I have been brain storming again and think that I may have found a feasable MFI pump for our beloved TIV 4cyl.

Any interest?

would it happen to be made by MSD???

Brand new MFI assembly...slick

That would be somewhat simple. Get one from a BMW 2002 Tii. They were MFI on a 4 cylinder, and the space cam should come close to what is needed for the type IV.

Mount it where the AC compressor is mounted. Install a cogged gear behind the fan.

Run the throttle cable to the pump and links from there to the butterflys.


I looked into this years ago, before I decided on a /6.


Finding the MFI pump is the problem. The BMW guys don't like to part with them.

Ron Mistak in Sandy Eggo has MFI on his four-cylinder race car (#22). Or had, several years back. It was pretty neat--I think it may have had slide valve throttles; the MFI "injectors" were right at the top of the velocity stacks--no filters. Looked trick as hell, but didn't run very well during the time I saw it. Had some "traditional" 914 problems (vacuum leaks, throttle linkage issues) so it wasn't running full-song.

--DD

Clay, you remind me of someone...

And they cost something like $1500-2000 to rebuild, and if the space cam isn't right NO ONE can help you fabricate a new one.

The pumps from '69 to '79 US Alfa Romeos would be cheaper and more plentiful, and there are versions with space cam suited to 1.8 and 2.0 engines. The main problem here is they have a huge port on the back that's meant to be open to the oil system on the block, which would require fabricating a plate with an oil feed and a return. Again, no help if there's any issues with tuning the space cam.

I like MFI a lot, but it does have some serious shortcomings (like they're all just alpha-N systems, with no load sensing). Now, marrying an MFI pump with full electronic control I'd find VERY interesting. Super high pressure injection for excellent vaporization with no problems handling full sequential injection under all loads. All you need to find is a fast actuator that can position itself to, oh, about 0.005", and can sweep through maybe 0.050" in under 1ms (rough guesses). The rest is easy.

QUOTE (lapuwali @ Jun 22 2005, 08:58 PM)

...marrying an MFI pump with full electronic control I'd find VERY interesting. Super high pressure injection for excellent vaporization with no problems handling full sequential injection under all loads.

sounds like VW TDI 'pumpe-deuse' system.

3000 Bar, computer controlled...

Formula Atlantic and SuperVee ran MFI.

the problem with crankshaft drive is that the pump is timed to the cam, so you'd need a crank drive cog with half the circumferance of the pump drive gear - putting the big gear on the pump could make it interesting to package.

QUOTE (ArtechnikA @ Jun 22 2005, 05:38 PM)

QUOTE (lapuwali @ Jun 22 2005, 08:58 PM) ...marrying an MFI pump with full electronic control I'd find VERY interesting.  Super high pressure injection for excellent vaporization with no problems handling full sequential injection under all loads. sounds like VW TDI 'pumpe-deuse' system. 3000 Bar, computer controlled... Formula Atlantic and SuperVee ran MFI. the problem with crankshaft drive is that the pump is timed to the cam, so you'd need a crank drive cog with half the circumferance of the pump drive gear - putting the big gear on the pump could make it interesting to package.

Crank drive is exactly what Alfa did with their pumps. They operate at "only" 1000psi. 3000 Bar sounds very high. That's 43,500psi, which is getting up into the tensile strength of some common materials (over that of aluminum). Sure you didn't mean 300 bar, or 3000psi? Common-rail diesel systems (aka EFI for diesel) are running about 20,000psi, which is still pretty damned high.

QUOTE (lapuwali @ Jun 22 2005, 10:35 PM)

3000 Bar sounds very high. That's 43,500psi, which is getting up into the tensile strength of some common materials (over that of aluminum). Sure you didn't mean 300 bar, or 3000psi? Common-rail diesel systems (aka EFI for diesel) are running about 20,000psi, which is still pretty damned high.

the first search i did mentioned "2000-3000 Bar" but

This Article quotes "only" 2050 Bar. i think the tuners are pushing that up a bit...

Pumpe Düse is the step beyond common-rail...

The pumps are Kugelfisher and they aren't that hard to find. I did do one of these about 20 years ago, but it was on a type 1. The cams weren't really that difficult, and at that time were several (probably all dead now) people that could do them. One of the best projects I ever did, was converting a TII pump for use on my 13B rotary MG Midget. Now that was fun. Did not start for shit, but once it was warm, it hauled ass. Then there was the tapered roller bearings for the eccentric shaft Blah , blah, blah.... As for Mechanical injection, I bet after you bought a mech pump, and spent the time to get it right, you would spend more than a basic electronic system.

Erik M

QUOTE (redshift @ Jun 22 2005, 06:53 PM)

Clay, you remind me of someone...

Naaaa....

No mullett. And I don't carry a Swiss Army Knife.

THe Alfa pump would be the low buck way to go due it's availability. There where only small changes in the space cam over the years with the '74 2.0 one being the best for Alfa performance. The oiling thing is easy to solve too. Just make a cover for the port in the side and fill the pump with oil. The oil barely circulates with the rest of the engine in the stock set up so I bet if you just changed it when you changes the engine oil there would never be an issue. It you wanted you could just have a high pressure line with a very small restrictor go the the pump and a drainback line. The pumps are very adjustable if you know how. There is a company called Wes Ingrham Inc. that rebuilds the pumps and can make custom cams etc.

I have a V8 though so don't ask me.

QUOTE (avendlerdp @ Jun 22 2005, 08:52 PM)

THe Alfa pump would be the low buck way to go due it's availability. There where only small changes in the space cam over the years with the '74 2.0 one being the best for Alfa performance. The oiling thing is easy to solve too. Just make a cover for the port in the side and fill the pump with oil. The oil barely circulates with the rest of the engine in the stock set up so I bet if you just changed it when you changes the engine oil there would never be an issue. It you wanted you could just have a high pressure line with a very small restrictor go the the pump and a drainback line. The pumps are very adjustable if you know how. There is a company called Wes Ingrham Inc. that rebuilds the pumps and can make custom cams etc. I have a V8 though so don't ask me.

I didn't know Wes could make new cams...Hmm.

No reason you couldn't do MFI with a V8. Alfa did with the Montreal. Two 4-cylinder pumps end to end, with only one logic section.

MFI, electronic or no, would be more expensive than an EFI system (and you don't really have to rebuild anything regularly on the EFI), but it has a number of inherent advantages. EFI has a really hard time doing sequential at anything over a 40% load or thereabouts, unless you use multiple injectors per cylinder and stage them. Vaporization is better with MFI just from the higher pressures involved. Since all gas MFI systems have just been modified diesel systems, and diesel is nearly always direct injection into the cylinder, you can make MFI into gasoline direct injection relatively easily. This brings a number of advantages in operating at the edges of the envelope, and can bring big gains in emissions while retaining wild valve timing.

Ha! I knew that I would get the brains smoking on that one. i currently own a 4 cyl Diesel MFI Mercedes 240D. It has a 4cyl pump and they are stupid easy to find in the junk yards. My favoite shop here in SD (Mototrworks) redoes them for less than probably anyone. You can get gears made if you want. THe resources are out there. It was just a flickering thought that I had.

The problem with using a diesel pump is that a diesel engine will run with a wider variation in mixture. A diesel doesn't need the thermostat or altitude compensator that is on the gasoline version of the pump. The whole concept behind a diesel is that it injects the absolute leanest mixture possible to keep running. That's why most diesels dont' have a butterfly in the intake runners. Just limit the fuel, and it won't accellerate. With a gasoline motor, you have to stay real close to stociometric to get the engine to run without either burning the motor up due to a too lean condition or flooding the motor and washing the oil out with a too rich condition. Even with the Bosch MFI on the 911, the oil change interval has to be very low due to the oil getting contaminated with gasoline.

My thought is to keep the actual injection portion of the pump, but make an electronic control unit to do all the warmup, altitude compensation, and add an O2 sensor to allow it more accurate mixture control. I was thinking a megasquirt modified to run a servo that moves the control rack on the MFI pump.


Thoughts still wandering around my brain, but no forward progress. Maybe after I get my 914 on the road.

QUOTE (ClayPerrine @ Jun 23 2005, 08:54 AM)

My thought is to keep the actual injection portion of the pump, but make an electronic control unit to do all the warmup, altitude compensation, and add an O2 sensor to allow it more accurate mixture control.

have a look at how the TDI works.

very sophisticated electronic control, excellent efficiency, good emissions.
(boost control built in...)

the guys at TDIClub are really up on this stuff.

chances are what you need adready exists.

What about a CIS instead off a Jetta or Golf?

4 cylinders appoximately 1.7 to 2.0, tons of them made and laying in Pick and Pulls all over the country. Volvo's too. I have a High Pressure electric fuel pump and accumulator off a Volvo running my 928 motor. Could this be an option?

Bosch CIS MFI for 78' thru 84' VW GTI's is a proven inexpensive system . It is extremely relable and parts are plentyful.
That's what is on my 914 Turbo GT.

QUOTE (rick 918-S @ Jun 23 2005, 07:08 AM)

What about a CIS instead off a Jetta or Golf? 4 cylinders appoximately 1.7 to 2.0, tons of them made and laying in Pick and Pulls all over the country. Volvo's too. I have a High Pressure electric fuel pump and accumulator off a Volvo running my 928 motor. Could this be an option?

CIS is not the same thing as MFI. CIS is a continuous spray of fuel into the intake runners even when the valves are closed. If you want to see what I am talking about, take one injector out of your car, turn the fuel pump on and lift the airflow flap. You will see that injector (and the other 7 that you can't see) spraying fuel like a fire hose.

MFI is different. It uses a camshaft like an engine cam to spray a high pressure stream of fuel into the cylinder when the valve is open. It is a true sequential injection. It will make more power than a CIS engine, and it doesn't have the drawbacks of the air flow meter flap in the air stream. A CIS engine cannot use and engine camshaft with a radical profile. The flap will ossilate (SP) at low RPMs and cause the mixture control to freak out. MFI doesn't have an air flow measurement, so it doesn't care if the cam has a lot of overlap.

QUOTE (ArtechnikA @ Jun 23 2005, 07:00 AM)

QUOTE (ClayPerrine @ Jun 23 2005, 08:54 AM) My thought is to keep the actual injection portion of the pump, but make an electronic control unit to do all the warmup, altitude compensation, and add an O2 sensor to allow it more accurate mixture control. have a look at how the TDI works. very sophisticated electronic control, excellent efficiency, good emissions. (boost control built in...) the guys at TDIClub are really up on this stuff. chances are what you need adready exists.

The VW TDI is a great design. But I don't think adapting a Diesel injection is the right thing. The TDI uses a small shot of fuel to set off the fuel burn in the cylinder before it injects the main charge of fuel. Plus, it's direct injection. I don't think that would work too well on an air cooled engine. We just can't control the engine temp well enough.


My thought was to retain the factory 911 pump, and make it Look like it's completely stock. I just want to improve the efficency of the MFI system. Reomve all the mechanical controls in the front of the pump, install a crank angle sensor, a tps, a servo to drive the control rack, and any other sensors necessary for the proper operation of the injection. Then run a discrete wire harness out of the pump to a programmable ECU. The advantage is that you retain the MFI's high pressure injection, and the apperance of the factory MFI system. But you get the advantages of the programmable computer controlling it.

This would allow the owner to hook up a laptop, and adjust the virtual space cam on the pump. Adding the O2 sensor also allows the ECU to adjust the mixture for more effeciency on the fly. It's still an Alpha-N system and not a speed density or afm system, so it will never be as effecient as a true EFI system. But a true EFI system will not match the MFI for pure power output. And a EFI system running through one throttle body will never Sound like the 6 MFI stacks when the engine is at full song!


BTW.. CIS is Great for Turbo motors. A turbo motor does not need, and cannot use the cam overlap that a high strung normally aspirated motor thrives on. They are different engineering approaches, and each works great for what it's designed for.

I am not trying to reinvent the wheel. I am just trying to make it better.

QUOTE (ClayPerrine @ Jun 23 2005, 05:22 AM)

QUOTE (rick 918-S @ Jun 23 2005, 07:08 AM) What about a CIS instead off a Jetta or Golf? 4 cylinders appoximately 1.7 to 2.0, tons of them made and laying in Pick and Pulls all over the country. Volvo's too. I have a High Pressure electric fuel pump and accumulator off a Volvo running my 928 motor. Could this be an option? CIS is not the same thing as MFI. CIS is a continuous spray of fuel into the intake runners even when the valves are closed. If you want to see what I am talking about, take one injector out of your car, turn the fuel pump on and lift the airflow flap. You will see that injector (and the other 7 that you can't see) spraying fuel like a fire hose. MFI is different. It uses a camshaft like an engine cam to spray a high pressure stream of fuel into the cylinder when the valve is open. It is a true sequential injection. It will make more power than a CIS engine, and it doesn't have the drawbacks of the air flow meter flap in the air stream. A CIS engine cannot use and engine camshaft with a radical profile. The flap will ossilate (SP) at low RPMs and cause the mixture control to freak out. MFI doesn't have an air flow measurement, so it doesn't care if the cam has a lot of overlap.

I'm not questioning your knowledge just looking for more substance in your theory. I'm certainly no mechanic really. I hate greasy grimmy stuff.

MFI has a HP robbing pump that is engine driven right? How much more HP can you make within the limits of the 2.0 or even a slightly over bored 2.0 if your driving a mechanical pump. Hell and alternator causes a loss in hp.

True the CIS does feed the fuel, but isn't a lean condition a concern with the air cooled motors?

My 928 is a 10/1 stock compression motor with "S" spec cams. It needs all the fuel I can feed the beast.

My thought on CIS is it's compact packaging, no complex pump to try to fit into the confines of the engine bay, no HP robbing pump to drive, and it should work with the stock 914 injection cam profiles and accept a nice mild cam change and more compression.. Just some thoughts...

QUOTE (rick 918-S @ Jun 23 2005, 09:43 AM)

MFI has a HP robbing pump that is engine driven right?

MFI is worth a clear 10-15HP net GAIN.

excellent, sequential fuel distribution, superior atomisation, much more cam selection opportunities, and The Biggie:

No Venturi, no airflow metering cone and flap - nothing in the airstream but the throttles. and - if you're running slidevales - not even that, at WOT.

FSV engines were making about 200 HP from 1,6 liters...

CIS would work fine on a stock or slightly modified Type-IV. It's readily available, and it's easy to install. It will also handle a slightly more radical cam than the stock D-Jet. I am not debating the merits of CIS. It's a good system. And when you get to the CIS-E, you get an even better system.

When Porsche put CIS on the 911, they had to detune the motors to get it to work. Special pistons and cams made for a drop in performance from the MFI versions of the same motors.

The load the MFI pump puts on the motor is very negligible in comparison with the gains made by the ability to put a much more agressive cam in the engine. On the Type-IV, you would be able to put a very radical cam in the engine, and still run injection.

That's the reason that MFI (aka Kugelfischer injection)was used on all the normally aspirated Porsche racing motors until somewhere near the 1990s. Then they went to a computer controlled programmable sequential system.

Your 928 motor was designed for CIS. That's what it will work best with. If you design a motor for MFI, and put CIS on it, it wont' work worth a .

QUOTE (ClayPerrine @ Jun 23 2005, 09:39 AM)

it's direct injection. I don't think that would work too well on an air cooled engine.

i wasn't suggesting direct injection; i don't think it'd be difficult to locate injectors in the ports, a la MFI - or even into the stacks, a la high-butterfly MFI.

QUOTE (ArtechnikA @ Jun 23 2005, 05:55 AM)

QUOTE (rick 918-S @ Jun 23 2005, 09:43 AM) MFI has a HP robbing pump that is engine driven right? MFI is worth a clear 10-15HP net GAIN. excellent, sequential fuel distribution, superior atomisation, much more cam selection opportunities, and The Biggie: No Venturi, no airflow metering cone and flap - nothing in the airstream but the throttles. and - if you're running slidevales - not even that, at WOT. FSV engines were making about 200 HP from 1,6 liters...

Good points well taken. This is interesting to me as I am starting on a 57 Oval with a typeIV.

I do question a 10-15 hp gain by simply in stalling MFI as the only mod though. Our stock FI can't be holding back our little motors that badly...

QUOTE (rick 918-S @ Jun 23 2005, 08:07 AM)

I do question a 10-15 hp gain by simply in stalling MFI as the only mod though. Our stock FI can't be holding back our little motors that badly...

the stock D-Jet was a marvel of it's time. Very effecient and very advanced. It was not the best at performance, though. Just changing the stock injection to Weber downdrafts will gain HP ( and ruin the fuel economy). It has to do with airflow. The stock injection breathes through a relatively small throttle body. The Webers breathe through 4 bigger throttle bodies. Being a very high pressure sequential fuel injection, MFI is more efficent than the Webers, and it breathes through 4 throttle bodies too. 10-15HP is a reasonable expectation of gain from MFI. If you change the cam in the motor, you can get even more HP, and still have a drivable motor.

Clay, you and I are thinking along identical lines (see my earlier posts). I had two Alfas with their version of MFI (very similar to the Bosch pump on the 911 [which, btw, is NOT the same as the Kugelfischer pump used by BMW and on some racing Porsches]).

My thought was first to find a way to measure the movement of the control rack in real time, and log this along with other measured parameters (rpm, throttle angle, engine temp) while still using the stock logic unit. Use that to determine what the first stab at a map should look like, and also characterize how quickly and accurately the actuator on the control rack would need to be.

Just finding the actuator would be the hard part. With no data but some rough guesses, it looked to me like the actuator would have to move relatively quickly to be able to handle stomping on the gas from idle AND it would have to move quite accurately over a pretty small distance. You can find servos that do one or the other pretty easily, but you only find actuators that can do both in scientific and aerospace catalogs with four-figure price tags attached (or did 5 years ago, when I first started looking into this).

Another related angle is to do something similar with CIS. Use the fuel pump, plumbing, injectors, and fuel distributor, but actuate the distributor control plunger with a servo. Drive the servo with speed/density electronics using a MAP sensor and rpm, and you eliminate the major downside of CIS (no wild cams, and the blockage of the flapper), but retain its primary virtues (very simple and reliable, high pressure for good vaporization). This could be tackled in the same way as MFI (measure the plunger movement in real time), but runs into the same problems with servo selection. This would be vastly easier to fit to a Type IV than MFI, since you no longer need to work out a pump drive or an oil feed. You could use individual throttles with this modified CIS to gain the top-end airflow advantages, too. You do lose the sequential operation of MFI.

The VW TDI example may very well provide a servo that meets both needs at a reasonable price. I'll have to look into that.

QUOTE (ClayPerrine @ Jun 23 2005, 06:17 AM)

QUOTE (rick 918-S @ Jun 23 2005, 08:07 AM) I do question a 10-15 hp gain by simply in stalling MFI as the only mod though. Our stock FI can't be holding back our little motors that badly...  :shades: the stock D-Jet was a marvel of it's time. Very effecient and very advanced. It was not the best at performance, though. Just changing the stock injection to Weber downdrafts will gain HP ( and ruin the fuel economy). It has to do with airflow. The stock injection breathes through a relatively small throttle body. The Webers breathe through 4 bigger throttle bodies. Being a very high pressure sequential fuel injection, MFI is more efficent than the Webers, and it breathes through 4 throttle bodies too. 10-15HP is a reasonable expectation of gain from MFI. If you change the cam in the motor, you can get even more HP, and still have a drivable motor.

Ya, That seems to make sense. But how adjustable is MFI? Isn't our stock D-jet set fat on the top end. How does MFI compensate over the range.

I though I read about how most injections systems guys are trying to adapt are too lean. Just more non-sensical questions... I guess they use it on the 911 stuff so the theory and practice is all there and inplace.

This may begin to lool like a thread hijack but . . .

I have Bosch MFI on my 914 too. 2.2 E. Problem is that by definition our systems are'nt stock: intake and exhaust, at least are different than what the system was designed for.

So I've been thinking of using some electronics too, with Megasquirt. My plan is to use a servo motor to replace the warmup regulator. The basic injection and "alpah / n" component would be handled by the pump and the regulating components. The servo would adjust for O2, head temp, cranking mode (maybe) and maybe one or two other things.

Servos work by sending them a positive voltage pulse of between 1 and 2 ms every 20 ms or so. This is very similar to what MS sends to injectors (except MS works by grouding the injectors, as opposed to sending a positive voltage - I think that can be fixed with a simple FET transistor)

My current thinking is that that something like this may be implementable quite easily, perhaps even without any coding. I tried to buy a servo to do a bench top test last weekend, but ended up doing some work around the house instead. May have better luck this weekend. Tomorrow is a holiday, here, so I'll have a little more time.

I think that using something that replaces the warmup regulator has a couple of potential advantages: It is minimally invasive pump-wise, can be undone easily if it does'nt work, would allow for a limp-home mode. It requires a servo that can exert a few pounds of force over 3/8 or 1/2 inch of length. It may be useful to build-in some mechanical leverage to help with resolution and force. From what I saw when playing around with the wideband O2 sensor, the warmup regulator can easily make 3 or 4 percentage points of difference on mixture.

I'll report any progress that I make. My problem is that I can't really write code, so any help in that area would be appreciated, if it becomes necessary.

QUOTE (rick 918-S @ Jun 23 2005, 06:08 AM)

What about a CIS instead off a Jetta or Golf? 4 cylinders appoximately 1.7 to 2.0, tons of them made and laying in Pick and Pulls all over the country. Volvo's too. I have a High Pressure electric fuel pump and accumulator off a Volvo running my 928 motor. Could this be an option?

CIS has been done half a dozen times or more on a Type IV, nothing novel or new.....to me it's boring

nothing looks as sexy as individual volocity stacks sticking up towards the sky inviting the air to fill the engine chambers

QUOTE (lapuwali @ Jun 23 2005, 08:36 AM)

Clay, you and I are thinking along identical lines (see my earlier posts). I had two Alfas with their version of MFI (very similar to the Bosch pump on the 911 [which, btw, is NOT the same as the Kugelfischer pump used by BMW and on some racing Porsches]). My thought was first to find a way to measure the movement of the control rack in real time, and log this along with other measured parameters (rpm, throttle angle, engine temp) while still using the stock logic unit. Use that to determine what the first stab at a map should look like, and also characterize how quickly and accurately the actuator on the control rack would need to be. Just finding the actuator would be the hard part. With no data but some rough guesses, it looked to me like the actuator would have to move relatively quickly to be able to handle stomping on the gas from idle AND it would have to move quite accurately over a pretty small distance. You can find servos that do one or the other pretty easily, but you only find actuators that can do both in scientific and aerospace catalogs with four-figure price tags attached (or did 5 years ago, when I first started looking into this). Another related angle is to do something similar with CIS. Use the fuel pump, plumbing, injectors, and fuel distributor, but actuate the distributor control plunger with a servo. Drive the servo with speed/density electronics using a MAP sensor and rpm, and you eliminate the major downside of CIS (no wild cams, and the blockage of the flapper), but retain its primary virtues (very simple and reliable, high pressure for good vaporization). This could be tackled in the same way as MFI (measure the plunger movement in real time), but runs into the same problems with servo selection. This would be vastly easier to fit to a Type IV than MFI, since you no longer need to work out a pump drive or an oil feed. You could use individual throttles with this modified CIS to gain the top-end airflow advantages, too. You do lose the sequential operation of MFI. The VW TDI example may very well provide a servo that meets both needs at a reasonable price. I'll have to look into that.

James,
I suspect that you are a mechanical engineer. If not, you think like one. I on the other hand am a mechanic and a jack leg engineer. I don't do paper designs. I just start trying to fit things together.

I never did any research on the servo, but I suspect that the MFI mixture control rack could be controlled hydraulically (SP) by using a frequency valve from a CIS-E system, and a servo plunger. The rack is spring loaded to move in one direction,and hydraulic pressure could be used to hold the rack against the spring. The frequency of the valve would bleed off pressure, allowing the rack to move via spring pressure, just like using control pressure in the CIS system to alter the mixture. I suspect that we could use the CIS-E control module to run the MFI pump. It already has most of the sensors that we would require, and it already has both closed loop via the O2 sensor and warmup components packaged into it. We could leave the throttle position sensor mechanical and just replace the warmup and mixture control components.

If you used a megasquirt instead of the CIS-E control box, it only has to drive one injectior, and you can remove the mechanical throttle position sensing.

The only real question I have is... what do I do for control pressure? The electric pump on an MFI system only supplies 10 PSI. The high pressure is supplied by the cam followers when they are compressed.

Waddya think?


BTW.. sitting in a box in my garage is an old MFI pump that I aquired so I could look into this. Let me finish my car and I am going to build it. Then I can put it on my car and try it out.

QUOTE (ClayPerrine @ Jun 23 2005, 07:19 AM)

The only real question I have is... what do I do for control pressure? The electric pump on an MFI system only supplies 10 PSI. The high pressure is supplied by the cam followers when they are compressed.

Not sure why fuel pressure is an issue. Electric pump supplies 10 pounds to the injection pump and the latter takes it to 300 pounds, or whatever, to go through the hard lines to the injectors.

If fuel pressure is an issue, I must be missing something. There is a fuel pressure regulating thingy inside the electric fuel pump, as well as inside the big fuel filter assembly, I believe.

Michel

QUOTE (michel richard @ Jun 23 2005, 09:31 AM)

QUOTE (ClayPerrine @ Jun 23 2005, 07:19 AM) The only real question I have is... what do I do for control pressure?  The electric pump on an MFI system only supplies 10 PSI. The high pressure is supplied by the cam followers when they are compressed. Not sure why fuel pressure is an issue. Electric pump supplies 10 pounds to the injection pump and the latter takes it to 300 pounds, or whatever, to go through the hard lines to the injectors. If fuel pressure is an issue, I must be missing something. There is a fuel pressure regulating thingy inside the electric fuel pump, as well as inside the big fuel filter assembly, I believe. Michel

In a CIS system, the fuel is used as hydraulic fluid for control pressure. The control pressure opposes the Air pushing up the air flow meter. the higher the control pressure, the leaner the mixture. But it takes 70 PSI to run the CIS system due to the control pressure. In an MFI system, the injection pump takes the 10 PIS from the electric pump and when one of the cam lobes in the injection pump comes up, the fuel is pushed out at about 3000 psi. Becasue the pressure is only 10 psi until it's injected, I can't use it for control pressure.

Look what I started This is an interesting read indead.

QUOTE (Dr Evil @ Jun 23 2005, 09:39 AM)

Look what I started This is an interesting read indead.

You better hope you never meet my wife. She will kick your ass for getting me started on this!!!!

Yeah, very interesting. A thanks to the members of the 914Club for being so civil in their conversations. It's topics like these that explode into major arguments and name calling on other sites. Yay 914Club.

QUOTE (ClayPerrine @ Jun 23 2005, 08:49 AM)

QUOTE (Dr Evil @ Jun 23 2005, 09:39 AM) Look what I started  This is an interesting read indead. You better hope you never meet my wife. She will kick your ass for getting me started on this!!!!

So Dr Evil, what is your idea? Obvioulsy number 1 interest level is going to be price with us cheap fuckers. The cool factor is there for sure.

These guys have gone farther than I have thought. My only thought was, "hey, I gat a 4 cyl MFI pump in my MBZ. I wonder if it could be adapted to the TIV?"

Thats about as far as I got. I honestly did not know of all of the little intricacies that have been mentioned like mixture control. I am finding this all very interesting, and as Mark reffered to refreshingly civil.

Thanks for continuing to share your info, fellas.

I like the CIS idea, too.

I'm not a mechanical engineer (not by training, anyway), I'm a software guy. I suppose I think like one, anyway.

I'd considered programmable control of the CIS-E system (there are actually two versions, one early system used, for example, on the SC, and a later version I think was only used by VW). You still need the flapper to give you the primary movement of the plunger. The servo valves didn't actually move the plunger, just altered the control pressure to change the flapper/plunger curve. Lower the control pressure and the flapper can swing more for a given airflow, so enriching the mixture. The early, simple version just used an electronic fuel injector (may even have been a D-Jet injector) as the servo valve, so driving it was trivial. The later version used a more sophisticated actuator whose details I've since forgotten.

I don't know nearly enough about hydraulics to know if you could control the MFI pump rack directly with them with sufficient accuracy. Speed, sure, accuracy I'm not so sure on. You could get lots of accuracy using a geared drive for the pump rack, with the big gear on the servo. 0.001" would be pretty easy to reach, and a wild guess tells me that would be more than enough. However, I don't know if you could move the rack from full closed to full open fast enough when the throttle is moved quickly.

Pure speculation here, but I suppose if you sprung the pump rack so it was normally at full lean, then provided some control pressure (perhaps just plumb into the oil system and use oil pressure) then used two servo valves, one to add pressure, one to release it (both closed to hold the rack at one position), you could get pretty good control. You could also use one valve you make yourself: a rotary valve with drilled passages that either allowed oil in from the pressure side, drained oil from the rack side, or closed it off, then the servo itself wouldn't have to be all that fast or accurate. It just has to spin the valve to one of three positions. The open questions at that point is could full oil pressure pin the rack against the spring to full open with sufficient speed? You'd also want to close the feedback loop by having a way to measure the pump rack's position, so you just add or drain pressure until the rack is in the right place. This would require a pretty high precision method of measuring the rack's position in real time. I'd probably use something like a magnet and a pair of Hall Effect sensors.

Seriously, I'd look at the VW TDI system. There are much more clever engineers than me in the world (I wouldn't have thought of CIS in a bazillion years), and they may have come up with something really simple and adaptable.

In response to an earlier post, I'd also considered doing just electronic control of the "warmup" valve. On the Alfa pumps, this is just a solenoid that only has two positions. Adding a linear actuator to this would be relatively easy. You could then add load-sensing via a MAP sensor, and alter the fueling the space cam provides to some extent (almost certainly enough for typical engine mods) w/o physically altering the space cam itself. This would also help to compensate for space cam wear over time. Temperature compensation with the Alfa pumps was always a big problem with old pumps (used a wax bulb actuator with a Bourdon tube hooked to a coolant passage), enough that many people replaced it with a manual "choke" mechanism to just alter it from the driver's seat.

I've thought about this stuff a lot, and I'm pleased to know there are some other wackos out there thinking of the same thing.

QUOTE (lapuwali @ Jun 23 2005, 11:36 AM)

Pure speculation here, but I suppose if you sprung the pump rack so it was normally at full lean, then provided some control pressure (perhaps just plumb into the oil system and use oil pressure) then used two servo valves, one to add pressure, one to release it (both closed to hold the rack at one position), you could get pretty good control. You could also use one valve you make yourself: a rotary valve with drilled passages that either allowed oil in from the pressure side, drained oil from the rack side, or closed it off, then the servo itself wouldn't have to be all that fast or accurate. It just has to spin the valve to one of three positions. The open questions at that point is could full oil pressure pin the rack against the spring to full open with sufficient speed? You'd also want to close the feedback loop by having a way to measure the pump rack's position, so you just add or drain pressure until the rack is in the right place. This would require a pretty high precision method of measuring the rack's position in real time. I'd probably use something like a magnet and a pair of Hall Effect sensors.

Why do you need a way to measure the rack position? Just use the O2 sensor and temp sensors for feedback. It doesn't have to be as precise as you describe. A fixed mixture curve until the O2 sensor comes online, and then read the O2 sensor to adjust the rack. As for accelleration enrichment, reverse the spring to drive the rack full rich (you would need this for startup anyway)and all you need is to dump the control pressure (i.e. change the frequency of the valve) briefly, and the rack would move via spring pressure. After the initial accelleration, the pressure comes back up, and the mixture returns to normal.

I wonder if the control rack spring can be compressed with 10 psi?

Now I am going to be doodling on the back of a piece of paper the rest of the afternoon.

QUOTE (ClayPerrine @ Jun 23 2005, 07:37 AM)

QUOTE (michel richard @ Jun 23 2005, 09:31 AM) QUOTE (ClayPerrine @ Jun 23 2005, 07:19 AM) The only real question I have is... what do I do for control pressure?  The electric pump on an MFI system only supplies 10 PSI. The high pressure is supplied by the cam followers when they are compressed. Not sure why fuel pressure is an issue. Electric pump supplies 10 pounds to the injection pump and the latter takes it to 300 pounds, or whatever, to go through the hard lines to the injectors. If fuel pressure is an issue, I must be missing something. There is a fuel pressure regulating thingy inside the electric fuel pump, as well as inside the big fuel filter assembly, I believe. Michel In a CIS system, the fuel is used as hydraulic fluid for control pressure. The control pressure opposes the Air pushing up the air flow meter. the higher the control pressure, the leaner the mixture. But it takes 70 PSI to run the CIS system due to the control pressure. In an MFI system, the injection pump takes the 10 PIS from the electric pump and when one of the cam lobes in the injection pump comes up, the fuel is pushed out at about 3000 psi. Becasue the pressure is only 10 psi until it's injected, I can't use it for control pressure.

I was, indeed, missing something (ain't being civil fun !). Using CIS parts sounds very complicated to me. If I understand correctly, you would still be using big chunks of an MFI injection pump, so that the wide availability of CIS parts would not be big help from a system wide perspective i.e. you would still need to find an MFI pump to implement the system. Still, if you meet some success, I'll be very interested to hear about it.

As far as accuracy of rack positionning is concerned I figured the following eyeball estimation in the case where the warmup regulator actuation lever is used: if 3/8th of an inch is 3 points of mixture (from say, 11 to 14) , and you aim to adjust mixture to within 1/10 of a point, you need resolution of about 12 thou. Not sure if that estimation has any relevance when considering a system that acts directly on the main rack.

If you then have 4 to 1 leverage, mechanically, you need approximately 50 thou of resolution. I think that should be achievable.

In any event, I'll try my best case scenario for controlling a servo motor using Megasquirt over the weekend. If I'm successful, I will certainly let you know.

Michel

well this rapidly excellerated way beyond my ability... I nominate this for classic thread status. I need more time to read this stuff than my fast speed reading will allow. Some good and interesting info. Keep it going.

I tried to figure out how I could do this a couple years ago, and I got stuck at trying to figure out where my cam tower is.




M

QUOTE (ClayPerrine @ Jun 23 2005, 09:55 AM)

QUOTE (lapuwali @ Jun 23 2005, 11:36 AM) Pure speculation here, but I suppose if you sprung the pump rack so it was normally at full lean, then provided some control pressure (perhaps just plumb into the oil system and use oil pressure) then used two servo valves, one to add pressure, one to release it (both closed to hold the rack at one position), you could get pretty good control.  You could also use one valve you make yourself: a rotary valve with drilled passages that either allowed oil in from the pressure side, drained oil from the rack side, or closed it off, then the servo itself wouldn't have to be all that fast or accurate.  It just has to spin the valve to one of three positions.  The open questions at that point is could full oil pressure pin the rack against the spring to full open with sufficient speed?  You'd also want to close the feedback loop by having a way to measure the pump rack's position, so you just add or drain pressure until the rack is in the right place.  This would require a pretty high precision method of measuring the rack's position in real time.  I'd probably use something like a magnet and a pair of Hall Effect sensors. Why do you need a way to measure the rack position? Just use the O2 sensor and temp sensors for feedback. It doesn't have to be as precise as you describe. A fixed mixture curve until the O2 sensor comes online, and then read the O2 sensor to adjust the rack. As for accelleration enrichment, reverse the spring to drive the rack full rich (you would need this for startup anyway)and all you need is to dump the control pressure (i.e. change the frequency of the valve) briefly, and the rack would move via spring pressure. After the initial accelleration, the pressure comes back up, and the mixture returns to normal. I wonder if the control rack spring can be compressed with 10 psi? Now I am going to be doodling on the back of a piece of paper the rest of the afternoon.

Response time. If you measure O2, you've now got a delay between the time the rack position affects the mixture and the time you read the new mixture, based on the distance between the exhaust port and the O2 sensor. This means the rack will now be set to too rich, requiring you to reverse the rack, which will overshoot on the other side, and back and forth. Closed loop O2 sensing only "adjusts" the mixture even on OEM systems, it's not the only measure, and it's not the only control mechanism. At WOT, the O2 sensor is completely ignored. Narrow-band sensors also only allow you to control the mixture very close to 14.7:1, which is too lean high engine loads on any engine, let alone an air-cooled engine.

A wideband sensor can be used to fix that, but you still have the delay. Again, this is all speculation on my part. I have no idea if the delay is significant, nor do I have any idea if you'd need the complication of a hydraulic actuator over a simple stepper motor. Until there's actual data on how the pump rack moves in the real world with the real logic section, there's no knowing what's necessary. If you build a measurement device to log what the rack does just to choose an actuator strategy, then you can use the same device to control the actuator once you're doing that.

One thing that's always stopped me from doing this is that it probably wouldn't actually make any difference, or not enough difference to make the effort worthwhile. A staged injector EFI system could be much more easily devised to test if full sequential operation made any real difference under heavy loads (I strongly suspect it doesn't). The vaporization benefits are, I expect, primarily in the realm of emissions, which wouldn't be a bad thing at all, but you'd clean up the emissions dramatically on a Type IV by just fitting a catalytic convertor for a lot less money and effort. It would be a fun project, but it isn't my thing to build something for huge effort that's no better than something else that's cheaper and easier to do. Guess I'm really too much of an engineer that way: always pursuing the best efficiency in all ways, including time and money spent.

Lots of fun to think about, though.

Not sure that MFI will give you a big boost in power...

Dad had a 1973 Alfa 1750 Berlina with the Alfa Spica MFI...

A bug could beat it off the line,

But at about 50, that sucker would blow the doors off almost anything else.....

Loved the line in the handbook .... "Kindly do not exceed 132 mph in 4th gear....."

Then you go for 5th.....


Ken

my dad said he rode in a tricked out austin healey (sp check) and the driver was doing some 170 in it....in 4th gear. getting close to redline and so he reached down and flicked on the electric overdrive....and back down to 2000RPM(UPM in our cars ) ready to go even faster....

I would have crapped my pants.

QUOTE (kwales @ Jun 23 2005, 06:29 PM)

Not sure that MFI will give you a big boost in power... Dad had a 1973 Alfa 1750 Berlina with the Alfa Spica MFI... A bug could beat it off the line, But at about 50, that sucker would blow the doors off almost anything else..... Loved the line in the handbook .... "Kindly do not exceed 132 mph in 4th gear....." Then you go for 5th..... Ken

I think you're misremembering something. The '73 Berlina had a 2000 engine, not a 1750. The 1750 was '69 and '71 only (there was no US '70 model). On both the 1750 and the 2000, switching to Euro cams makes a HUGE difference. The US 1750 only made 105hp, where the Euro version made 130. On the 2000, it was 115 v. 150hp. There's a guy in TX that has a 1750 engine in his car with re-tuned SPICA, Euro cams, and some mild porting, and it makes 165hp at the wheels.

I had a '73 2000 Berlina just 3-4 years ago. Exactly the same drivetrain as the GTV and Spider, just a wheelbase that was 2" longer. Despite it's smog-choked stock tune, it would still out-drag my 1.7. A hot Bug probably would, too...

Wyane Baker ran MFI on his 220hp, 2.2L, t-4 race car.
BMW Kultznswiner (I could be wrong with the speeling )

I think that is the car that Mistak has now--or had, back when I first met him.

Yeah yeah yeah, Kugelfischer Kugelfischer Kugelfischer.

--DD