Using MS to control CIS/K-Jetronic FI systems

[R2.0]

Over at Shop Talk Forums we are discussing the possibility of using MS as a controller for a CIS or CIS-E system. One guy already is using some rigged electronics to control a CIS-E frequency valve, but he wants to move to MS control down the road. My ideas are more ambitious: get rid of the CIS air meter component and just use the fuel distributor. This is mainly to allow mounting of a CIS system on air cooled VW's, which are space challenged. Here are my thoughts:

First, get rid of the air meter. The highly calibrated cone, plate, and arm need to be replaced by something. I was thinking a simple spring. This would allow all flow adjustments to be made on the other side of the fuel distributor. This would be accomplished by:

1) Regular CIS uses a piston controlled by a Control Pressure Regulator to trim fuel flow for warmup, etc. It routes fuel pressure to the back side of the piston and modulates it - it it reduces pressure on he back side, more fuel flow through the distributor and the mixture is enriched. So one option is replace the control pressure regulator with a PWM or other variable valve driven by the MS. The "VE" map would basically just be a duty cycle map for the valve. This would translate to pressure on the piston, which, counterbalanced by the spring on the other side, would control flow to the injectors.

2) CIS-E uses a PWM frequency valve in place of the piston in basic CIS. It is uses to trim injector flow in response to an O2 valve, via a controller. So if the controller is replaced by MS, it could control the mixture directly.

Questions:
1) Is the frequency valve capable of full range flow adjustment?
2) Anybody got an appropriate PWM valve for controlling regular CIS?
3) Does anyone know the forces applied to the piston in the fuel distributor? Or have a reference?

Open for comment.

[ModMech]

The Bosch CIS system is a constant injection system. The quantity of fuel is MOSTLY determined by the air meter/fuel distributor, and the Lambda (freq valve) was added to later systems only for fine tuning the AFR.

It is my understanding that the CIS injectors offer superior atomization over even the newest EFI injectors. Therefore it may be *possible* to use very low pressures (and resultant low flows), if the freq valve will produce them, and maintain acceptable atomization.

[lapuwali]

I investigated this very problem several years ago. The main problem with controlling the distributor plunger directly is trying to find a servo that would move both fast enough and precisely enough that didn't cost $4000. You can get fast, you can get precise, but both together require lab-grade stuff that was not only really expensive, but probably not up to the rigors of a hot, vibrating engine bay.

There are actually two "frequency valve" controller setups on K-Jet. Both vary the control pressure (effectively altering the resistance offered by the air flapper to air flow, and thus the airflow to fuel curve). The earliest just used a fuel injector as a PWM valve. The later systems used a more sophisticated servo valve. Both systems, however, require the airflow meter remain in place. They don't control the plunger directly. I'm not certain a PWM valve and a spring are sufficient, but it's worth trying.

If you want to experiment, the place I'd start is to set up a good measurement system that doesn't restrict flapper movement, to measure plunger movement. Using a magnet on the meter arm and a pair of Hall sensors differentially should be sufficient. This will give you an idea of how far and how fast the plunger needs to move, and if you combine logging of this with logging a few other variables, like throttle position, RPM, and manifold pressure, you should get a good curve to start with for an initial map. You'll also get the data you need to design an actuation mechanism for the plunger.

Best of luck. I thought it was a fabulous idea then, and still do, but it was beyond my abilities.

[R2.0]

I'm home for the holidays and I dug up my Bosch FI book by Probst. The "frequency valve setup actually does control the plunger position, but indirectly. It alters the control pressure, which moves the plunger a small amount to trim the AFR.

KE jetronic doesn't regulate control pressure at all. It uses an electromagnet to drive a valve with a variable milliamp signal. In addition, it has an o-ring in the fuel distributor to provide a bottom stop for the piston. It also has a potentiometer on the AFM that could be used for your logging idea.

I think the K-Lambda, with the frequency valve will be the one to play with. One simple test would be a static test: fire up the fuel pump and measure the force required to move the plunger. Then, figure out the deltaP between system pressure and control pressure to move the plunger full range. Then it becomes a pressure regulation problem.

[T3Bunny]

Hmmm... I have played with a few cis systems and can think of another possible and dirt simple method.

Why not create a simple manifold for the fuel supply and have a PWM valve feeding this? Then it wolud be a PIS system!

Yeah, I know cis works so well due to its atomization as its runs continously. But I bet if you insreace the fuel pressures and have an accumalator attatched to the manifold that it would reatain enough pressure to never fully turn the fuel flow off to the injectors.

Best part is you could build a setup to test for SUPER cheap. Here is a cool place with lots of valves that could well perform this job and not break the pocketbook doing it! http://www.stcvalve.com/I-Solenoid.htm#0

[chois]

I was considering a simple approach.

Let the airflow plate make the 'macro' fueling adjustments, and use MS to control the DPR for 'micro' fueling adjustments based on where you are in the VE tables. We just need to produce a 0-20mA signal to send to the DPR.

I know that this leaves the big nasty air plate in the system, but I am really just looking for a better way to tweak fueling than sticking a resistor on the coolant temp sensor (and the ignition control will be nice too).

Thoughts?

[T3Bunny]

Honestly I am still for getting rid of the airplate all together. Its the biggest restrictor in the CIS setup. I am sure somebody more electronics savvy than me could creat the 0-20ma setup your thinking about. In fact I know that there is actualy a programable setup out there that does this. I think if one were to go MS with a CIS car, do it all the way. Figure out how to get the MS controlling them CIS injectors.

The setup doesn't have to be hard even. Simply gut the air-vane and some other parts out of there, then I would retain the fuel manifold with possibly some bits inside removed. PWM a solenoind feeding it and your pretty much done. Plus your retaining the look of having a CIS setup in the car.

[hobieboy]

A tangential comment...

With CIS, you can't really control the amount of fuel going into individual cylinders. Doesn't that somewhat defeat purpose of EFI?

[lapuwali]

No, L-Jet doesn't do this, either, nor does MS. Even full sequential systems usually don't control fuel to the cylinders individually, they just calculate a value for the whole engine and divide by the number of cylinders (conceptually).

To get per cylinder fuel control, you need to measure each cylinder individually, either using an O2 sensor per cylinder, or some other measurement method.

This is still on the drawing board even for OEM systems.

As for controlling CIS without the plate. You can simply use a MAF sensor in place of the plate. However, you still need some way of controlling the plunger through it's whole range of movement, which leads you straight to needing a fast, accurate servo with the appropriate stroke, which costs big money.

The CIS injectors are 70psi pop-off valves, and the system pressure is roughly 100psi. So, the injector pressure is between 70-100psi. This is better than the 45psi used by most EFI injectors, but it's not a gigantic improvement. Direct injection systems use far higher pressure. You could also get much higher pressure by electronically controlling Bosch MFI or SPICA mechanical injection, which use system pressures in the 1000psi range. They're also genuinely sequential at all engine speeds. You still need a fast, accurate servo (again), this time to replace the logic section on these units.

[oughtsix]

I am just finishing up replacing the CIS injection system on my 1980 mercedes 450slc with a megasquirt setup. Good ridance to the CIS! The hardest part was crafting the fuel rails and fuel plumbing. I was able to turn down the lower diameter of Ford 19lb injectors a couple thousands of an inch on my lathe and they fit perfectly in the CIS pockets. I don't know what CIS looks like on a VW as far a injector mounting but I have never been happier with my Mercedes!

[T3Bunny]

As for controlling CIS without the plate. You can simply use a MAF sensor in place of the plate. However, you still need some way of controlling the plunger through it's whole range of movement, which leads you straight to needing a fast, accurate servo with the appropriate stroke, which costs big money..

Okay, a few holes here. And the WHOLE point of my idea is being missed. First WHY use a MAF sensor? Use a standard MAP like in the MS. A MAF is going to be a pain to install and be obvios, pluss its still a restriction.

Second, rip out the plunger and forget about the servo. They are not needed. What you would do is simply use the fuel distributer as bascialy a fuel manifold. Have it fed by ONE solenoid that you PWM like a standard fuel injector. If your going to retain the CIS system for the stealth look or for the atomization of the injectors, then retain it. Simplify it. Why make it hard with control algorythums when everything you need is already there?! Just use a solenoid to modulate the feed pressure.

Granted, this is all theoretical. BUT, I would spend the 30-50 for experimenting with a valbe, that can be made to work without code changes, or makeing compleatly new code, and all that. As opposed to a system that does NOTHING too simplify or address the issues of the CIS.

Take tha advantages of both systems, and combine them. GET RID of the disadvantages of the CIS. Mainly its SUPER restrictive airplate. TOSS IT OUT! Use the better atomizing CIS injectors. If the concept above still isn't making sense, take a look at the Aquamist Water Injection systems with multiple jets and PWM injection contol. Its super simple and can likely be applied to this concept. The only tricky issue I see is matching the flowrate of the solenoid used for fine idle control. One might have to use two solenoids and say the Roger Enns staging control to use a smaller solenoid for idle.

As for the pop off of the CIS injectors, I don't know if Benz runs theirs higher, bu the Bosch injection manual says between 45 and 50 psi is the pop off. And I have found that they will frequantly pop off and start spraying as low as 35psi.

IF, and this is a BIG if, somebody were to go through the trouble of doing this weird conversion.... Keep it simple. Forget about expensive servos and components controlling a bad setup. Get rid of stuff like that. Keep it looking as stock as possible... Get the MS to do what it already can do. PWM control solenoids. Sure tuning will be entertaining until you get the hang of things. But the headaches for trying to create new code for a servo won't be there.

I can't think of any reasons anyone would want to retain the look of a CIS system though

[chois]

I appreciate what you are saying, and agree that the air flow plate is a restriction. This is why I am putting MS on my CIS-E Motronic 16 valve.

However I have another application that I would like create a 'programmable' CIS-E system, maintaining the airflow plate. It is for a competition engine that must retain 'CIS type' injection. If we can simply have discrete control of the fuel it will help a lot, we are not airflow limited at the plate on this application as it is a small engine. Is there anyone that can help me with this, or anyone you would recommend I approach?

thanks,

[T3Bunny]

Do a little research, and also look up Peter Tong on the VW vortex. I know he is running some sort of fully programable setup on his own CIS car.

And I would still use a MS to control the spark. Its deffinatly a possibility to still use the MS to control a later motronic style regulator, I am not sure what it would involve though.

[chois]

No surprise that Peter Tong came up, he really seems to have a soft spot for CIS sysetms. If anyone stumbles across this thread and thinks they can help me generate a 0-20mA signal to feed the DPR based on VE table location, please contact me.

Just to be clear on my 16v there is no way I am leaving the CIS fuel delivery hardware in place. It will be a 'typical' Megasquirt conversion.

[R2.0]

A tangential comment...

With CIS, you can't really control the amount of fuel going into individual cylinders. Doesn't that somewhat defeat purpose of EFI?

That's not exactly true. On some (most?) CIS fuel distributors there is a trim screw that will adjust pressure for each injector. Theoretically, they are used with EGT probes (or, for air cooled, CHT sensors), to trim individial cylinders. But it isn't dynamic, and from all reports one can seriously screw up your system bymessing with these arbitrarily.

R2.0

[R2.0]

As for controlling CIS without the plate. You can simply use a MAF sensor in place of the plate. However, you still need some way of controlling the plunger through it's whole range of movement, which leads you straight to needing a fast, accurate servo with the appropriate stroke, which costs big money..

Second, rip out the plunger and forget about the servo. They are not needed. What you would do is simply use the fuel distributer as bascialy a fuel manifold. Have it fed by ONE solenoid that you PWM like a standard fuel injector. If your going to retain the CIS system for the stealth look or for the atomization of the injectors, then retain it. Simplify it. Why make it hard with control algorythums when everything you need is already there?! Just use a solenoid to modulate the feed pressure.

A couple of things. First, I don't think a servo is the answer either. On Cis-E, the plunger is controlled by the frequency valve, which looks and acts suspiciously like a regular Bosch injector. So I think its' possible to use MS and a frequency valve or injector to do full range fuel control.

Second, the idea of using one big fat injector to feed the CIS distributor is intrigueing. 3 questions:

1) Flow - That's a pretty damned big injector. Adapt a Ford CFI?
2) Pressure Pulsations - at low flow, the injector will be off most of the time, so we are losing the advantages of the "continuous" aspect of CIS.
3) Total Pressure - While CIS injectors pop at pressures in the normal operating range of a CFI injector (I think), they don't necessarily like to *operate* at those pressure. How will a CFI injector take to operating at standard CIS pressures.

One solution would be to use 2 or 4 smaller injectors manifolded together, and use both banks alternating to trigger them. This would smooth out the pulses, and maybe alowa wider injector selection?

R2.0

[T3Bunny]

A couple of things. First, I don't think a servo is the answer either. On Cis-E, the plunger is controlled by the frequency valve, which looks and acts suspiciously like a regular Bosch injector. So I think its' possible to use MS and a frequency valve or injector to do full range fuel control.

A servo is definatly not the way. It would be expensive to impliment and it would likely require special coding.

Lemme address your second question first. Mainly we are dealing with very light throttle and idle at these flow rates. I am not sure that its going to matter much here. I feel its a non issue. No matter what system you go with there will probally be a bad point or comprimise.

Questions 1 and 3 are assuming a familiarity with the ford injector. And are realy directed at that. I posted a cool link on the first page: http://www.stcvalve.com/I-Solenoid.htm#0 I think that one or two of these might have the flow capability needed.

As for the low load situation, I can see something like an accumalator being used between the solenoids and vlaves. The real idea though, would be a setup that allows relitivly cheap experimentation to occur.

And it would be super cool, to try and simply impliment an existing setup using either the frequancy valve ot the electro hydrolic actuator used on the more modern Motrinic CIS rigs. That would actualy have broad appeal to the VW croud.

[R2.0]

Questions 1 and 3 are assuming a familiarity with the ford injector.

Sorry - the Ford CFI injectors are basically (very) large Bosch style injectors with different ends - they can flow upwards of 50#/hr a piece. And they can be modified to standard ends.

And it would be super cool, to try and simply impliment an existing setup using either the frequancy valve ot the electro hydrolic actuator used on the more modern Motrinic CIS rigs. That would actualy have broad appeal to the VW croud.

That's kind of what I was getting at in my first 2 posts My problem is lack of time and a proper CIS setup for experimentation. That's why I was more throwing it out there as an idea.

[lapuwali]

I love this. Multiple injectors to control the pressure? Why not just move those injectors to the ports and forget the CIS fuel distributor altogether?! CIS is one of the simplest FI systems available, and you guys are dreaming up all kinds of ways to make it very complex. Trying to control the entire system pressure with one injector would be very challenging.

The only reason CIS atomizes better than EFI is the higher system pressure. I have no idea where T3 gets his data, but the Bosch books all speak of 100psi system pressures, not 50 or even lower.

The plunger is NOT directly controlled by the PWM valve on CIS-E. The only thing that's attached to the plunger is the airflow plate. The PWM valve alters the control pressure, which alters the air/fuel curve by altering the resistance to movement of the plunger by the airflow plate. Controlling this valve IS the way to control fuel flow while retaining the airflow plate (which a previous poster listed as a goal). This would not be at all hard to do.

MAF sensors don't have to be restrictive. There are 400hp OEM cars out there with MAF sensors. Just use one that's big enough and there won't be any restriction.

[T3Bunny]

What is the point of running the MAF? It adds to the cost, and there is a nice MAP sensor in the MS already. The point here would be to retain the STOCK apperance of the CIS system and to gain some further measure of control over it. Also to try and maintain the charachteristics of the fuel atomization that the CIS system has.

Anyone familiar with the water injection systems by Aquamist, will understand what the thought here is. Not using the injector to control fuel pressure exactly, but to control the flow. Of course in a CIS system they basically become the same thing.

And the other point, was to fully retain the CIS setup and use the MS to control it VIA a frequancy valve or the hydrolic actuator.

Whatever the case, this is a theoretical discussion still. As nobody here realy intends on running this, yet.