Finally drove MFI

[winders]

Originally Posted by jluetjen

Just to clarify -- "EFI" covers a wide range of systems, everything from D-Jetronic and L-Jetronic up to the latest systems. Most people are not putting the most recent direct injection on their air-cooled 911s.

I don’t know of anyone putting direct injection on an air-cooled 911. Do you?

I was not talking about throttle response.

I was talking about the idea that EFI does not get fuel into the port in time for combustion to happen when it is supposed to happen. The engine is going to fire the plugs at the right time in the cycle. The EFI has to have the fuel in the port in time. “In time” is all that is required.

For quick throttle response, you must have an ITB setup with the EFI. Fully-mapped Alpha-N systems are not the best unless they are able to compensate for atmospheric conditions, engine conditions, and fuel temperature and pressure.

[Jeff Higgins]

Originally Posted by jluetjen

...

Wow. Fantastic stuff, John. Very clear, understandable explanation of the various systems under discussion, how their various strengths and weaknesses relate to one another, their theoretical advantages/disadvantages and how they relate to what is actually available out there in the real world.

I only have one point on which I would like to differ - the opening pressure for the Bosch MFI injectors used by Porsche is much higher, at 230+ PSI, than you indicate (120 PSI). And yes, this very high pressure does indeed result in better atomization of the fuel, "which results in better combustion and better throttle response and more HP" as you stated.

Modern EFI, as available to us for our applications, still runs at that 30-40 PSI, achieving a lower degree of atomization. And irrespective of the notion that, in some systems, their injectors may be aimed at the back of the intake valve, they do introduce fuel at much lower pressure far higher up, thereby negating any "atomization" the injectors themselves may have achieved. That fuel charge is still largely dependent upon the incoming air to help it along its way into the cylinder after splashing up against many obstacles along the way. Directly into the cylinder, past the intake valve, at five to six times the pressure has notable advantages that we can feel.

[targa72e]

Ok Now that we are back on track lets talk about perception vs reality. Porsche MFI as installed on early 911’s is nothing more than a mechanical computer. So lets break down the actual control elements of both MFI and EFI.
Full disclosure I have worked with MFI on a non-stock application and done the tuning to get it to work correctly. Also did first programable EFI system 20 years ago. That said my day job is as an application engineer for industrial control systems. Basically ECU’s selection and troubleshooting for industrial processes is what I have done for the last 20 years.
So lets look at MFI as delivered from Porsche vs latest EFI sequential system with ITB’s
1: Power generation at a given time (what is throttle response?)
I think we can look at two options. One is how quickly can I rev the engine as in RPM/Second. The fastest rev is going to be the system that makes the most power at a given instance to accelerate the engine thru its rev range.
Second would be in gear (loaded response) same premises how efficient is the engine at making the most power as the engine has changes in throttle.
Since MFI has no way to detect load other than throttle position and Speed the system will be at less than optimum fueling when not under load (generally rich).
MFI system as designed by Porsche assumes that the engine is under some sort of load and fueled appropriately. When not fully loaded runs rich.
EFI ECU in both cases can measure Throttle position, Load (assuming map is installed) and RPM.
2: Design of system:
So all original MFI systems were designed from Porsche for the engine they were going to go on. We can assume they spent a lot of time and money to design the system for the engine they were installed on. For 911 T, E and S each engine had different stack bore configurations, different cams, different compression and different space cams in the MFI pump. Each was designed as a system to work together.
ITB EFI system:
Never sold by factory on common street cars. So all ITB EFI systems are designed by aftermarket. They are either developed by a company, shop or by an individual. Assumption is only or rare occasions does engineering the system get close to factory. Plenty of opportunities for miss matched parts that are not optimized to work together.
3: What does it take to maximize #1.
Fuel, Air and spark
We have to have appropriate fuel, appropriately designed intake and also very important correctly timed ignition.
Fuel has the most forgiveness. Gasoline engines will make the same power over a range of Air. Fuel mixtures (Best torque lean vs rich is from approximately 11.5-13.5 mixture)
Air and sizing of manifolds is important. Lots of things to get wrong. Intake port diameter, intake runner diameter and length all have to be matched to engine displacement and cam shaft to work correctly. Get any of these wrong and you will get less than optimum performance. (not even going to discuss the exhaust side for this discussion).
Ignition. Probably one of the most critical parameters. A few degrees off of optimum can make a huge difference in power and response.
Ok, so lets see how these two computers (mechanical vs electronic) compare on speed.
First neither is instantaneous and both systems have lag. Let compare (warning math involved)
EFI.
So if we assume that we are using a sequential EFI system with ignition control reading a trigger wheel with 60 teeth and the ECU needs to be able to repeatably control ignition timing to with in one degree or less (360 degrees of rotation) and has the capability to run a engine at 8000RPM then we can assume the following. The minimum computational speed would be 360deg x (8000rpm/60sec)=48,000 calculations per sec. Since the ECU actual has to make decisions in that time frame actual execution has to be much faster. So most of the time calculations are happening way faster than needed for the RPM.
MFI.
MFI system is timed to engine since it is driven off of the cam shaft by a timing belt at half crankshaft speed. Internally the MFI pump looks like a little engine with 6 pistons. Like the EFI system it is always operating on past information and correction time is driven by engine speed. The way the pump works is that fuel is drawn into the pump cylinder on the down stroke and expelled to the injectors on the upstroke. The amount of fuel pulled into the metering piston of the pump is based on throttle and RPM (determines position on space cam). Cylinders are variable displacement depending on rack position. So if we take worst case scenario in that the fuel for cylinder #1 was just metered (downstroke) as the throttle starts to move how long until it can it take to recalculate with new throttle and RPM position for fuel metering on next down stroke? This is variable dependent on RPM. So lets look at two cased 1K rpm and 8K rpm. Since the pump is already driven by the cam at half crank speed and only 3 cylinders are on compression stroke each revolution, calculation for 1000 RPM update time (1000rpm x 3)= 3000 injection events every minute/60 = 50 events per second / 6 cylinder so each cylinder update at 8.3 times per second (or every .12 seconds) . At 8000 RPM (8000 rpm x3) = 24,000 injection events every minute / 60 400 events per second/ 6 cylinders so each cylinder updates at 66.6 times per second (or every .015 seconds).
So my take away is:
#1 power generation, EFI ECU should be better able to maximize the power at a given point as it has more information to determine best outputs for power and can update fast (see #3).
#2 Design, MFI (for stock engines) wins. Its very easy to mess this up. Especially if design is bolt on a bunch of parts. Porsche engineering should be able to make a better system, designed to work as a system. You could design a proper system with EFI but you would need to match displacement, cam choice, runner length and size, compression ratio, valve sizes and ports sizes etc to all work together.
#3 Speed. EFI ECU wins hands down. It is orders of magnitude faster the MFI system at being able to update fueling needs.
Feel free to comment on my assumptions.
Next let’s think about fuel delivery from injectors, Injection timing and ignition timing.

john

[Tremelune]

Originally Posted by jluetjen

Throttle response while not in gear is kind of a useless measure.

Well...I would say that throttle response is important for crisp, quick shifts.

Throttle response implies the engine is under load.

I really think there is a confusion of terminology here. I'm not certain I'm in the right, but I've only ever heard people discuss throttle response as a function of a quick-revving engine not under load...for the purpose of shifting.

Response under load is torque, power, and perhaps lag. If you're driving along in a 3.2 (which is not renowned for its throttle response) at 4,500rpm and you floor it, it will immediately leap forward without a hint of lag. I wouldn't describe this as throttle response at all, it just has good power in that part of the rev range.

Whatever term is used, I think it really needs to be stated explicitly what people mean, 'cause this thread was began under a title of throttle response, but noted primarily that it was great how the car rushed forward above 4,000rpm...By this definition, putting 3.2 in a lighter car would somehow improve the "throttle response"...

If nothing else, there are definitely people arguing with each other about different things in this thread...

Edit: Heh, I see I'm not alone: Dumb ? What is throttle response?

[winders]

I look at throttle response from the perspective of a racer. How the throttle responds is critical when dancing at the edge of the friction circle. I prefer the throttle response to be immediate. I don't want to have to predict when the throttle will respond.

I don't really care much about throttle response when the engine is not under load. I can rev match just fine even with a single throttle body EFI setup. My current EFI/ITB setup has immediate throttle under load and with no load.

Yes, lowering weight would improve throttle response under load. To see how this is case, take it to the extreme. Take an engine that has excellent throttle response under load in a 2,000 lbs car. Now put that same engine in a 20,000 lbs car. How is the throttle response under load going to feel? Much more sluggish....

[dannobee]

Even more telling would be to take 10 lbs of weight off of the flywheel. The "throttle response" would improve, but certainly not related to anything fuel related.

[Tremelune]

Originally Posted by dannobee

Even more telling would be to take 10 lbs of weight off of the flywheel. The "throttle response" would improve, but certainly not related to anything fuel related.

Well. Yes. This is precisely what people are talking about when they discuss throttle response.

Sometimes.

I think we need new terminology...

[acme911]

Tremelune,

When talking Throttle response does the actual design and implementation of the mechanical throttle linkage come into play? Both on MFI and EFI. I ask this because after trying out many different marketed linkage kits I ended up designing my own due to what I felt were poor modulation, binding and sticking. In general sloppy. Some of this was related to poor choice of ITB's as well. Which is another whole subject.
Also ended up designing my own intake adapters (I tapered them to provide a smooth transition from custom manifold to ITB's to intake port. ) ((small port 3.0)) ((41mm-34mm))
I honestly do not know how it stacks up to MFI in feel and would welcome a chance to drive a car with MFI.
The most honest feedback I have had was from a renowned local custom Porsche builder. Who really enjoyed the response and feel of my car once I got it dialed in.
Thanks for your explanations.

[winders]

Throttle response = reaction to pedal modulation

[winders]

I use a push/pull (thanks, Carroll Smith) cable throttle setup with my AT Power ITBs. I find the cable setup to be more direct and less sloppy than the Porsche rod linkage system. The AT Power cable throttle hardware has a cam integrated into the design that opens the butterflies more slowly at first and ramps up the opening speed as the throttle pedal is pushed down. DBW setups should feel just as responsive as the cable setup.

The AT Power ITBs are direct to head units.

[panzerfaust]

Originally Posted by targa72e

Ok Now that we are back on track lets talk about perception vs reality. Porsche MFI as installed on early 911’s is nothing more than a mechanical computer. So lets break down the actual control elements of both MFI and EFI.
Full disclosure I have worked with MFI on a non-stock application and done the tuning to get it to work correctly. Also did first programable EFI system 20 years ago. That said my day job is as an application engineer for industrial control systems. Basically ECU’s selection and troubleshooting for industrial processes is what I have done for the last 20 years.
So lets look at MFI as delivered from Porsche vs latest EFI sequential system with ITB’s
1: Power generation at a given time (what is throttle response?)
I think we can look at two options. One is how quickly can I rev the engine as in RPM/Second. The fastest rev is going to be the system that makes the most power at a given instance to accelerate the engine thru its rev range.
Second would be in gear (loaded response) same premises how efficient is the engine at making the most power as the engine has changes in throttle.
Since MFI has no way to detect load other than throttle position and Speed the system will be at less than optimum fueling when not under load (generally rich).
MFI system as designed by Porsche assumes that the engine is under some sort of load and fueled appropriately. When not fully loaded runs rich.
EFI ECU in both cases can measure Throttle position, Load (assuming map is installed) and RPM.
2: Design of system:
So all original MFI systems were designed from Porsche for the engine they were going to go on. We can assume they spent a lot of time and money to design the system for the engine they were installed on. For 911 T, E and S each engine had different stack bore configurations, different cams, different compression and different space cams in the MFI pump. Each was designed as a system to work together.
ITB EFI system:
Never sold by factory on common street cars. So all ITB EFI systems are designed by aftermarket. They are either developed by a company, shop or by an individual. Assumption is only or rare occasions does engineering the system get close to factory. Plenty of opportunities for miss matched parts that are not optimized to work together.
3: What does it take to maximize #1.
Fuel, Air and spark
We have to have appropriate fuel, appropriately designed intake and also very important correctly timed ignition.
Fuel has the most forgiveness. Gasoline engines will make the same power over a range of Air. Fuel mixtures (Best torque lean vs rich is from approximately 11.5-13.5 mixture)
Air and sizing of manifolds is important. Lots of things to get wrong. Intake port diameter, intake runner diameter and length all have to be matched to engine displacement and cam shaft to work correctly. Get any of these wrong and you will get less than optimum performance. (not even going to discuss the exhaust side for this discussion).
Ignition. Probably one of the most critical parameters. A few degrees off of optimum can make a huge difference in power and response.
Ok, so lets see how these two computers (mechanical vs electronic) compare on speed.
First neither is instantaneous and both systems have lag. Let compare (warning math involved)
EFI.
So if we assume that we are using a sequential EFI system with ignition control reading a trigger wheel with 60 teeth and the ECU needs to be able to repeatably control ignition timing to with in one degree or less (360 degrees of rotation) and has the capability to run a engine at 8000RPM then we can assume the following. The minimum computational speed would be 360deg x (8000rpm/60sec)=48,000 calculations per sec. Since the ECU actual has to make decisions in that time frame actual execution has to be much faster. So most of the time calculations are happening way faster than needed for the RPM.
MFI.
MFI system is timed to engine since it is driven off of the cam shaft by a timing belt at half crankshaft speed. Internally the MFI pump looks like a little engine with 6 pistons. Like the EFI system it is always operating on past information and correction time is driven by engine speed. The way the pump works is that fuel is drawn into the pump cylinder on the down stroke and expelled to the injectors on the upstroke. The amount of fuel pulled into the metering piston of the pump is based on throttle and RPM (determines position on space cam). Cylinders are variable displacement depending on rack position. So if we take worst case scenario in that the fuel for cylinder #1 was just metered (downstroke) as the throttle starts to move how long until it can it take to recalculate with new throttle and RPM position for fuel metering on next down stroke? This is variable dependent on RPM. So lets look at two cased 1K rpm and 8K rpm. Since the pump is already driven by the cam at half crank speed and only 3 cylinders are on compression stroke each revolution, calculation for 1000 RPM update time (1000rpm x 3)= 3000 injection events every minute/60 = 50 events per second / 6 cylinder so each cylinder update at 8.3 times per second (or every .12 seconds) . At 8000 RPM (8000 rpm x3) = 24,000 injection events every minute / 60 400 events per second/ 6 cylinders so each cylinder updates at 66.6 times per second (or every .015 seconds).
So my take away is:
#1 power generation, EFI ECU should be better able to maximize the power at a given point as it has more information to determine best outputs for power and can update fast (see #3).
#2 Design, MFI (for stock engines) wins. Its very easy to mess this up. Especially if design is bolt on a bunch of parts. Porsche engineering should be able to make a better system, designed to work as a system. You could design a proper system with EFI but you would need to match displacement, cam choice, runner length and size, compression ratio, valve sizes and ports sizes etc to all work together.
#3 Speed. EFI ECU wins hands down. It is orders of magnitude faster the MFI system at being able to update fueling needs.
Feel free to comment on my assumptions.
Next let’s think about fuel delivery from injectors, Injection timing and ignition timing.

john

That’s the key isn’t ? The MFI can’t determine load and the ignition can’t adjust for it. There isn’t an acceleration pump circuit like a carbs or snap-WOT enrichment program into the EFI mapping to excesses intake charge. There also is isn’t that compensating draw you get with carb venturi’s. So MFI are set up comprised to run rich.

Who here has a MFI car set up to run stoic at cruise mode please raise their hands ? ��That’s why they tend to have poor fuel economy and gum up plugs faster than cis,motronic and properly tuned stand alone. But the boon is great throttle response and power.

On the flip side yes the MFI throttle response is immediate intoxicating for a 50yr car. I do believe the high nozzle pressures (80-100psi?) aid in combustion swirl and have self cleaning properties. But these attributes are nothing a modern EFI can’t overcome or easily exceed in which they can predict the load with a TPS, AFM, RPM, Temp, barometric, atmospheric, timing, knock sensing etc.. please keep in mind modern 500ho plus supercars program there fly-by wire for a smooth tip-in and to keep it from crab walking sideways with the TC off like an old viper in the rain on bald tires.

Having said that, I have a MFI fetish. For its looks, sound, response and it’s prestigious motorsports link, it’s hard not to be smitten. I love conceptual simplicity that’s it’s just a diesel like jerk pump. At the same token I’m in awe the complexity of logic circuit on how this mechanical computer is able meters fuel in various conditions Ive owned at least one MFI car for over 35yrs and have driven itb butterfiles, slide throttles and even rotary barrel throttles for reference. Too me there is nothing more sexy than a high butterfly MFI car but it hasn’t clouded my judgment.

My 2 cents

[47silver]

I would think that a 225# fuel pump and injectors that could handle that pressure would be the way to go with EFI.
The only advantages to MFI are the fuel atomization is better and the fact that once it is set up it is bullet proof. And you do not need a laptop, any special software, wires and cable. A few simple tools and an a/f mixture guage and you are done.

[jluetjen]

Originally Posted by 47silver

I would think that a 225# fuel pump and injectors that could handle that pressure would be the way to go with EFI.

Easier said then done. My understanding is that the technical obstacle that that prevented this from being done was the ability to control injectors at that pressure in a timely fashion. This is where the somewhat recent development of reliable piezoelectric injectors was the breakthrough that allowed modern high pressure injection systems.

[jluetjen]

Originally Posted by Tremelune

If nothing else, there are definitely people arguing with each other about different things in this thread...

Edit: Heh, I see I'm not alone: Dumb ? What is throttle response?

Guilty as charged!

[Jonny042]

Originally Posted by jluetjen

Easier said then done. My understanding is that the technical obstacle that that prevented this from being done was the ability to control injectors at that pressure in a timely fashion. This is where the somewhat recent development of reliable piezoelectric injectors was the breakthrough that allowed modern high pressure injection systems.

Modern direct injection systems have high pressure fuel pumps..... usually driven off the camshaft!!!!!!

[jpnovak]

Originally Posted by Jonny042

Modern direct injection systems have high pressure fuel pumps..... usually driven off the camshaft!!!!!!

So... Jonny you are saying that we have have come full circle? Isn't that the definition of logical thinking.

[Jeff Higgins]

Originally Posted by Jonny042

Modern direct injection systems have high pressure fuel pumps..... usually driven off the camshaft!!!!!!

Originally Posted by jpnovak

So... Jonny you are saying that we have have come full circle? Isn't that the definition of logical thinking.

"Everything that was old is new again"...

But, yes, few would argue the demonstrated advantages of direct injection and its superior atomization. Combined with the fidelity engendered through modern electronic controls, we truly have "the best of both worlds".

[winders]

Originally Posted by jpnovak

So... Jonny you are saying that we have have come full circle? Isn't that the definition of logical thinking.

OMG!

Hardly. MFI does not use direct injection. It still injects fuel outside the combustion chamber in the intake port. You guys can't equate MFI with DI just because both use higher fuel pressure than modern port EFI on ITBs. DI fuel pressures are 10 times higher than MFI. MFI is only 5 times higher than EFI systems.

The ONLY reasons to use MFI is to be period correct or because you want to. Modern EFI/ITB setups are significantly better than MFI.

[winders]

Hell, DI systems have more in common with EFI port injection systems than they do with MFI.....

[jluetjen]

Originally Posted by jpnovak

So... Jonny you are saying that we have have come full circle? Isn't that the definition of logical thinking.

Keep in mind that the pump that you're referring to just pressurizes the fuel rails. The actual injection pulses are controlled by the injectors.

There are no fuel rails in an MFI system. In an MFI system, the increasing pressure resulting from the plunger stroke in the pump eventually overcomes the spring pressure in the injector and pushes the injector valve open.

So conceptually the systems work differently.