Saving Fuel With Advanced Sensors And An Arduino

When [Robot Cantina] isn’t busy tweaking the 420cc Big Block engine in their Honda Insight, they’re probably working on some other completely far out automotive atrocity. In the video below the break, you’ll see them take the concept of a ‘lean burn’ system from the Insight and graft hack it into their 1997 Saturn coupe.

What’s a lean burn system? Simply put, it tricks the car into burning less fuel when it’s cruising under a light load to improve the vehicle’s average mileage. The Saturn’s electronics aren’t sophisticated enough to implement a lean burn system simply, and so [Robot Cantina] did what any of us might have done: hacked it in with an Arduino.

The video does a wonderful job going into the details, but essentially by using an oxygen sensor with finer resolution (wide-band) and then outputting the appropriate narrow band signal to the ECU, [Robot Cantina] can fine tune the air/fuel ratio with nothing more than a potentiometer, and the car’s ECU is none the wiser. What were the results? Well… they weren’t as expected, which means more experimentation, more parts, and hopefully, more videos. We love seeing the scientific method put to fun use!

People are ever in the quest to try interesting new (and sometimes old) ideas, such as this hot rod hacked to run with a lawnmower carburetor.

37 thoughts on “Saving Fuel With Advanced Sensors And An Arduino

    1. If you don’t control for temperature and detonation, which is what happens when the fuel map tuning is all over the place. You’ll burn the valves or crack something from engine knocking – but if you don’t let it go that far then the only problem you’ll have is potentially increasing NOx emissions.

      1. There isn’t really any potential about increased NOx emissions. Doing this is basically guaranteed to increase them. Those are the ones that are damaging to human health and contribute to acid rain.

        1. Not quite. NOx starts to go down with more oxygen. It peaks at a typical AF ratio of 16 and then starts to go down with leaner and leaner mixtures – if you can pull that off.

          1. Basically, adding excess air works a bit like EGR where you’re adding some “inert” gas into the cylinder to absorb the heat. The combustion temperature goes down and NOx generation is suppressed. However, HC emissions also increase because of incomplete combustion at the very lean end and you start to get misfiring.

    2. You can massage your engine a bit to make it more favorable. Like radius everything to avoid hotpots, make more of a wedge entry into the quench area to slow down the flame front. Ideally you want to take a stratified charge approach where you get a pocket of near stochio mixture and a pocket of air or exhaust gas as a working fluid. Possibly you can achieve this by meddling with exhaust valve timing and/or replumbing a 4 valve head to only use one intake valve for mixture and either allowing air or exhaust gas bleed in the other. It’s all fluid dynamics in your head though figuring the best approach. What might be interesting is sticking a high bleed down lifter in one intake valve, so it has assymetric opening at low load. But ideally all this should be sorted out on a flow bench before committing it to your daily driver.

      Injector timing hacks could be tried though, but it probably needs a complete rebuild of the injector drivers which are usually paired so you’ve actually got injectors squirting at closed valves. So your duty cycle relies on using the time the cylinder isn’t firing to meter some of the fuel in. Ergo, on single timed drivers, where you’d try and squirt all the fuel in in the last half of the intake valve cycle, so you’ve got your stochio pocket in last (and hopefully somewhere near the plug) you’d need to double up on pressure or injector size.

      Anyway, if that’s all greek, best just stick with keeping your plugs clean, tires pumped, oil changed on time with synthetic.

    3. Power production decreases dramatically with a leaner mixture first. Conventional ICE’s (internal Combustion Engines) for those into acronyms, run best in the range of ~12:1 air/fuel. the lower #’s equate to higher power up to a point and lower #’s obviously give better econ. with less power, again up to a point. The down side is with higher speeds, higher power is required along with increasing weight, drag, rolling resistance and inclines, opposite for declines, less power and all that goes with that. N2O is a result of combustion pressure and temperature along with the chemical make up of the chamber gasses. EGR introduces burnt residual gasses into the chamber to slow combustion, lower temperatures and reduce N2O’s which is confusing because we think isn’t exhaust hot already which would make the air temp going into the chamber increase in temp and to a point it does. Cold air charges are denser than hot charges which means more O2 available for supporting combustion which should mean increased power and it does, it also means the compression must raise the air temp to a point that makes the fresh charge easier to ignite! This is problem where fuel octance comes in. Lower octane will ignite at lower temps and pressures which is major factor in producing power, determining the timing of the ignition and controlling the combustion. The entire dynamics of internal combustion engines and all the factors of designing, tuning them is so complex, few people can undertake a project like this and successfully complete it without a multiple exhaust gas sensor, a dyno, and a high likely hood of having a masters degree in automotive engineering. Look how long it took the manufacturers to get where they are now from the days of the Model A’s. How much engineering time was devoted to all that progress. Granted a lot of that time was spent idling along not putting design into more efficient units but simply getting a product they could sell to an unwitted market that had little concern about the environment or resource reserve. I can say this about engine design and a few other things, we should have been a lot farther ahead by now. We all know the big oil companies had a major hand in everything, keeping things as they were. They’re still trying. As they say, it’s all about the money honey.

    1. He is actually using a 420cc air-cooled engine from Harbor Freight. He’s added a supercharger and has gotten the car up to 70mph on that little engine. He got it up to 50mph with the 220cc engine. He uses the term “big block” as a joke, as it’s a lot bigger than the 220cc engine. The series is very interesting.

  1. products like the Apex’i SAFC(and the SAFC II, AFC NEO) have been doing this since the 90’s. It’s arguably cheaper and usually only a 6-8 wire splice. Though aimed at the other direction (fine tuning or running richer), it works by intercepting the upstream o2 sensor and altering the signal that the ECU receives based on a user define map either in low load or high load mode. You could have a lean low load for MPG cruising but still have ample acceleration when you need to accelerate rapidly without hurting your engine.

  2. I’ve seen that tried before, usually with poor results. The ECU often interprets the signal as a failing sensor instead. Tweaking one sensor is a poor substitute for real tuning. You could fake a lower Map sensor reading, but that also adds ignition timing. If you’re trying to monkey with one variable and want less fuel, I would try reducing fuel pressure.

      1. This is something that may work at part throttle, but is a bit iffy. Whether it works depends on the ECU’s internal logic. And I don’t know enough about OBD2 Saturn specifics to know if any of these are in play, but here are some features that can prevent this from working:

        1. The spoofed O2 signal may fail to meet some criterion the ECU uses to detect sensor failure, typically response time or heater current.

        2. The ECU may limit how much it can change fueling in response to the O2 input.

        3. If the ECU is already targeting something other than 14.7:1, it may ignore the O2 sensor. Less likely to happen at cruise; this happens more often when someone tries to add more full throttle fuel.

        4. The ECU may compare several sensors and treat the O2 sensor as the odd one out.

        I can’t say for sure if he will run into any of these. But it’s more certain to modify the ECU outputs (unfortunately, it’s easier for a piggyback unit to add fuel than remove it), change something the ECU doesn’t control (I believe fuel pressure control on these cars is purely mechanical), or change the data in the ECU. Outright replacing the ECU would also work, but may not be street legal.

        One other thing to consider: water injection can help reduce combustion chamber temperature and oxide of nitrogen issues that often happen with lean burn.

    1. I’d go the other way, smallest 4 hole injector you can fit with highest fuel pressure. Lower just makes your atomisation nastier, carbs up your intakes and introduces potential percolation problems in hot weather.

      1. I guess percolation is more a carb thing whereas vapor lock would be more the right name for the issue in injectors, anyway, gaseous phase stuff in liquid handling systems don’t not work no good.

        Anyhoo, reason we ended up with fuel pressures of 40 or so PSI typical in 90s up cars was they tried 27psi, had problems, they tried 35ish, had problems.. 40+ golden. But some went higher, 55 or 60 PSI and AFPRs are available. Need your maff brain to work out if you’re delivering same fuel at lower cc and what actual PSI your injectors are specced at. 20lb/hr at 40 psi is a bigger injector than 20lb/hr at 55psi.

    2. The right sensor would be coolant or air temperature as those result in a fairly direct change in fuelling without any associated changes in timing as they’re not related to engine load etc.

      Or he could just throw a Speeduino on it and control everything properly with no guessing.

  3. I did a similar thing with a Toyota.. But no additional computer needed.. I did a Negative Voltage Offset to the Ground side of the Existing Narrow Band O2 Sensors of about .05 Volts.. It was on a Pot fed from a Battery. I was able to keep the Existing ECU in Closed Loop up to about 18:1 AFR while driving ‘Normally’.. It was just an experiment to see if I could do it.. I did not have any EGT’s to check to Engine Exhaust, or the Cat Temps, so I took it out..

    Because I could..


      1. I did not see an improvement in Fuel Millage with just messing with AFR. I switched over and started fooling with Ignition advance and got improvement I was after.. The millage went from 32 to 36 MPG. So I did not mess with AFR after that..

    1. When I was a kid we did something similar.

      We had a tow truck coming for a junker (worthless 4 banger). We got a keg, put a brick on the gas pedal and had an engine blowing party.

      It was epic. Never been so impressed by a MOPAR. It finally siezed after an oil fire then cooling hose failure (that put out the oil fire).

  4. I have a 5-speed manual and CVT Insight. The 5-speed manual manages a ULEV designation while the CVT ekes out SULEV. Yes, NOx output goes way up in lean burn conditions which go way beyond stoichiometric 14:1 A/F ratio to around 25:1. The Insight has hard valves, valve seats and platinum spark plugs which are designed for the harsh conditions. The other piece of the puzzle is the use of special (as in also very expensive) adsorption catalytic converters which can store up NOx in lean burn and then deal with it under more favorable conditions.

    The Insight’s engine control computer isn’t super smart by modern standards, designed with mid 1990s technology. The DLC bus is pre-CAN ODB-II, J1850 at least in the US models. The hybrid system is parallel mode “assist” where the engine is always running when the car is in motion. The integrated motor assist (IMA) motor-generator is a three-phase permanent magnet type capable of about 7.5 KW and is connected to a 6 AH 144 Volt NiMH battery pack of 120 EV rated D-cells. The maximum discharge to the assist motor is about 50 Amps and the recharge rate is around 25 Amps (the meter in the car has asymmetrical calibration on the two sides to make it look like regenerative braking provides as much power as the acceleration assist uses. A 12 Volt conventional starter is provided for starting the car in temperatures below 10 degrees F. or if the hybrid battery is somehow totally depleted. It is in fact possible to operate an Insight without the hybrid battery if the feedback wire from the inverter is open, allowing the 12 Volt battery to charge normally. The hybrid battery condition monitor (BCM) and inverter talk to the instrument panel and ECU via a 9600 bit/S serial data link. The most common means of temporarily putting an Insight with a dead hybrid battery back into service is to switch off the main circuit breaker and simply unplug all connections to the BCM which allows the inverter to properly handle the 12 Volt battery as noted above.

    Another interesting fact is that Honda apparently had a marginal trust level of CPU chips in the 1990s. The engine control units of the Insight and several other conventional gasoline Honda models are capable of running the engine in limp-home mode without a working CPU chip, using just the combinational logic in the controller.

  5. I’m pretty sure [Robot Cantina] knows what he doing with his cars. But for the rest of us, especially if you have a more modern car…

    Messing around with the air/fuel ratio etc. outside the OEM Engine Control Unit’s (ECU’s) programming is asking for trouble. Running too lean or rich trying to save fuel or increase power in a modern engine unless you have total control over the ECU software and all the real-time sensor data and you know exactly what you are doing is a very bad idea.[1][2] Leave that witchcraft to professional automotive engineers.

    There are a lot of shady black boxes on the internet that claim to increase mileage or power by messing with the mixture and/or timing and/or injecting water.[3] Most (but not all) of them at best don’t work or at worst are harmful. Install one of those things in a modern car and ignore your forever-lit red check-engine-light, good luck ever passing an emissions test again, and that strange smell? Yeah that’s your toasted red-hot catalytic converter.

    One story you may hear is that it is OK to tweak the automobile engine’s fuel-air mixture any time, after-all pilots do it all the time with piston aircraft engines. That is true in-part, a common technique is to manually adjust the aircraft engine’s mixture using the exhaust-gas-temperature (EGT) and cylinder-head-temperature (CHT) readings as guidance. But that is in an airplane. Your automobile is NOT an airplane!

    An airplane engine usually sees a fairly consistent and slowly varying load. But an airplane engine will be subject to widely variable but slowly changing air temperatures/density and humidity values. Under those conditions manually tuning the mixture is justified. An automobile on the other hand sees rapidly changing widely variable loads plus somewhat variable air temperatures/density and humidity values. Under those conditions it is best to let the OEM ECU do in real-time what it was programmed to do at the factory. That way your check-engine-light (if lit) will still really mean something, and you will not damage the engine or catalytic converter.

    * References:

    1. Lean vs. Rich Fuel Mixture (What’s the Difference?)

    2. Signs Of Your Engine Running Lean

    3. Fuel Savers

  6. GM actually has a lean cruise algorithm built into the computers used in the TBI pickups and it can be activated by changing a bit in the calibration chip. It was never turned on due to the need for the storage catalyst required to pas US emission standards, but they had it available if they deemed it the better way to meet their CAFE numbers. I have heard of people turning it on when used to convert a carb equipped car to TBI.

    1. Its YT, and a a video series at that. Its not meant to accomplish a goal, but to learn step by step and educate/entertain. Im sure they will try standalone ecu at some point.

    1. It’s not too bad over a several full tank average if you try to use the same pump… worst case is when ppl do a 20 mile loop and “only got 0.1 gallon back in” because yeah, gas gets warmed up as you drive, either with the pump being stalled because it’s returnless, or with the gas getting warmed up in the engine bay and coming back in the return line. Thus expands a bit. Add that to the pump cutoff variance and you get ppl claiming 200mpg from sticking a fan blade in their intake.

  7. I have a megasquirt FI on my 1964 VW bus, 1776cc engine. I lean burn approx 16:1 at hiway cruise 60mph. Ignition advance is 40 degrees. Mileage is about 28mpg. I have CHT sensors to avoid overheating the heads. So far not a problem. 28mpg is unheard of in a VW bus.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.