Factory MAF theory

[DucatiRdr]

Here is a question that I didn't see raised:

What was Ford's thinking behind putting a 55mm MAF sensor on a 60mm throttle bodied motor? How about the 94/95 70mm MAFS?

What is the actual flow capacity of the 55mm and 94/95 70mm MAFS?

Per the Ford Racing catalog:

If the MAF meter is not large enough, it will become a restriction in the intake path and will limit the overall horsepower output that the engine is capable of. The MAF meter’s calibration is equally important. If you are using a different fuel injector than the engine was originally equipped with, the computer has no way of knowing this. By introducing a properly calibrated MAF meter matched to your injectors, the computer can be ‘tricked’ into thinking that the engine is receiving less air. If properly calibrated, this will cause the computer to generate a smaller injector pulse width, allowing the larger injectors to function properly throughout the entire RPM range. The potential down-side to this method is that on Ford electronics, the PCM schedules spark advance as a function of (among other things) engine speed and engine load. Engine load is defi ned as the mass of air entering a given cylinder divided by the mass of air that can fit in the cylinder, and is calculated directly from the MAF signal. If the MAF sensor is “tricked” into thinking the engine is receiving less air, then load will be artificially low. Since required spark advance typically decreases as load increases, this “tricked” MAF will result in additional spark advance being scheduled, which can result in spark knock.

Be aware that the MAF meter houses the single most important sensor in a Ford fuel injection system! The engine’s air/fuel ratio and spark advance are primarily determined by the computer from the input received by the MAF meter. This is why it is of the utmost importance that there are no vacuum leaks present in a MAF-based EFI system.

There are several ways to “trick” the MAF meter. The fi rst method is to change the MAF meter’s voltage output by manipulating the electronics of the meter. For EEC IV electronics, this can be an extremely accurate way to calibrate a meter, since the meter’s “curve” can be precisely targeted to refl ect the needs of the new application. A second method is to manipulate the signal from the meter by mechanically changing the amount of air that is permitted to pass by the filament of the MAF meter by installing a different size ‘sampling tube’ or restricting the fl ow through the tube with a screw. While this method can be effective, and can work quite well at wide open throttle, it is frequently the source of idle and low-speed drivability concerns. Another source of idle and low-speed drivability issues is using a meter that has far too much capacity for the application. For example, don’t use a meter good to 750 HP if you only
plan on making 200 HP. The low-flow resolution of the high horsepower meter will not be as good as a meter designed for a 200 HP application, and issues can result that are impossible to tune around.

There is also a method of using a mass air meter that has not been calibrated at all. However, these meters will usually require the use of an aftermarket “chip” or “flash” to work properly. Using an “off-the-shelf” or “mail order” computer chip is discouraged with these types of mass air meters. If you do choose this type of meter, it is strongly advised that you have the overall combination tuned on a dyno while monitoring the air/fuel ratio. Provided this is performed by a competent tuner, this is the best method and will result in the best part-throttle drivability and idle. Additionally, before tuning on a dyno, you should BE ABSOLUTELY CERTAIN that the ground circuits for the EFI system are in pristine condition. Otherwise, you are merely tuning around a problem, and a tune that works well one day, can be substantially different if the ground signal varies. You can actually have a tune that works properly when tuned ‘around’ a poor ground, and it is then possible to make it perform poorly by simply correcting the ground signal! It can’t be overstated that prior to the vehicle being tuned in any way, all vacuum leaks, electrical issues, etc., need to be resolved. Fixing them before you go to the dyno will always be cheaper than paying for dyno time while you’re wrenching on your car.

As a general rule of thumb, the following mass air meters will support the corresponding horsepower:

MASS AIR METER Horsepower Level
55mm (Stock 88-93 Mustang) 275 HP
70mm (Stock 94-95 Mustang) 350 HP
80mm (Stock Ford) 425 HP
90mm (M-12579-54) 540 HP

 

[seijirou]

Here is a question that I didn't see raised:

What was Ford's thinking behind putting a 55mm MAF sensor on a 60mm throttle bodied motor? How about the 94/95 70mm MAFS?

Driveability.

A smaller diameter MAF gives more notable, and therefore more detectable changes as the throttle blade is opened.

Remember, the size of the inlet needed is decided by the combination. Just because the TB is 60MM doesn't mean it actually needs to be that big.

On a stock 5.0 the 55MAF is hardly a restriction if any at all. The bottleneck is in the lower intake manifold, not the upper, not the tb, and not the MAF. Once you being opening up one bottleneck, something else becomes the bottleneck. But stock for stock, the stock MAF is just fine.

 

[smokin91']

I noticed an improvment in my throttle response when i switched over to the 70mm mass air unit from a 94-95.

 

[seijirou]

I think the sn95 units actually use different sensor hardware that's supposed to be better in one way or another.

 

[smokin91']

Or it could be the fact that its 55mm VS 70mm.

 

[Daggar]

Your "response" won't have anything to do with the size of the opening in the meter. Not in that context, anyway.

 

[seijirou]

Think of it like this. Try sucking air as fast as you can through 3 objects. A straw, a paper towel tube, and some 3 inch diameter pcv pipe. You'll feel that the straw is holding you back. The paper towel tube won't slow you down at all because it can easily carry all the air you're asking it to. Sucking air through a PCV pipe won't let you draw the air any faster either, because after the paper towel tube, it's not the tube that's holding you back, it's your lungs that can't pull air faster.

So in performance, there is no difference between the paper towel and the pcv tube. If you stick the paper towel tube on the end of the PCV pipe and seal up the gaps, and then suck air through that, again, you won't notice a difference. The paper towel tube isn't going to hold you back even though it can't flow as much as pcv pipe can, because your lungs are the inferior air movers not the tubes.

A 55mm MAF on a 60mm TB on a STOCK MOTOR is the same thing as sticking the paper towel tube on the end of the PCV pipe and sucking air through them. Yes the paper towel tube can't flow the air the PCV pipe can, but, you can't suck that hard anyway, so it doesn't matter.

Get some big ass lungs, and then we have a different story.

Your MAF is more than bigger, it's also better electronics, more accurate measurements, closer signal steppings, more sensitive instrumentation. You changed alot more than size, so you can't logically try to give all the credit to the size alone

 

[vristang]

I think...
The performance gain experienced by smokin91' is due to the larger bore size of the 70mm MAF.
The MAF Transfer (output voltage) of the 55mm and 70mm Mustang MAF meters is virtually identical.

Also, the dyno book by Holdener showed a 5-7hp gain on a completely stock Fox Mustang and a larger MAF.
The 55mm MAF did sacrifice a little power, but it wasn't very much.

As for why the Mustang had a 55mm MAF and 60mm TB -
Who knows?
Unfortunately Engineers don't always get the final say, so it is possible that budget, politics, or packaging played a factor in that decision as well.
Perhaps there are some noise restrictions that could only be met with this setup?
Most of us have experienced a noisey aftermarket TB.

jason

 

[DucatiRdr]

What about the design differences between the 55mm and 70mm. My 55mm is a straight shot where the 70mm has the backflow preventer down the middle. How does the backflow preventer work? On the 70mm, why doesn't the sampling tube run straight through - it looks like it is routed through the backflow preventer.

 

[stang&2Birds]

It was because of cost and the abilities of that time in history. And, guess what - humans do not know everything that is possible to know!

Gee, today, for $15, I can have a programmable chip that has the equivalent of over 2,000 1985 chips and it's super fast (Xilinx Spartan 3E's vs 244's (etc)).

Look back to when the Fox MAF was designed. How would you like to design a complex part that deals with fluid dynamics without computers! Gee, people won't even take 2 hours to read basic fluid dynamic theory because "it hurts their brain".

Ford did many many cheap *ss things. Uhm, make that a TON of cheap *ss things. But, the Ford Fox MAF design was pretty much at the max ability of engineers at that time to make a *cost effective* design.

The EEC was done because of cost. It was ~$10-$40 cheaper than the GM PCM. Figure that Ford makes a few million EFI cars a year, and you can see why saving $10-$40+ per vehicle is *important*. And, the EEC-IV "works". It may be pathetic compared to the GM PCMs, but they still work fine! As an Engineer, it's important to make cost vs performance trade-offs.

As someone that's been a principal engineer for over a dozen years, it's *my job* to define architecture and to come up with a GOOD cost/performance trade-off.

FYI:
~1.2 Meg
Fox - LMAF- 94/95 MAFs
http://www.veryuseful.com/mustang/tech/tmp/mafs-fox-LMAF-94.jpg

Last: No-one is preventing "you" from designing and selling your own MAF/ EEC/ CAR/ what-ever. So, if it's so "easy", then go make a zillion dollars! Also, *anyone* can completely rewrite rewrite the EECIV code and then sell an add-on chip to replace the EEC program. So, go for it!

 

[jrichker]

It was because of cost and the abilities of that time in history. And, guess what - humans do not know everything that is possible to know!

Gee, today, for $15, I can have a programmable chip that has the equivalent of over 2,000 1985 chips and it's super fast (Xilinx Spartan 3E's vs 244's (etc)).

Look back to when the Fox MAF was designed. How would you like to design a complex part that deals with fluid dynamics without computers! Gee, people won't even take 2 hours to read basic fluid dynamic theory because "it hurts their brain".

Ford did many many cheap *ss things. Uhm, make that a TON of cheap *ss things. But, the Ford Fox MAF design was pretty much at the max ability of engineers at that time to make a *cost effective* design.

The EEC was done because of cost. It was ~$10-$40 cheaper than the GM PCM. Figure that Ford makes a few million EFI cars a year, and you can see why saving $10-$40+ per vehicle is *important*. And, the EEC-IV "works". It may be pathetic compared to the GM PCMs, but they still work fine! As an Engineer, it's important to make cost vs performance trade-offs.

As someone that's been a principal engineer for over a dozen years, it's *my job* to define architecture and to come up with a GOOD cost/performance trade-off.

FYI:
~1.2 Meg
Fox - LMAF- 94/95 MAFs
http://www.veryuseful.com/mustang/tech/tmp/mafs-fox-LMAF-94.jpg

Last: No-one is preventing "you" from designing and selling your own MAF/ EEC/ CAR/ what-ever. So, if it's so "easy", then go make a zillion dollars! Also, *anyone* can completely rewrite rewrite the EECIV code and then sell an add-on chip to replace the EEC program. So, go for it!

Since you seem to have spent more time digging in the EEC innards than most of the rest of us, I have a question. Was the binary operating code entirely in the EPROM soldered into the PCB or was any part of it in mask ROM that was part of the CPU chip itself?

Intel made a series of CPU chips that included scratchpad memory and a mask ROM that was programmed during chip fabrication. The cheap way to do the EEC would have been to build one computer PCB and then change the way it worked by using different EPROMS for different engine/emissions requirements. Build one baseline PCB and add components as the number of cylinders and controlled accessories increased. Or build one PCB that had everything and change the EPROM the target engine/chassis.

Seeing that there are always Ooops and improvements in the development cycle of any computer controlled device, using an EPROM would have been the thing to do. Once the design is cast in concrete and isn't changing, then switch to a mask programmed ROM on the CPU chip die that would be part of the chip fab process. To the best of your knowledge, did Ford ever use the combined mask prom/CPU approach in the designing of the EEC-IV?

 

[stang&2Birds]

It was all in the PROM. (EPROMs are erasable through a window on the top of the chip and using UV light)

Using PROM and not a socketed PROM was one of the *cost* decisions that Ford made. Again, looking at the *cost* savings, it can be debated if it was a smart thing to do or not.

GM made a number of PROM changes for their cars. For GM, all it meant was a simple chip replacement. For Ford, since it meant replacing the whole EEC, Ford just left the problems alone. The hunting idle problem (even with a new IAC, EGR, TPS) is a classic example. That's also why Ford made their "idle adjustment plate" a Ford service part number and part of their TSBs.

There are many trade-offs in building a generic PCB. It's a longer design cycle. It takes much longer to upgrade. the board is bigger. The power supply has to be designed for the worse case system. It takes much longer to debug. EVERY group wants ("must have") their own little thing that is "critical". However, the other groups often see that feature as worthless and a waste.

Think of PC's. Even today, how easy would it be to build *one* PC that fit 99% of everyone's needs? So, it's best to *try* to do "modular" designs. That way, one group can leverage what another group has already done. The same is true for software. Today, well defined routines are called to create and control a Window.


One of the BAD decisions that Ford chose was to not include a live data stream (extended diags) in all of it's EEC-IV systems. IMHO, very stupid! IMHO, that clearly shows the influence of a manager that was *totally clueless* on software/ hardware/ computer/ control system debug. That decision cost Ford $$$$$ in unneeded replacement parts, long mechanic hours, bad reputation, flaky problems that they didn't have a CLUE about the causes for a long time, and so on.

Overall, I *am* very impressed that Ford was able to "pull it off" with their low cost EEC-IV EFI systems. With the exception of being able to send real-time sensor data for all of the EEC-IV systems, they did an impressive job at a low cost!

"Thankfully", the government required the basic OBD-II stuff in '96. It's one of those cases where a Federal requirement likely *saved* companies money in many cases (easier/faster debug for the dealer mechanics).

Then, there's the EEC-III pure cr*p. The government should have sued Ford $$$$ for putting out such pure cr*p. The ECT sensor (or many other sensors) goes, or the wire breaks, and the car dies/won't start. Again, it was all done for reasons of *cost* (smaller PROMs and simpler/ quicker/ cheaper software development).

I'll go into more detail about the disadvantages of generic board in the new post.

 

[stang&2Birds]

This is directly related why the "cheap" MAF was used on a Fox.

Here are just some of the direct cost disadvantages of using a single generic board design:

1) Bigger circuit board. Circuit boards are done on "panels". Back then, I *think* the biggest panel was approx 12" x 12". If not, it's good enough for an example. Each panel costs money. It's best if you can get as many boards out of a panel as possible. If a circuit board is size "X x Y" and that allows 4 circuit boards per panel, then even a small increase in board size will reduce the number of boards from 4 to 2. That's double the cost for the raw board!

2) More design options on a board means more drill holes. A drill hole is needed for each pin and for each "via" (transfer of a trace from one layer to another). Drill holes take time. that means money. Drill wear out, that means money.

3) Raw boards need to be tested to make sure that the traces are good. More components means more pins, that means more money. Bigger boards means a bugger tester, that means more money. More test points take more time to test, that means more money.

4) A generic design would increase the board size from 2x to 4X.

5) There's very little room for the EEC as it is in the cars.

6) A bigger board means a bigger metal enclosure and a stronger bracket. Again, more money.

7) When a change/update is done, it would be a much bigger effort to fit it in on a big design. More components and traces would need to be moved.

8) A generic board would have meant an even bigger connector. That means a bigger board and cable connector. A significant cost increase.

9) .... as I (or others) think of them.

 

[jrichker]

My main interest was to understand if it was possible to unsolder the prom from an A9L, copy it on a burner. Then put the copied prom in one of the more common 5.0 computers from that era and have an A9L.

I also though about using an EEPROM and some add on components and have as result a flashable computer that didn't need an external J3 card to use a different program.

 

[Daggar]

A flashable EEC-IV! Hmmmmm... At first I thought "Sweet! I'll be the first customer.", but I wonder if that would be any better than having a chip on the back.

What would be the advantage over using the J3 port?

 

[jrichker]

A flashable EEC-IV! Hmmmmm... At first I thought "Sweet! I'll be the first customer.", but I wonder if that would be any better than having a chip on the back.

What would be the advantage over using the J3 port?

Maybe we could get Stang&2birds to whip us up a Xilink PLC array that connects directly from the J3 port to the USB connection on a laptop...

Think about instant changes to the base binary code that didn't depend on a device permanently attached to the J3 port. The abliity to modify any part of the binary code on the fly and have it remain until it was reprogrammed. You never have to replace the chip on the J3 port. You just plug the device on the J3 port, make your changes and hit save. It writes them to the EEPROM on the EEC and you are done. Disconnect your Xilink PLC and cables and store them away with you laptop until the next changes are needed.

 

[310stanger]

i think jrichker made a friend. I need to read this thread when I havent drank a 6er of beer cause this is a great thread. I LOVE technical discussion like this. Ill be lucky to make sense of it sober though. Can this possibly become a sticky? Theres lots of good theory on the eec computer and MAF in this thread.

 

[JChalfan]

i think jrichker made a friend. I need to read this thread when I havent drank a 6er of beer cause this is a great thread. I LOVE technical discussion like this. Ill be lucky to make sense of it sober though. Can this possibly become a sticky? Theres lots of good theory on the eec computer and MAF in this thread.

I think this would go well in the "Useful Technical Thread Index" sticky.

There's a lot of good info in this thread... It's definitely given me some more insight into how the EEC-IV system works.

Jeff

 

[Daggar]

Maybe we could get Stang&2birds to whip us up a Xilink PLC array that connects directly from the J3 port to the USB connection on a laptop...

Think about instant changes to the base binary code that didn't depend on a device permanently attached to the J3 port. The abliity to modify any part of the binary code on the fly and have it remain until it was reprogrammed. You never have to replace the chip on the J3 port. You just plug the device on the J3 port, make your changes and hit save. It writes them to the EEPROM on the EEC and you are done. Disconnect your Xilink PLC and cables and store them away with you laptop until the next changes are needed.

You'd still need some way to clear the KAM after making changes. That is one of the things that is such a PITA with the TwEECer. You make a change then have to disconnect the battery and clear the KAM in order to see if the changes are getting you closer to where you need to be.

You can diable the KAM but then you've just tossed out a good tool for letting you know whether you're headed in the right direction.

 

[stang&2Birds]

Maybe we could get Stang&2birds to whip us up a Xilinx PLC array that connects directly from the J3 port to the USB connection on a laptop...

My own custom setup does something similar. But, I yanked the PROM, and soldered in my own home built adapter that has flash memory. I don't trust using the EEC finger connectors in an automotive application. Since my setup is older, it uses serial RS232. But, upgrading to USB isn't that hard with the *current* products.

It's very trivial to do an "EEC memory bus" to EPROM interface. Many high school students could whip one up and get it working in a few days.

But, I know what Brian has gone through with the DCC. So, I don't have any desire to sell my design. Beside, there's no need or reason.

Sorry, I thought everyone already knew about using flash memory on an EEC-IV. See:
http://www.moates.net/product_info.php?cPath=25_35&products_id=101
$60

For software, see:
http://tunerpro.markmansur.com/downloadApp.htm

I know that a couple of people use the moates setup on the tweecer forum.

The disadvantage of using the flash memory adapters is the inability to see the trim values and the other internal values. I started working on adding those abilities, but I lost my notes on where the values are stored. I did a lot of my initial stuff under win95 and a little under windows3.1.


Since my Win95 days, a screwed up disk-backup program nuked my back-up disk and my daily disk. I did backup my stuff onto floppies. But, I wasn't good at marking the floppies since I don't like writing on the floppy covers. I have hundreds and hundreds of floppies. Many floppies are no longer readable.

So, I now use the tweecer RT to do my EEC internal value data logging. I have custom hardware and software to data log the external analog values into / out-of the EEC (very fast and very deep memory) . I also have a few commercial products and I can data log with my "ancient" HP 65000B logic analyzer and ~$2K of option cards. New, my 2 HP logic analyzers and option cards went for well over $100K. Gotta love ebay!