperature, and we can be not only at sea level, but higher - we need to add two multipliers to our X - temperature and altitude correction. Air temperature could be done via simple sensor, altitude - sensor, or manual.

Last one is throttle - fully closed, it will left 0.2-0.3 Bar, from 1 of absolute pressure. Or ~4psi from 14 psi absolute. And engine will breathe not 1 litre per revolution of 14 psi, but only third of it.

So - we got following equation = for each rpm we have X * (temperature correction) * (altitude correction) * (manifold pressure)

There manifold pressure could be sensed - sluggish control without fuel enrichment, or "dictated" from tps position, which provides kind of enrichment.

And - all this could be done using few ne555 timers, or smallest microelectronic.

#### Vigilant1

##### Well-Known Member
Note--All below was written before stanislavz's Post 212
As promised : SW-EM Bosch D-Jet Notes
(Schematic here)
But it still have accelerator enrichment controll too. Remove it, make it from smd components with some potentiometes (after it to be replaced by solid ones) for pre-tuninig - and it is ok for aircraft.
Thanks. That's sheet 1 of 3, I wonder what else is out there!
As you say, the D-Jetronic did include some accelerator enrichment. It also appears to include some temperature compensation.
+++++++++++++++++++++++++++++++++++++++++
About this idea of an electronic analog EFI. It won't be as precise as an an EFI running a program (which can take into account more inputs, can refer to a detailed fuel map, etc). It also won't be as easily tuned, requiring hardware or at least changes in potentiometer/variable condenser changes with different installations. It is also likely to require some manual fine-tuning in flight (e.g. mixture adjustments, as we now do when flying a carburetor). On the "plus" side, there's no program to "crash," no re-boot time, no complex code to understand, and fewer components to buy/fail. Some people might find a simple analaoge EFI setup to be most suitable as a backup to a more "full featured" fully digital EFI system.

I wonder if we couldn't get almost all the EFI info we must have >>for fixed-pitch prop airplane use<< from RPM. Our needs are much different from a car. The gears (and neutral) of a car transmision can allow it to be at any RPM with 10% fuel flow. In an airplane with the crankshaft firmly attached to a fixed-pitch prop, the HP needed to turn the prop is tied much more tightly to the RPM. The RPM to HP relationship is not a direct one, though, with HP varying by the cube of RPM (per the "propeller load equation):
HP = K x (RPM^3)
If we use the formula alone (and make some simplifying assumptions e.g. that volumetric efficiency and BSFC are constant, etc), then we end up with the following fuel requirements for an engine with max rated power at 3600 RPM:

 RPM​ % Max HP (and GPH)​ 3600​ 100%​ 3500​ 92%​ 3400​ 84%​ 3300​ 77%​ 3200​ 70%​ 3100​ 64%​ 3000​ 58%​ 2900​ 52%​ 2800​ 47%​ 2700​ 42%​ 2600​ 38%​ 2500​ 33%​ 2400​ 30%​ 2300​ 26%​ 2200​ 23%​ 2100​ 20%​ 2000​ 17%​ 1900​ 15%​ 1800​ 13%​ 1700​ 11%​ 1600​ 9%​ 1500​ 7%​ 1400​ 6%​ 1300​ 5%​ 1200​ 4%​ 1100​ 3%​ 1000​ 2%​

The analog "no program" system wouldn't use a fuel map, just a "best fit" adjustment to have the injectors squirt in the right amount of fuel for each RPM.

So, what would we need to correct with this approach?
1) The fuel flow at very low power levels is obviously too low. The propeller power equation only tells us how much power it takes to turn the prop, and at low power settings a disproportionately large amount of fuel is used to do non-useful work. Even on a per-rev basis, we'd need to increase fuel flow at these lower RPMs--we could determine how much with field tests (using a wideband O2 sensor, etc). The good news is that, in actual flight, we'll probably be between 50% power and 100% power almost all the time, and the fuel calcs above likely aren't too far off. Better yet, the published Lycoming and Cont. fuel flows at reduced RPMs/partial throttle ops with FP props would probably be more useful and accurate than fuel flows derived from the propeller load equation.
2) Altitude compensation. With increasing altitude, it takes less HP to turn the prop a given RPM in the thinner air. Our engine will automatically compensate to some degree (an equal volume but lower mass of air will be drawn in at the same RPM as altitude increases), but our mixture will become increasingly rich if we keep injecting the same mass of fuel per RPM. So, we either need to manually adjust (with a mixture knob), or we can get fancy with another circuit using a barometric sensor. We could put that in the intake manifold (where it will sense both changes in the air charge due to altitude and also those due to vacuum from throttle setting), but putting the sensor in the manifold will lead to big differences in reading depending on induction phase. If we run a separate system (and MAP sensor) per cylinder (handy with a V-twin), it will be at/near ambient pressure for about 75% of the time, then decrease (maybe markedly, for partial throttle ops) during the intake phase. Without a camshaft sensor, there's no simple way to make the reading occur during the induction cycle.
3) There's no automatic enrichment for starting and cold engine warmup. Probably need at least a prime button, maybe some sort of electric choke with a temp sensor.
4) There's no automatic adjustment when we move the throttle--our fuel only changes when the RPM changes. If we run in the "best power" mixture area (which is richer than stoichiometric), then the RPM will automatically increase with larger throttle opening, and it will also eventually decrease with throttle closure.
5) Prop load due to flight conditions: During descent, potential energy of the plane is traded for kinetic energy and RPM increases with no concommitant need for additional fuel. So, our RPM-derived fuel squirts will result in a rich condition. Similarly, if cruise speed is used to determine our fuel requirement per RPM, then during TO and climb the mixture willl be lean (because the lower airspeed results in a prop experiencing higher AoA=more load per the same RPM). For these reasons, it might be best to do our calibration at climb speed (so we'll be at best power mixture there) and then lean when we reach cruise (as with a carb).

Bottom line: It would take a little experimentation, but I suspect an electronic analog "RPM with manual adjustment" system might be sufficient for some flight uses. Or, integrate some automatic trimming using external barometric pressure or MAP.

The previously mentioned "555 analog EFI system" could be a start. It uses a crank sensor (to determine RPM and to trigger the injection) and also includes a MAP sensor. Instead of a crank sensor, an induction loop on a spark plug wire or coil might also work (realizing that then an ignition failure o that wire results in loss of fuel flow). More explanation and schematics are at the link above. Probably about $10-$15 in parts per cylinder (plus injector and, if desired, a MAF or barometric sensor). At that price, a separate board/circuit for each cylinder seems fine. I don't know if this has ever been flown, so a lot of work would be needed to make it reliable, attend to the EMI/noise/power filtering issues, get the slope of the injector open time per RPM right, etc.
As an alternative, as above, use the RPM as the primary sensor but modify the injector opening signal using the TPS (rather than MAP). You'd lose altitude compensation, but maybe gain some simplicity.

Main schematic of the 555 EFI from Paul Lamar's Rotary Engine site, plus ancillary schematics:

#### Attachments

• 174.1 KB Views: 1
Last edited:

#### Vigilant1

##### Well-Known Member
View attachment 98276
Flat curve shows - for open throttle, we could have fixed amount of fuel, per revolution, or fixed opening time (X) of injector per revolution.
I'm not an expert, but I don't think it will work like this on an airplane, at least one with a fixed-pitch prop. The dyno torque curve shows us the maximum torque (and power) an engine can make at each RPM. To build the dyno chart, a variable load is needed so that at each RPM the engine's maximum torque is determined. But, in our airplane with a fixed pitch prop, we can optimize that prop only for a single RPM. Below that RPM, the throttle will be less than fully open and the engine won't be producing all the torque (and power) that it is capable of. So, below that RPM, in the real airplane the fuel required will be lower than would be calculated using the dyno chart. And, at least in level flight and at our calibration speed, the throttle will be partially closed.
Per the propeller load equation, if we go from 3600 RPM (100%
hp) to 3100 rpm (a 14% reduction in injection events), our HP (and presumably fuel flow) drops by 36%. It looks like a fixed amount of fuel per revolution won't work quite right, we'd be better off with some sort of scaling (using an opamp?)
The above is a bit (over?) simplified, as prop load changes with airspeed as well as RPM.

Last edited:

#### Vigilant1

##### Well-Known Member
More on the above topic: Relationship of RPM to required fuel per revolution with a fixed-pitch prop.
The graph below is from Lycoming, it is for the O-360 engine:

Note how the curves for the propeller loaded engine differ from the full throttle curves. Also, for the prop loaded engine, BSFC gets worse when we reduce RPM below 80% (= power below about 55%).

I made the chart below based on the info in the graph, then added some analogous figures for the Briggs and Stratton 810cc engine under the assumption (?) that it will have similar RPM-to-HP and HP-to-BSFC relationships as the Lycoming.

 O-360 RPM​ RPM %​ O-360 HP (Prop Load)​ HP Percent​ O-360 BSFC (Prop Load, Full Rich)​ V-twin RPM​ B&S 810cc HP (est)​ B&S 810cc GPH (at O-360 BSFC) ​ B&S 810cc fuel per rev (ml)​ 2700​ 100.0%​ 180​ 100.0%​ 0.55​ 3600​ 30.0​ 2.8​ 0.048​ 2600​ 96.3%​ 161​ 89.4%​ 0.51​ 3467​ 26.8​ 2.3​ 0.042​ 2500​ 92.6%​ 144​ 80.0%​ 0.48​ 3333​ 24.0​ 1.9​ 0.036​ 2400​ 88.9%​ 128​ 71.1%​ 0.47​ 3200​ 21.3​ 1.7​ 0.033​ 2300​ 85.2%​ 113​ 62.8%​ 0.47​ 3067​ 18.8​ 1.5​ 0.030​ 2200​ 81.5%​ 100​ 55.6%​ 0.47​ 2933​ 16.7​ 1.3​ 0.028​ 2100​ 77.8%​ 88​ 48.9%​ 0.48​ 2800​ 14.7​ 1.2​ 0.026​ 2000​ 74.1%​ 73​ 40.6%​ 0.48​ 2667​ 12.2​ 1.0​ 0.023​ 1900​ 70.4%​ 62​ 34.4%​ 0.53​ 2533​ 10.3​ 0.9​ 0.023​ 1800​ 66.7%​ 50​ 27.8%​ 0.57​ 2400​ 8.3​ 0.8​ 0.021​

For the O-360 with a fixed pitch prop (and maybe for other air-cooled pushrod engines like the B&S's), as we move from half power (77% RPM) to full power, the amount of fuel required per cycle approximately doubles. So, an RPM-based EFI system would benefit from having some type of scaling to increase the fuel per injection event in this area. Could be based on MAP, maybe based on throttle position, or maybe (in the very simplest case) even just a hard-wired (or coded) enrichment based only on RPM. Below about 50% power, the amount of fuel per rev seems to be more constant, though I couldn't find very low RPM fuel burn numbers for the Lycoming. Maybe a fixed injector pulsewidth per rev would work okay at these lower power settings.

#### Attachments

• 557.8 KB Views: 12
Last edited:

#### TiPi

##### Well-Known Member
Log Member
Nice work Mark
This would be the baseline fuel map, then correction factors for DA and IAT would be needed for automatic fine-tuning and engine temp for the warm-up phase. Starting can be done with a few extra squirts while cranking.
Just don't add a variable pitch prop.

#### stanislavz

##### Well-Known Member
HBA Supporter
All agree an above chart - but you miss throttle/intake absolute pressure - on 3600 it is 100% opened, on 2800 you have 0.6 abs pressure my guestimate. Because poweris halved, but rpm is not, only 27% less.

#### TiPi

##### Well-Known Member
Log Member
Throttle position and/or MAP is not needed as primary signal in this case. The rpm achieved is the result of the power generated which directly links into the fuel map. There is a very strong relationship between rpm and power, good enough for the baseline mixture. TPS and MAP would just confirm the running parameters at each rpm.

#### Vigilant1

##### Well-Known Member
This would be the baseline fuel map, then correction factors for DA and IAT would be needed for automatic fine-tuning and engine temp for the warm-up phase. Starting can be done with a few extra squirts while cranking.
Just don't add a variable pitch prop.
For a "no software" analog EFI (like the one from the old Rotary Engine site), in principle you could set up a simple N-alpha like this:
1) Set the baseline injection dose as something fairly rich (to give, say, .05 ml per rev per the above chart, = .025 ml per cylinder).
2) De-enrich from the baseline dosing with three in-series (so, additive) resistance inputs:
A) a barometric sensor that decreases signal voltage with decreasing atmospheric pressure. So, as we climb, fuel dose is reduced across all RPMs.
B) a potentiometer on the throttle (anywhere: at the throttle body/carb, at the cockpit throttle knob) that reduces the baseline signal voltage as the throttle moves down from WOT. The rate and throttle opening range of the pot actuation could be adjusted with changes in the length and eccentricity of the arm on the pot, and with stops. Primative. Maybe there's a good electronic way.
C) A manual mixture control knob on the panel.
If a MAP sensor can be made to work (i.e. give us a true indication of the charge density going into the cylinder, which requires that we take the reading at the right point in the cycle), then that single sensor would perform functions A and B above, and we'd have a primative speed density system (with no fuel map, just an adjustment of some kind) rather than an n-alpha rule system.

A " limp home" mode on the same board could use an independent RPM sensor (inductive pickup on a spark cable?) and a default "good enough" injector pulsewidth. Other adjustments (mixture knob?) only if found to be necessary to make climb power.

Some variant of the above might be an okay "stone knives and bear skins" approach to EFI. The hard part would be making the adjustments work at the right ranges and rates, and in making it reliable. It will never be as hands-off and fuel efficient as a modern speed density system with more inputs and a real fuel map. But, it might be good enough, cheap enough, and reliable enough to serve as a backup, or a primary + backup if fiddling with knobs in flight is acceptable.

Last edited: