Maybe this has been posted before but I haven't seen it. This is one of the clearest explanations I have come upon so I thought it would be appreciated. The whole thread is here, https://www.m5board.com/vbulletin/e39-m5-e52-z8-discussion/531905-alpha-n-explained.html
But here is the meat and potatoes,
Many posts on this forum discuss what what Alpha-N is. Some accurate, some not. It is common on S62’s so I think its important to discuss and that accurate information is widely available. I will try my best to stick to strict terminology here.
I’ll first talk about Alpha-N and then talk about how it relates to S62’s (to the extent of my knowledge of S62’s, because Alpha-N and how its actually working on the S62 is more nuanced). I’ll use real world examples also.
Before I start discussing what Alpha-N is, let me explain what Alpha-N isn’t.
Alpha-N by itself is not a “tune” and its not better or worse inherently by itself. It has specific applications and benefits/drawbacks just like anything else.
Alpha-N is nothing more than a method to calculate engine flow. Thats it. The ability to properly calibrate an engine or “tune” is entirely dependent on the ECU’s ability to calculate the flow of air through the engine at any given tiny interval of time. The more accurately the ECU can do this and the smaller the time interval it can update this flow number, the better.
What do I mean by “Flow”? The ECU’s job is not to calculate the volume of air flowing through an engine. Technically its job isn’t even to calculate the mass of air flowing into the engine. Its job is to calculate the mass of Oxygen flowing into the engine. That only makes up about ~21% of the air in the atmosphere here on planet earth. This composition can slightly vary. Oxygen is the reactant thats used with fuel during combustion.
That is where we convert chemical potential energy into usable heat energy. The more heat energy combustion generates, the more average force it applies on the piston through the relevant time interval and well…you can take it from there.
How does the ECU calculate the amount of oxygen flying through it?
Two things: Sensors and Assumptions. In reality everything is based on assumptions at a certain extent.
A lot of load calculation by the ECU is done by assumptions. Now, a lot of people are saying
“Stop right there, how can that be a good thing? That can’t be an accurate method can it?”.
It can and it is...IF its done correctly.
Assuming everything is working right, ECU’s make assumptions about things that they don’t expect to change and make measurements (with sensors) of things that are expected to change.
A big example of an “assumption” the ECU makes is injector flow. An ECU is programmed to know that its injectors will ALWAYS flow a certain amount of a fuel when it commands a certain pulse width. It makes this assumption because it doesn’t expect the fuel pressure or injectors to change. If those two things do not change then the ECU can always assume that at a certain pulse width the flow will be a certain mass of fuel. If you change the injectors or fuel pressure, it makes its assumptions inaccurate. It will get different fuel flow for the same injector pulse width as before. As a result in closed loop, your trims will start to change to adjust for this. Your WOT AFR’s will be different too. If you had a fuel flow sensor, then you can program an ECU so that a change in the fuel system can be accounted for. The S62 ECU and most ECU’s however do not expect the fuel system to change because thats modification and so therefore do not need sensors to take measurements. They can get away with assuming its an unchanging variable.
That is the entire point of “tuning”. Tuning is done when you change things the ECU assumes shouldn’t change though not exclusively. That is fuel flow or airflow. With enough sensors and advanced enough software you can make a self-tuning ECU and that is actually not that difficult to program.
Anyway, lets continue:
An example of something thats measured with a sensor is mass air flow via voltage. An ECU has a MAF flow curve that tells it at Voltage=X then Flow=Y. In reality however this too is kind of an assumption. What can cause this assumption to be inaccurate? A leak after the sensor perhaps, or maybe buildup of any insulating residue on the wire in the MAF sensor. You’re measuring it, but you’re also assuming that you have no leaks or any other factors that can change. As I said before, we’re assuming everything is working properly.
A MAF sensor is one way to account for changes in the atmosphere that matter. A MAF sensor accounts for air density (which is affected by pressure and temperature). That is how it determines the mass/time of flow.
If you increase the flow of air through the wire, it cools it. If you flow the same volume of air but increase density (either by pressure or temperature) then the wire once again gets cooler as the air becomes more thermally conductive so it pulls heat away. All of these things affect the voltage and therefore the MAF sensor can account for the mass of air flowing through at a pretty astonishing accuracy in steady-state conditions.
The best part of a MAF sensor is that it not only accounts for atmospheric changes, but it directly sees the amount of air your engine is pulling in. Open the throttle further? More voltage. Increase RPM? More voltage.
In fact, accounting for atmospheric changes is REALLY easy no matter what you do: MAF, MAP, or Alpha-N.
The beauty of a MAF is that it accounts for changes in engine VE and that is why its so common on street cars for emissions.
For example, clogged air filters change engine VE. A change in engine VE always requires a re-tune on a MAP sensor setup or Alpha-N, otherwise fuel trims start to wander and lambda control becomes less consistent. On street cars, many of which are neglected, a MAF is an amazing way to make sure the car drives good and keeps AFR’s at the stable Lambda=1 regardless of whats happening.
A quick note: A MAF cannot really account for oxygen content, but thats okay. Atmospheric Oxygen content hardly varies and we can leave that to an assumption because it will lead to unnoticeable margins of error.
There is another entirely different method of calculating airflow and this type of calculation is called “Volumetric Efficiency”.
A MAF sensor system doesn’t care about volumetric efficiency of an engine. Whatever the engine flows is what it flows and that will be measured by the MAF. When you don’t have a MAF and you’re not measuring flow…you need to somehow come up with a value of flow using more indirect methods.
These indirect methods are not inherently less accurate or less capable. It all depends on how they are done in practice.
The two ways I’m talking about are using a TPS or a MAP sensor. The former method is commonly known as Alpha-N and the latter is known as Speed-Density.
Lets take a step back and look at how they work on a fundamental level.
If you run an engine at a certain throttle angle opening and then run it at a certain RPM, its flow VOLUME will ALWAYS be the same assuming nothing else changes. If you take an engine and create a certain pressure in its manifold at a certain RPM, it will always flow the same in those characteristics assuming nothing else changes.
So what we can do is start building tables in the ECU basically that say this:
If Manifold Pressure=X AND RPM=Y THEN VE %=Z
with Alpha-N its the same thing except:
If Throttle Angle=X AND RPM=Y THEN VE %=Z
This is VE tuning. You have to measure the VE in as many possible combinations of engine operations as you can. The finer you go, the higher the resolution and the more accurate it is. For example you can measure VE in increments of 100RPM or increments of 10RPM. You can measure VE in increments of 1 PSI of pressure or every .1 PSI of pressure. You can measure VE in increments of 1% throttle angles or 10% throttle angles. A table of VE can be lets say…32x32 cells. Thats a resolution that gives you 32 increments of RPM and 32 increments of your controlling sensor which is a TPS for Alpha-N or manifold pressure for speed density. Anything in between can be interpolated or the closest cell is used so margin of error is within tolerances.
This can be used to to calculate the VOLUME of airflow at any RPM and throttle angle. Part of the work is done.
So how do we get mass? Easy, an IAT sensor and Barometric pressure sensor.
Our formula can look more like this now:
If Throttle Angle=X AND RPM=Y AND IAT=A AND Pressure=B THEN Mass flow = some value of mass/time.
you can also rewrite it to be
If Throttle Angle=X AND RPM=Y THEN VE %=Z and then take the Z value and correct that volume to temperature and pressure to find density. From there you get mass. It doesn’t matter what order, the logic works here.
Now lets to get to when Alpha-N is better and when MAP is better.
Alpha-N works just fine for NA and SC’d engines. The reason is because flow of the supercharger is constant and directly related to RPM so we can make an assumption about that without actively measuring it on the street just like a naturally aspirated engine. It can get a bit complicated during rapid acceleration of RPM, but it still works within tolerances.
On a turbocharger, boost changes all the time and is not consistent at all with throttle so making assumptions is impractical (but no impossible, with enough sensors you can do it). Thats why MAP sensors are used for turbos. A MAF also works for this, but due to a very long and complicated reason, in my opinion a MAP is better for turbochargers.
Using a MAP sensor on an S62 would actually be a challenge at partial throttle. S62’s do not really have an accurate vacuum source that can be used to reference engine load. A MAP sensor can certainly be used at WOT for boost, but Alpha-N or a MAF is the only option at partial throttle. This is clearly proven by the fact that M5 owners with FI are seeing boost registered in their sensors located AFTER the throttle even at light acceleration and load. This clearly means a MAP cannot be used realistically.
Real race cars using Alpha-N generally don’t have “closed loop” with O2’s. Street cars do and IMO a more accurate way of calling it would be MAFless. O2s are a big deal. Just like with a MAF sensor, the O2s can actively check AFR and make corrections for stability during part throttle operations.
Now lets move on to S62’s.
S62’s (as far as the evidence has pointed) still use O2s to make corrections in Alpha-N operation. With a MAF system, the ECU takes a reading of flow, then provides fuel and then sees feedback from the O2 to oscillate at the stoichiometric lambda. With Alpha-N the ECU uses its formulas (or you can call them look-up tables) to deliver fuel in the same way and then checks accuracy from the O2 sensors.
The downside to Alpha-N (and MAP sensors) is that ANY change at all to the engine will toss its VE assumptions and can lead to drivability problems if they are big changes. With Alpha-N the engine has no idea if you changed a flow characteristic. A MAF is a lot more tolerant of that because any change that changes flow will affect voltage and the ECU can see that to appropriately compensate. That is why fuel trims don’t really change as the weather changes or when you install mods on a MAF car.
On an Alpha-N car, every modification you do or even a clogged filter will blindside the ECU.
For example on a MAF car, I can take my S62 and add a completely overhauled exhaust with headers, X-Pipe, wider piping and higher flow cats. I can also installed bored throttles, better stacks, less restrictive filters…etc. It doesn’t matter what I change before or after the MAF. The MAF measures flow and the car will start right up and drive even though I may have just drastically changed the VE at every RPM and throttle combination.
On an Alpha-N car you might have horrible drivability if the changes are large enough and it may require quite some work to model the air flow. The ECU does not know whats going on and therefore must be calibrated. By the way, on an exclusively MAP sensor based ECU, the same rules apply.
Also, I do not know if S62’s have ambient pressure sensors, but that is okay because pressure really doesn’t THAT much. The main contributor to air density changes from weather is temperature anyway. In my opinion, an IAT is enough.
Ambient pressure can change a lot between regions (mainly altitude), but not usually so much in a single region so that can accounted for in custom tuning much like your modifications.
To summarize, if Alpha-N is done correctly then there is no problem or worry in running it. It works just fine for a naturally aspirated or supercharged engine. There is no inherent deficiency in this system over a MAF sensor for this type of engine. The key words are that it has to be done right and that is entirely dependent on the tune itself.
Let me know if you have any questions.
kjbruckner, liquidsmoke, Rontgen and 3 others like this.
SOLD: 2000-BMW-E39-540i/Biarritz-Blue/Gray-Interior/Black-Trim/Sports Package/6-Speed/DSP. Powertrain: TTFS Stage-2 VF-Engineering Supercharger | Dinan Manifold | Magnaflow 16858 Exhaust & X-Pipe | High-Flow Cats. Suspension: ECSTuning Cup Kit (Koni Yellow/H&R 50464) | BF-Goodrich Comp-2 A/S. Cosmetic: Front & Rear M-Tech Bumpers | Slimmbones Finned Rear Diffuser | Piano Black Interior Trim/Style 37's. Extra: S62 Clutch & PPlate | AEM 320 Fuel Pump | AEM Wideband/Boost Failsafe | E60 SSK | ZHP Knob | CDV Delete | Blitzsafe Aux | Wolverine Pan Heater Current: 2001-BMW-E39-M5/Carbon-Black/Silverstone-Interior/Sport Seats/M-Audio Powertrain: Dinan Exhaust YouTube Channel Instagram
Last edited by mcgnms; 30th August 2016 at 05:36 AM.
mcgnms is offline
Quote Quick Reply
The Following 3 Users Say Thank You to mcgnms For This Useful Post:
auaq (31st August 2016), El Diablo (30th August 2016), gsfent (30th August 2016)