Air Charge: Bosch's Secret Snail Sauce

People coming from older generation Hondas get thrown for a loop when they hear the Type R doesn't need a tune for most parts, but that it doesn't really make more power without one either. It runs all the same.. but how? How can you modify the airflow characteristics of an engine and not see significantly different results? It's thanks to our physically modeled or torque targeting ECU and the 'load' measurement being Bosch's air charge model. Other modern Honda’s with a Keihin ECU use a similar mechanism, but we’re just covering the FK8’s ECU in this article.

Instead of relying on static lookup tables primarily based on one main sensor's info like in the mass majority of ECUs, the Bosch ECU is able to precisely determine the exact mass of air in the cylinders and what air mass it should target in order to make a specified amount of torque, regardless of most modifications or environmental conditions.

An ECU's primary job is injecting a precise mass of fuel with a measured mass of air flowing into the engine in order to achieve the most efficient combustion, both for power output and emissions purposes. ECUs were first known as 'electronic fuel injection' computers, and to this day that is still their primary task.

There's two main ways other ECUs are tuned for this job: by measuring the airflow with a MAF sensor (also known as a MAF tune), or by measuring the manifold pressure (also known as a speed-density or MAP tune). Each method has its own pros and cons, each basing the majority of the table lookups off one particular metric and compensating with additional sensors via tables set up by the OEM calibrators or your tuner.

MAF and MAP model

With a MAF tune, a mass airflow (MAF) sensor can approximate the mass of air flowing through your intake at atmospheric pressure by using a simple table of estimated grams per minute, also known as the AFM table. This works very well for a naturally aspirated engine, but once a turbo starts compressing the air the actual mass and temperature of air after the turbo isn't equal to that measurement. There's compensation tables involving the manifold pressure sensor to make a turbo possible on a MAF based tune (assuming it supports a positive pressure MAP sensor, which many Hondas do not thus making turbocharging impractical if not impossible on an OEM ECU) but the tune is still largely based on airflow.

With a speed density tune, a manifold pressure (MAP) and intake air temperature sensor tells the ECU the density of the air in the intake manifold before the intake valves and allows for much more flexible setups than a MAF sensor, but density alone cannot tell you the mass of air inside of a space. You can't mix the proper mass of fuel by knowing solely the density of that air except by tuning it for a particular intake/exhaust setup (including intercooler, turbo, and anything else between the intake and exhaust tips). Given that constant, you can use density divided by engine speed as a metric for driving the engine as the mass flowing into the intake manifold should be almost the same from pull to pull despite not having an actual measurement of the mass. The problem is that swapping out your parts can heavily throw off the tune.

Air charge model

The Bosch ECU is neither a MAF tune nor a MAP tune, rather something very different and relatively new. It uses a combination of all these sensors along with some very clever time-domain, physics-based mathematics in order to more accurately control the engine than ever before seen in a Honda. Rather than basing it off one main sensor, the Bosch ECU combines data from the MAF, MAP, BP, barometric, IAT1, IAT2, and RPM sensors to calculate a single, precise metric that almost every other table or calculation uses: air charge.

Air charge is a percentage, representing the mass of air in the cylinders versus the mass those cylinders would hold at atmospheric pressure. At 50% aircharge, the cylinders are under vacuum. At 100% aircharge, the cylinder is not under vacuum nor under pressure. At 200% aircharge, the cylinder has twice the normal amount of air inside of it and the engine has an effective displacement of a 4.0L engine.

While the exact algorithm is proprietary and not known, after years of theorizing, researching, and reading tidbits of internal documents we think we have a pretty good idea how it works.

  1. The MAF sensor measures how much uncompressed air is flowing by it, corrected by temperature (IAT2) and atmospheric pressure (BARO). This flow data is tracked over time.
  2. The IAT1 sensor located after the throttle plate determines the temperature of the air inside the charge pipe and intake manifold. The temperature should effectively be the same between the BP sensor and MAP sensor so there is no need for three air temperature sensors.
  3. Using the ideal gas law (pressure + temperature = density) the temperature of the IAT1 sensor is combined with the pressure of the MAP and BP sensors to determine the density of the air that is building up before the throttle plate (BP) and before the intake valves (MAP). This density is also tracked over time to determine a rate of change.
  4. By tracking both the uncompressed mass airflow and the compressed density resulting from that flow, the ECU can build up a mathematical function to convert pressure into mass in the intake manifold.
  5. As the intake valves open, the ECU measures the resulting drop of manifold pressure divided by engine speed. Since the ECU can now convert manifold pressure to air mass, it is able to determine the precise air mass flowing into the cylinder at any given point in time. The effects of VTEC, cam angles, misfires, exhaust backpressure and more are automatically integrated into the calculations as a result. That is something that is not possible with other engine control approaches.
  6. This mass is finally converted into a percentage based on an internal model of the engine.

Changing most parts around doesn't hurt the air charge calculation because it's all being dynamically calculated anyways. It abstracts away the need for compensation and manual tuning of various parameters. The overall intake path and exhaust path largely doesn't matter because the ECU is automatically figuring out the only thing that really matters in an engine: how much air is in the cylinders. By knowing this, the ECU can calculate almost all the fueling parameters on its own, thereby accomplishing the primary task of the ECU automatically. This greatly simplifies the task of the OEM calibrators and by extension aftermarket tuners, allowing environmental conditions and airflow configurations to largely be a non-concern. The Bosch ECU was designed as a one-size-fits-all ECU for OEMs and luckily that directly benefits the aftermarket as well. Everything from ignition to cam angles, injector phasing to lambda targets, are looked up by aircharge.

However, the air charge model also has a downside: it requires perfect calibration of the sensors and absolutely no leaks between the MAF sensor and cylinders. Any errors are exponentially worse than on a MAF or MAP tune due to the way the sensors are so intertwined and how the data is integrated over time.

Air charge is a sensitive algorithm

Even the tiniest boost/vacuum leak or miscalibration/maxing out of the MAF sensor will cause the air charge calculation to be wildly inaccurate. It is easy to see in datalogs - air charge will be jumping all over the place and sometimes reaching numbers that shouldn't even be possible.

On a recent tune we did for a customer's FK8, everything was as smooth as possible and looking great.. until they replaced their inlet pipe. It wasn't that the inlet pipe itself caused a problem, rather that their silicone intake pipe didn't create a perfect seal around the inlet pipe which introduced a tiny vacuum leak after the MAF sensor.

This resulted in the turbocharger being able to suck in and compress air that the ECU didn't track the flow for, which threw off the entire algorithm and resulted in a significantly higher and unstable air charge calculation - so high, matter of fact, that it wouldn't have even been possible on the stock turbocharger. The default limit of 250% air charge was being reached and power was reduced, all from a teensy tiny vacuum leak.

 

As you can see, the air fuel ratio and short term fuel trim was going wild too. This is because fueling (and nearly everything else) is based on air charge - if the air charge is higher than reality, the ECU will inject too much fuel until the closed loop system kicks in and starts modifying the injector duration to reach the commanded lambda. Despite manifold pressure and MAF voltage remaining smooth, the calculated air charge was all over the place. Nearly every single pull would do this, with the exhaust going rich and then lambda starting to bounce all over the place while the closed loop system attempted to correct for it.

After reseating the hose and clamping it down a bit better, he sent us new logs and the issue was gone completely, back to making very smooth power that the ECU isn't constantly compensating for.

And yet the customer had no idea - it felt perfectly fine to him. This highlights a big problem with the Type R. We hear it all the time: 'my car is running fine', and then we review logs for them and it's anything but. You may think it's running fine, it may feel fine, but in reality, it could be going absolutely haywire.

This is why it is so critical to learn how to read datalogs and start actually logging every once in a while, especially after making modifications. Even with no CEL or no noticeable power difference, your engine could actually be having issues that greatly lowers the reliability and safety.

We're probably going to rustle some jimmies here, but all of this is also why a blow-off valve that vents to atmosphere is a terrible idea on this platform. Even a tiny leak causes issues, sometimes major, sometimes unnoticeable - but there is no debate that it messes with the core algorithm of this ECU. A VTA BOV is inherently a boost leak. There are some people trying to say that a BOV is fine, but they don't understand this algorithm or weren't even testing a Type R. On a MAF based tune it's bad enough, but it's even worse with Bosch's air charge algorithm.

Torque targeting

So, the Bosch ECU can figure out how much air is inside the cylinder.. but how does it figure out how much should be inside the cylinder?

Using a mathematical model of the engine, the ECU can estimate a desired mass of air that should be able to theoretically generate the requested amount of torque. More fuel is not what makes more power, rather it is the amount of air in the cylinders that determines potential power output. Having such an accurate measurement of air mass in the cylinders enables the Bosch ECU to use torque targeting.

Everything begins with the torque request map, one for each gear and drive mode for a total of 18 maps. RPM and throttle position correspond to a requested amount of torque in newton meters. From this, the ECU calculates a target air charge (which we cannot see in datalogs yet) up to the limits defined in the air charge limit table, and from that it determines the necessary boost pressure and wastegate command to reach that torque under the current environmental conditions. The ECU estimates the actual torque output of the engine as it builds boost, and modulates the ignition advance to achieve it.

We do not specify a particular air charge, PSI or turbo ratio when tuning a Type R, we can only set safety limits for them. Even ignition advance is not directly controllable.

Many tables in this ECU are requests and limits rather than final values, especially when it comes to ignition advance. Having better fuel like ethanol simply allows you to extract more power out of the air by being able to advance the ignition timing further, but the ECU will not continually increase ignition unless it actually needs it.

We hope this article will help people understand just how special their Type R is, especially other tuners that are wanting to better understand what's going on.

For further reading on how this and other functions of the Bosch ECU works, refer to our article on our FK8 Wiki.

If you'd like to have your FK8 running better than the rest, you can purchase a custom remote tune here.