So I have been thinking about the AFM removal and I can't see why you couldn't do it when you have a SAFC2 atleast and some electrical knowledge.

I have not yet done anything but I have studied the wiring diagrams, ECU pin chart and AFM specs.

The fuel pump switch is easy to bypass. Connect 2 wires and your done.

So I was thinking, there are 2 inputs to the ECU from the AFM and One output which is around 4.8v which goes to a few different sensors including the AFM.

On a Series4, Pin 2A is the output with about 4.8v. Pin 2E is the INPUT for the AFM Flapper signal which sees raw voltage and Pin 2J is for the AFM Intake temp sensor.

Now my idea is to have a preset temp, say around 70*F which would be around 2.5K Ohm, according to the FSM. Then the AFM flapper output I would have at around .50v which is around 4000rpm at WOT.

I would take the volts from Pin 2A and drop the volts down to .50v into the AFM input of the SAFC and then the AFM output of the SAFC to Pin 2E which should cause the engine to run without the AFM. My guess is that lower rpms would be very rich and high rpms would be very lean.

To correct this I am going to use my SAFC2 with my full range TPS and set the LOW throttle point to 20-30% and HI throttle point to 60-70%. Then add or remove fuel accordingly at specific rpms above and below the throttle points.

The SAFC will be able to calculate the fuel amount between the 2 throttle points.

I don't know if this will work perfectly but I want to try it and I already have a wiring diagram made and im going to see if I can get this to work.

I got the idea when I read an article about bypassing the AFM and still keeping the afm but correcting the fuel with a fuel computer. So instead of that, why not just remove the afm and manipulate the ECU into thinking its still there.

Well, I thought I would share. I will see if I can make it happen.

 

So how will you provide engine load data to the ECU? Does the S-AFC2 support a MAP sensor as an input?

By the time you pay for an S-AFC2, you can buy a Megasquirt and have a full standalone.

 

^Beat me to it lol

There's no way you can eliminate the AFM without going standalone.

 

I already have a SAFC2, I have had it for years. I will eventually sell that and get a Rtek. I don't feel like getting a megasquirt, for some reason, I just don't like it.

I was thinking of hooking the SAFC up top the pressure sensor for better load calculation, people have used the pressure sensor with success, i have not used the SAFC with the pressure sensor yet but it will be easy to do. The SAFC uses load and rpm which might work out ok.

 

I'm not sure that the S-AFC can translate the signal from the pressure sensor well enough to replace the AFM. For one, the voltage curve of the pressure sensor is opposite that of the AFM.

 

Hahaha, for once I can tell somebody that their crazy idea may actually work. There is a fuel injection method called "Alpha-N" in which the load is determined by the TPS, and the ECU compares this to an RPM input to create the X-Y fuel and ignition tables. I don't have the electronics background to know how to modify the SAFC2, or to know if it is even possible, but I can tell you that Alpha-N does work. The advantages of Alpha-N are that there are absolutely no restrictions in the intake (no AFM, MAP sensor, IAT sensor, etc.) and it is totally impervious to pressure reversions unlike a flapper AFM, MAP sensor, or hot-wire MAF sensor. The disadvantages of Alpha-N are the incredibly critical tuning of the TPS (the crummy S4 TPS isn't a good choice), and that the engine will need to be constantly re-tuned whenever weather conditions change, much like high-performance race engines that run carburetors. The 4-rotor Mazda 787 used Alpha-N fuel injection. For the trouble you are going to have with this conversion, it would be better to get a weekend job at McDonald's until you can afford a cheapie Microtech EMS, but you may find it a fun experiment.

 

The fundamental flaw in this plan is that it'll require so many compromises (reduced ignition timing control and fixed temp input being obvious ones) that the losses will almost certainly outweigh the gains. If your engine makes enough power for the AFM to be a significant restriction, you need to be looking at a far better tuning device than an S-AFC anyway.

Alpha-N is for race cars, not street cars...

 

The ECU will still have a varied voltage to Pin 2E from the SAFC for the AFM signal so it will still adjust timing accordingly. Also the fixed intake temp voltage for the AFM does not effect timing directly. The temp sensor is there to help the ECU determin the air mass coming into the engine and it does this in conjunction with the airflow meter portion. Between the vaired voltage from the SAFC to pin 2E and the solid voltage at pin 2J, the change in timing the ecu does in conjunction with those 2 figures should not be a problem at all.

I want to see what the gains are instead of all the words that get spread. Its a cheap mod for me if it works. Im obveously going to use caution since I don't need to pop my motor

Also, I will get the rtek some time and I hope that, IF this work, and thats a IF, that the rtek will be able to work better then the SAFC without the AFM.

 

I must not have been clear some where. sorry.

The TPS input wire of the SAFC will be hooked up to the pressure sensor. I was going to keep it on the TPS but that wouldn't work out at all for determining engine load.

The AFM input wire on the SAFC will be hooked up to Pin 2A with a resistor to drop the voltage. On my engine, the volts at Pin 2A are a steady 4.88v. I was concerned at first that the volts would vary and that could cause issues.

The AFM output wire on the SAFC will be hooked up to Pin 2E, just like it would be normally when installing the SAFC.

So the ECU will still get a varied AFM signal input from the SAFC. The SAFC will just see a steady voltage of around .50v from Pin 2A so I would have to remove and add fuel where needed. Like I said above, I think in this case, it would be rich below 4000rpm and lean above.

 

Sounds like what the SAFC does. Except you can't adjust individual throttle points with the SAFC

 

Wrong.
This is why FD's blow up left and right and FC turbos do not.
Since you're the mechanic, I don't need to explain why this is so.

I'm assuming you're talking about a non-turbo.
Why you want to remove the AFM for?
You don't like how it looks?
You're trying to make more power? I seriously doubt it.

So this ends up as an exercise in futility?


-Ted

 

I would also like to add that the AFM is a restriction on a n/a without porting. If you hookup a manometer to the intake after the AFM And do a full throttle run, you will see how about 470cfm is having a little difficulty moving through the small cross section of the AFM with the spring loaded flapper in its way.

 

Even at 100% VE, an NA 13B at 7000rpm only flows 323cfm. So either your VE is 145% or you're revving it to ~10,000rpm. Impressive for a stock ports...

I am genuinely interested in your manometer readings though. What were they and what did you use?

 

HELL NO! I have never and will never work and McDonalds. To many bills to pay, 12/hour that im at right now is as low as I can go. I like working on the air field. Its fun

But yes I will have fun experimenting. I do see how it can work, but I also see how it may not work so it will just be fun to do and see if I can really make it happen. I don't seem like it will be vary hard at all.

 

Those who have installed an aftermarket EMS over the last 15 years already know the answer. Sorry, removing the AFM is nothing new. The only new thing you are trying is to use Alpha-N, which does have the potential to work, but your configuration will not have the tuning capabilities of current high-resolution standalone EMS products. If you want to use Alpha-N and make real power gains, then you will need to install a standalone EMS that can run in Alpha-N mode (Haltech, Wolf, Motec, etc.), in which case you would be much better off using the popular speed-density mode if this is for a street car.

The problem with the SAFC, Rtek, and every other piggyback is that they rely on the horrible resolution and ancient design of the stock ECU. Their only advantage is that they are easy to deal with when compared to a standalone EMS.

I think you misinterpreted my post. You will end up making about 30 cents an hour when you consider your time invested vs. what you could have paid for a professional aftermarket EMS installation that will work better. You WILL come out ahead working at McDonald's. Therefore, if you choose to try out this project, you will be choosing to make less money than you would at McDonald's. The choice is yours, I am just telling you the economics involved.

 

This is hardly a new idea as has been pointed out.

Lots of cars run airflow meter translators. Supras and 3000GT's for example run the MAF-T pro, and there's also E-Manage Ultimate. They use a MAP sensor and have an internal algorithm to supply an airflow meter signal to the stock ECU.

No working translators exist for our car. The SAFC II is not a translator! It just corrects the airflow meter signal. It doesn't calculate load.

And Alpha-N is far too limited for daily driving.

 

The real issue with Alpha-N is that you can have multiple load points for one throttle setting and RPM range. It's really only the choice when you have radical porting or something like ITBs that won't provide a good vacuum signal.

Making a car running TPS referenced run smoothly at low throttles and around town driving is very frustrating. I know, I've done it.

If this is something that you really want to pursue, start with a generic MAP sensor that the voltage curve is available for like the GM unit. Rig up the S-AFC on the bench with the MAP as an input and see if you can adjust it to mimic the AFM signal based on various vacuum loads.

Or use a Megasquirt.

 

Never said removing the AFM is new. Most with standalone EMS have also tuned timing and fuel curves including removing the AFM so they will notice a gain from the combination and not from removing only the afm.

Seems to work good so far.


The experience is worth more.

You are right, it corrects the airflow meter signal which is what it would do. Instead of the SAFC measuring throttle opening, it would be measuring load with the pressure sensor instead. Like I said, this may or may not work.

IF there was a MAF translater for our cars, I would not be doing what im doing.

I am going to measure the voltage curve for the pressure sensor soon to see how it reacts to the engine load.

 

Also, I believe the rtek, when it does support afm removal, digital tuning said they were going to use the pressure sensor and I think the CAS and TPS to find out the load.

 

There is no way to determine engine load via the CAS. All it does is provide the ECU the position of the engine so it knows when to fire fuel and spark.

The TPS is also not much of a load indicator since it just measures throttle position. The throttle may be wide open, but the engine under very low load. Or the throttle may be nearly closed but the engine fully loaded.

Combining two inputs like TPS and RPM will give you a general idea of load. For example if the TPS is 50% open and the engine is at 5000RPM, you can make an assumption of high load.

Most standalones run MAP sensors is speed density configuration. This means that they use manifold pressure vs. RPM to determine load. It's an accurate and easy to tune method as you can generate a table of injector times with MAP on one axis and RPM on another. This ends up making hundreds (or thousands) of load points that can be individually tuned.

HOWEVER, there is a major disadvantage to this configuration. Since the ECU is not measuring air into the engine directly, you can increase airflow through the engine but still measure the same load point. 10 PSI of boost at 5,000 RPM on the stock turbo is not the same about of air as 10 PSI at 5,000 RPM on a GT45! The result is that injector open time is not increased, the engine runs lean, and then pops. This is why FDs are easy to blow up, and why standalones must be retuned when you do anything that increases or decreases flow through the engine. If Mazda had continued to use an AFM, they would have been reliable and easy to mod like the 2nd gen. As the AFM knows exactly how much air is entering the engine, the ECU can make adjustments to compensate for more air relatively easily if the map allows for it.

 

I know, I was just saying that I think they were going to use a combination of sensors to determin the exact load. Just from what I hear, don't know if thats how they will make it happen.

This will also be on a non turbo so I don't have to worry about the issue with FDs but I do thank you for giving that information.

 

Well, you need to know more than RPMs and pressure to determine the amount (mass) of air going through... you also need to know the temperature. If the stock turbo and a GT45 were both boosting 10 psi at 5000rpms, and the air in both was the same temperature, then they'd need the same amount of fuel. However, since at that pressure and flow rate, the GT45 is more efficient, it heats up the air less as it compresses it, making the air more dense, and thus have a higher mass for the volumetric flow rate at 5000 rpms.

Basically agreeing with what RETed said, but giving the non-dumbed-down version

 

Thanks for the information..again.

I would like to state one more time.....
This engine is a NON TURBO

 

haha, actually I think you are right. From what I remember I thought it was closer to 370CFM

I may have to retest it again, I tested it about 2 months ago and I can't seem to find my notes on it. The manometer is a home made mano meter with a tube that curves 180 degrees at the bottom. I actually have a better spot to hook it up this time so the results my be different, but infact shouldn't. Still, I am going to retest it.

Sorry for the delayed responce.

 

654cc x 2 rotors x 7000rpm = 9156L/min. 1cu.ft = 28.3L, so 9156 / 28.3 = 323cfm