MFI Delivery Rates
 

Hi

Has anyone ever measured/graphed the delivery rate of an MFI pump versus the Rack position?

If you have are you willing to share these measurements?


TIA

Neven

Neven,

The basic reason no one goes this far with MFI is that there isn’t anything you can do about it. The MFI “map” is cast in stone … well, steel. Short of regrinding the “space cam” there is not much way to tweak it other than the normal adjustments.

If you want to go that direction, EFI is the way to go. You can sit there with your laptop, engine dyno, and sensors. You can re-map it any which way you want.

You can run a MFI pump on a carbureted engine on the dyno and measure the flow vs. rack position, recording the number of revolutions and volume of fuel. Considering where I think you want to go, this is an important measurement. You can build a bench test also.

Best,
Grady

Grady

I think Henry might have beause I've seen a picture of a pump he's modified and pulled out all the volumetric correction. As an aside the 'racing' MFI pumps had no volumetric controls. The reason I ask is that I'm going to pull out the control circuit and control the rack with a servo driven by a megasquirt, an 'eMFI' system.

You might wonder why I'd bother, The MFI system has great fuel delivery (200 psi atomisation, sequential, independant of manifold pressure) , lousy control (levers, cams etc). If I do this I'll be able to 'tweek' the MFI system with the same accuracy of an EFI system, If I re cam the engine, I can remap the megasquirt. I'm also going to permanently install an LM-1 AFR
and run closed loop

Neven

Very nice idea to replace the space cam with an electric servo motor + control electronics. Are you measuring intake vacuum also?
If interested, I have access to the rebuild manuals and calibration procedures, but only in september. I can send you photocopies if needed. I like your project.

Andreas

Andreas

I have the factory stuff but that is all based in the space cam calibration, Whatl I really need is:

1/ Is the delivery rate linear to rack position
2/ What is the delta in delivery rate linear to rack position

The megasquirt uses a map sensor (intake vaccum)
Other inputs will be
1/ RPM
2/ TPS (Throttle Position, I'll put in inside the pump
3/ 'Coolant Temp', (again a sensor inside the pump)

If you are interested I'd suggest you read the megasquirt info

Neven

You could most likely measure this pretty quickly with some graduated cylinder(s). Suspend the injectors over the cylinders and turn the pump over manually 5 or 10 times until you have sufficient volume to measure in the graduated cylinders. Repeat with the rack full in, full out, and then some increments in between.

The pump doesn't even need to be attached to the engine, as long as it has sufficient fuel supplied to it. Someone should confirm this for me, but you might want to use some alternative and less flamible liquid then gas while you are doing these experiments.

Nelven,

I have a 2.2 E with MFI in my car, plus a megasquirt computer I assembled for another project, last year.

Megasquirt II will have stepper motor control (as you undoubtedly know), so I've been thinking along the same lines. Actually, I was thinking more in terms of keeping the MFI pump pretty much as it is and using the megasquirt to drive a stepper motor to act on the lever that the MFI thermostat normally works on.

With Megasquirt II, I figure that scheme is implementable by changing some code alone. Would be nice to have wide band O2 sensing too ! And then use all the unused computer power to work the ignition maybe.

My problem is, I can't really write code, although I might have a go at doing it in C, again with the new Megasquirt II. Can't wait for the new hardware to become available

Michel Richard

John

Yes, But its not quite that simple, fuel delivery is measured per
1000 revs, A bosch test bench does this automatically, I will get a FI tech to do this if I have to I just thought there may have been someone who'd already done this

Neven

Michel

You are on the same page! I thought of a stepper originally, The problem with stepper is that they move relatively, ie they can 'get lost' if they dont pass thru a reference point. So i decided to a futaba servo with a linear converter (see http://www.emsjomar.com/linear.aspx) The code is so
basic as to be trivial instead of triggering the injector, (timed to the points, for a duration) you simply output a pulse between 1 & 2ms long every 10ms, I dont think it would be more than a couple hour work

Neven

Neven,

I have'nt looked at my pump closely enough to figure out how to set something up to act directly on the main rack. Are you thinking of splitting the pump in two and removing the whole governor section ?

As I said, trivial and code don't go together well in my vocabulary, so any help from you in that regard would be very welcome. I would love to make a contribution to your effort in return.

I understand that the 3.0 SC/RS race cars used some sort of Kugelfishcer pump with electronic controls on it. Do you have info on that ?

I've only recently gotten my engine to run fairly well, so I'm not too keen on taking the pump off to measure flows.

The other poosibility I have dreamed up is to get rid of the pump entirely and fabricate short individual runners to fit between the heads and the MFI throttle bodies. The runners would have provision for installing efi injectors as well as a small port used to measure intake air pressure.

I have made drawings for the runners and I have a small machine shop in my basement where I could fabricate them fairly easily. Actually, that could be my contribution, if you need a part or two fabricated, let me know.

In all cases, head temp would be measured through some of those small wire contraptions that are screwed under a spark plug.

Michel

Neven,

I'm not sure the Futaba servo would be able to develop the force necessary to reliably move and hold the rack or that it could sustain the temperatures of an engine compartment. The pump does get pretty hot, and a servo screwed to it would get very hot too.

I would investigate that General Motors IAC motors. If they can develop the force required they would undoubtedly be much more reliable. Making a fitting to fit them in place of the MFI thermostat would be trivial.

Michel

Funny how things happen, I was thinking about this very same thing 2 days ago. Except it would be for boost application/correction. the Gm IAC was my first choice. You guys stole my idea!

Aaron

Michel

You'd be surprised about the futaba servos, go to http://www.futaba-rc.com/servos/futm0029.html, They
position with a very high degree of accuracy, can generate up to
194 ounce/inches of torque and are designed for hostile enviroments. (RC Cars and Helicopters). I've bought a couple of
pumps on Ebay and was going to gut the control portion, Ill then
design somewhere to mount the servo

A std servo with the linear converter generates 1.5Kg of force,
with a high power one that would be over 6Kg. The should be very little force required to move the rack, its only twisting the
pistons of the pump

Best part, the servo costs about $24USD with the linear converter.

If you go full EFI you lose all the MFI advantages. As for
measuring the manifold vacuum, Its already there, tap into the
vacuum advance pipe to the distrubutor.

Neven

Michel

I just had another thought, It is easy to determine the required rack positioning force as the racks are 'passive return' via a spring
Therefore the force of that spring is in excess of the required positioning force

Neven

Neven,

You are right, I am surprised by the Futama servos. I'll look into them some more.

As far as MFI is concerned, the only advantage I can see that will remain if the whole governor section is removed will be the high pressure of injection. I must say I've always thought that was a bit overated as an advantage.

My guess is that if the factory used electronic/MFI injection on the early eighties SC/RS cars, it's because they were getting more power than was available with the then available K-jet or electronic systems. I believe the factory racing teams have now moved to all-electronic, which would mean that current electronic systems are better.

Still, it's certainly an interesting project, and I'm actively looking into it. My 2.2 E is in a 914, which means I have neither stock 911 intake or exhaust . . . and means that I can't expect the MFI to run perfectly well.

As I saidc, if you can help me with some of the code stuff, I'll be happy to machine some parts, if you need help in that area.

Cheers,

Michel

Michel

Thanks, drop me an email off list neven@mwk.co.nz and I'll keep you informed. I personally think the MFI pump has great advantage, Sequential (MS isn't), High Pressure, Absolute Metering (doesn't require a constant pressure differential).

This is the main advantage of electronic control, these things alter your VE and therefore its just a matter of tweeking the VE table in the MS to get it running sweet

Look for an MFI pump on ebay should only cost $100-$200

Neven

It sounds interesting, but why not just use EFI with the MFI stacks?

The MFI pump costs horsepower and the drive belt can come loose.

Also, the rack controls fuel delivery to all cylinders equally, you don't have the ability to give individual cylinders more fuel in response to higher CHTs.

Cool idea, though!

EFI with the MFI stacks is a possibility. But The Megasquirt system currently available does'nt do sequential injection and EFI injectors don't use the sort of pressure that MFI does.

The idea is to try to devellop electronically controled MFI, by replacing the stock mechanical governor system by a computer and a servo motor.

Megasquirt, which is what we want to use because it's open source, won't let us customize injection for each cylinder.

The coolness factor, at least for me, is certainly a factor.

Michel Richard

John

I'm not really doing this as an all out racing solution. I'm
currently restoring a '73E. So if I get this going then I can
retain a 'stock' look and get added performance & economy.

Having said that I'm also interested to see how it goes. Electromechanical injectors are not perfect, They have an opening delay and can only deliver fuel at a constant rate.

The main limitation of the MFI pump is its fixed VE 'map' so this removes that and allows for me to 'play' with intake, exhaust and cams and still get an optimal afr

I'd ask the question, why do so many of you guys build 2.7/2.8 MFI Motors?

Neven

I've also had that thought.

If this can be made to work, MFI becomes available for a wide variety of engines.

Michel