Carl Fausett

Two very good books on Bosch Fuel Injection tuning are:

"Bosch Fuel Injection & Engine Management" by Charles Probst, SAE;
and
"How to Tune and Modify Bosch Fuel Injection" by Ben Watson

I found the diagram 5-5 with the caption that reads: "Even on stock systems, L-Jetronic air-flow sensor flap is fully open at 3500 to 4000 rpm. At higher RPM, it is no longer signaling increased air flow to the ECU" on page 7-12 of the Probst book abaove.

ViribusUnits

Really, I'll have to take a look. That is intersting. Any clue as to where I might find the book? I'm pretty sure my public libary isn't going to help me.

The cheif book I was looking at was "Gasoline Fuel-Injection System L-jetronic" published by, edited by employese of, etc, for Robert Bosch GmbH. No writer though.

Have you verified that information on your test stand engine? Like I said, my experments don't agree, but who knows, maybe I screwed up.

Anyways, would it be too much to ask for to have an a/f ratio chart for the before mentioned dyno run? Also, maybe a chart of the boost level at that rpm?

I know 60 questions, but Thanks.

ViribusUnits

Here's what I did in my experements to my car.

First I did the duh thing. I mannaged to hook a volt meter up to the AFM, and went driveing about. The voltage wasn't very stable, as the engine's demand was constantly changeing. After reachin around 3500 rpm, the voltage seemed to stabalize, but wouldn't ever just stand on a number. I would expect it to stand on a value, probably the maximum or minimum value if I'm maxing the AFM out, right? Well, it wasn't exactly doing that.

So I thought, well, if the ECU is following a map over around 4,000 rpm, the amount of fuel that is injected should be the same at 5500 rpm, no matter how the throtal was possitioned, as long as the air flow rate was over that acheived at WOT, 4000 rpm. To test this I decided to check the a/f ratio at 5500 rpm, and nearly WOT. Howevere, I've got some limitations, as I've only got a narrow band O2 sensor. My solution was to dissconnect the WOT enrichment switch. I did a couple test runs to confirm no detonation, to be sure it was safe. After that I hooked my volt meter up to the O2 sensor, and did a base like WOT run. The a/f voltage returned remained under like .5 volts, or about stoc, maybe a little lean. As sensitive as the narrow band is to when you go slightly rich of stoch, I expected a the O2 sensor to return an elevated voltage on the part throtal runs. I did it several times, but the voltage returned remained the same.

Uh, why is that? Clearly some sensor was telling the ECU that the engine was under part load. My next question was how. I couldn't find a MAP sensor, anywhere, and the WOT sensor was disconnected. The fuel pressure regulators just keep a constant pressure, and the temp sensors are not doing it. That just leaves the AFM. What the heck is going on.

So I went the above mannual. In the above mannual, they have a chart that shows the voltage returned by the AFM is a linear fuction of the angle the flap is at. However, they have the angle of the flap at a logarithmic relationship to the air flow rate. That cought my attention, as I had not expected that. That makes alought of sence. As the flow rate increases, the sensitivy of the AFM is reduced. This explains why my voltage seemed to stabalize, but not "top out". The sensitivity was being reduced, makeing the voltage returned "look stable".

So to test this, I opened the AFM up. I had done this a couple times before, and I had a spare AFM. I adjusted the carrier arm so that when the AFM was absoutly maxed out, the arm would fall off of the black carbon conductor, makeing the AFM seem like it was unpluged. Well, I did this twice. One where it almost fell off, and one where it did. The one where it almost fell off was to make sure the rich mixture wouldn't cause a problem. The one where it did was to check to see if the AFM would be maxed out. The mixture didn't cause a miss, or any other problems, other than it was running really rich, and likes to carbon foul the plugs. That and I had to advance the idle speed so it wouldn't try to stall out. This was not a suprise.

When I adjusted it to fall off the black carbon conductor when the AFM was maxed out, I kept going up past 3,000 rpm, to about 5,500 rpm. Some times, somewhere between 5,000 rpm, and 5,500 rpm, the engine was stop running for a moment, and then start again, often repeating this several times before accelerating through 5,500 rpm if I stayed on the gas long enough.

My conclusion was that the AFM was not being maxed out at 3,500-4,000 rpm, but was getting close to being maxed out at 5,500 rpm, if not maxed out entirely. Instead of maxing out, the voltage was following a logarithmic curve that flatten out as the engine gets close to the max flow rate of the AFM. I propose the theory that the ECU reacts faster, and more accurently to the voltage returned than I was able to measure with my volt meter, given the circumstances of driveing a 928 at high rpm, and high load. Well, I had a buddy helping me, but he was able to do no better.

I'm fairly sure that the AFM will not be maxed out by a stock 928, but how much it would take to max the AFM out, I did not test, and I will not speculate on.

Of course, I could have screwed it all up.

2V4V

VU/Carl,

Depends. Depends on load, condition of every stupid freakin' vacuum holding device attached to the plenum, every seal, and the condition of the AFM (ie, is the spring fading?). LOTSA "x" in the equation...

Vacuum leaks reduce flow through the AFM, simple as that.

The other thing to know about L-Jet is the AFM is as much about driveabilty as it it is about actual air flow - not too many scenarios where one is over 4K RPM and not heavily, or completely into the throttle. Much over 3K, you can run Alpha-n and it works quite well.

On L-Jet, the fuel injection pulse is simply a modified pulse from the EZF ignition unit. The AFM is one of the correction factors, like TEMP I and II, full throttle switch, idle switch, and battery voltage.

Anyway, back to the science stuff. It depends. I have been datalogging L-Jet for a while now and I would say on the 928 the answer is somewhere in the middle. This is based on reading AFM output voltage into a datalogger -which also gives me RPM/WBO2 logging and some other stuff that I'm concerned with. I'm north of 3500 RPM but well south of 6K when the AFM maxes out. The AFM is a brand-new unit aquired specifically to have quality baseline data - and interestingly enough, the old unit aged pretty well - readings are were very similar to the brand-new unit.

Some quick bevnap calcs would put the AFM maxing out in the range Carl is suggesting under any kind of worthwhile boost (5+ PSI) and WOT. But ONLY in conjunction with WFO throttle position. (This is the often quoted "why you can't blow through L-Jet AFM"myth). Just remember, the amount of air flowing through the AFM is STILL regulated by the throttle, regardless of whether one is at atmo, or one is blowing 5PSI, it's about flow. If there were no throttle plate at the other end of the AFM, then it would just peg the minute you put about 1 PSI boost to it, but there IS a throttle, so it just doesn't happen that way. If it did, Carl's CIS cars would be pegged all the time too - they aren't. (rant off)

VU, amazon.com if they don't have a McBookstore in yer neck-o-da-woods (Borders, Barnes&Noble). You'll drive yourself crazy trying to read volts from the AFM off a DVOM - they just do not have the attack/decay resolution necessary for the task - you could never read it if they could anyway, the V-out dithers quite a bit on even a .1V basis, let alone on the .01 V sampling I do. You could use a hand held o-scope (Fluke, Velleman) but that would be serious $ unless you got one handy. Or, spend a bit and go WBO2 with datalogging capabilty. Between the LM-1 wideband and my laptop, I can do 8 channels plus WBO2 and RPM.

Greg

ViribusUnits

All right, will start looking. I have to go into Houston to get a hold of a Barns and Noble, or something like that. That is a 30 minute to an hour trip. So be it. Augest looks to be the next time I'm headed in that diriction.

Yea, I went nuts watching a volt meter. I've had enough of that!

Something to note, the horsepower curve typicaly follows the air flow curve. I'd expect the maximum air flow to occure at the maxinum hp produced point, or around 5500. That would be a good bit below 6,000 rpm right?

The final thing, is that if the AFM is not supplying the information required to maitain mixture control above rpm X, what is? The L-jet 928 is not set up to run alpha-n, except at idle, so it's not that. There is no MAP sensor, or any other sensor to reduce the fuel flow under less than full throtal. And yet, something is causeing the mixture formation to remain stable. Suggestions?

928ntslow

Whoa...you guys are freaking me out

Carl, I will email you directly over what we can work out on the SC kit. I am intrigued to say the least.

I have a base dyno before I did any modifications. Now it's time to get back to the dyno for the recent mods. Then, if I install the SC, the dyno numbers will be quite interesting....and yes, they will be yours to publish.

2V4V

VU,

Remember, the L-Jet fuel pulse is derived from modifying an ignition pulse. Perhaps this'll help make things clearer, here's the "pulse flow" from igniton to fuel...

1. IGNITION - this is the signal from the infamous "green wire" that gets fed to the L-Jet brain. (The one most folks see anyway. According to 1983ish+ DIN wiring norms, the feedwire from the ignition is green throughout the harness. YMMV.)

2. PULSE SHAPER - So, now our igniton pulse is at the L-Jet brain. Ignition pulses are notoriously "dirty" so we need to run it through this section to turn it into a nice square wave that will be easily used by the next section.

3. PULSE TIME - (or as Bosch likes to call it 'division control mutivibrator')
Here's where the basic pulse time is generated using RPM and the AFM. Air mass is know by referencing volume from AFM and TEMP I. RPM is by frequency of arriving ignition pulses. This is the "baseline" pulse - ie, if everything else is "normal" (warmed up, battery voltage nominal, average speed and throttle) this is the appropriate average kinda pulse to run the car optimally under those conditions.

4. MULTIPLIER - OK, so driving isn't optimum. Pulses need to be modified for a cold motor, 3/4 throttle, and a battery that's missing .5 Volt because it was really, really cold last night. This is where those calcs take place. Also, where the full throttle, idle switch, and O2 sensor come into the mix.

5. FINAL OUTPUT- In L-Jet the injectors are hot all the time. They just get grounded by the ECU for the appropriate duration.

ENRICHMENT- So how does a stone-age piece like L-Jet provide enrichment when you mash the throttle? Mechanically. One of the neat-o things that happens inside an AFM is the vane will 'overshoot' a bit if you slam the throttle open. Thus, you get sudden throttle enrichment - without the use of TPS (throttle postioin sensor) or MAP. CIS does the same thing - works pretty well actually. Mass and momentum.

So, the basic fuel injection pulse is pre-determined (mapped out if you will) by RPM and AFM. It gets modified a bit by the various temps and O2 kinda stuff then, out it goes - "two squirts per putt". Anything over 4000 RPM is gonna have a pretty finite amount of throttle settings to get you there. At full throttle, the microswitch pulls O2 correction off the table and the car goes to a pretty fixed multiplier. At full throttle, you really need only one map with a temp I mod to achieve whatever max power you've decided on.

Once again, it's an Air Flow Meter. If your car has less than perfect gaskets, rings, vacuum devices, vac hoses, etc - it will pull some of the air from somewhere other than through the AFM. If the valves and such are not sealing correctly - you can lose some of the ability to generate the manifold vac necessary to open the AFM. It is entirely possible that your AFM is not fully open till 5K. Maybe even more. Mine is WFO at lower revs, but I have some air-flowin' mods.

Greg

ViribusUnits

My motor is in reasonable shape, and I've recently redid all of the vacuum lines. I notice no lean conditions, no missing, and a general avalibility of power. I figure it's safe to say the car is running good. However, other than a slightly moded exaust, the engine is 100% stock.

Right, I understand how the ECU works. I'm sure if I really wanted to, I could get one of the EEs to tell me how the sepreate componenets of the ECU work. Notice, no where in there do you mention the ECU going to a Alpha-n, MAP based, or other type of system. This could be an oversight in the documentation, right? However, in this case, we can treat the ECU as a black box, and work back what inputs result in what outputs. Thankfuly, it's a reletivly simple system!

The air flow rate at say 4,000 rpm WOT, and 5,500 rpm WOT are sigificantly diffrent. Depending on the ratios of the volumetic effecencys, it's a bit less than 37% more air.

Thus, with the throtal body, I can reduce the flow rate of air at 5500 rpm down to the same rate at 4,000 rpm, and WOT. If the computer doesn't know that the engine is injesting less air, it will continue to inject the amount of fuel required for 5,500 rpm, WOT. As such, the a/f ratio should go to like 10:1 or 11:1, if it was stoch at WOT, which is unlikely. I have seen that this doesn't happen. Therefore something is informeing the ECU to pull back the pulse length. If it is not the AFM, what is it?

Next question, after your AFM maxs how, how does the a/f ratio react? Expecaly at high rpm, high, but not full, load situations?

Carl Fausett

928NtSlow -

I'm ready for you!

2V4V

VU,

Some of the following is semi-conjecture, since I do not have a full bench set-up to run L-Jet through all it's machinations and stare a my O-scope like a stoned monkey.

Based on empirical evidence -ie, data logging 2 L-Jet cars with vacuum checked manifolds/accessories (seal it up at the curve and at the end of the runners, pump it down) I can say at this point that when the full throttle microswitch is activated, the open loop starts. Once the open loop starts and the AFM hits north of about 7.5 V/ 5K RPM, the AFR goes from in the low 12's gradually up toward the high 13's as one approaches (and occasionally exceeds) redline of 6K. It would appear that as the AFM maxes out, the pulse width is whatever the CPU can spit out maximum. No, I don't have quite enough injector pulsewidth sampling yet to put a bunch of data up on that - so it's still a bit of theory in progress - but not terribly important to me at this point.

But, it would be a rather simple circuit - if ignition pulses are x, and full throttle swith is active, and AFM is maxed out, pulse width is maxed out too. BUT, you still get more fuel (to a point) because the RPMs continue to rise, and the number of injection pulses is determined from the number of ignition pulses. Comprende?

Greg

mspiegle

Nice work. I think your warantee and payment plan are pretty kick ***.

John..

I would not even consider keeping the AFM in the equation on the setup. Just go MAF and install the larger sprayers and you will be much happier. The benifits of the MAF are numerous and include:

Much improved throttle response
The ability to tune within low and high limits of the injector
Less intake restriction

It is just much better to upgrade the car when making those kinds of mods. It doesn't cost all that much in the grand scheme of things, you could do it all for under $700.00. The ability to program the entire fuel curve will beat the FMU approach.

Remember, it isn't just about a peak number, but rather the transient response in getting to that peak number.

The L-Jet is mapped on the top end off of RPM and max AFM voltage. The AFM voltage will swing wildly under light to moderate loads. I have the entire conversion map done now on two cars with MAF and it works very well.

2V4V

John,

We would both agree that losing the 'barn door' is a good idea. MAF too has it's shortfalls when compared to MAP, but that's a whole different thread.

The AFM voltage does vary widely at low revs/loads which is it's primary function - it's more a transition tool than a 0-redline feedback device. By the time you're pushing 3500 RPM steady-state, rolling on the throttle just requires an increased number of fuel pulses to keep up with the revs (esp. since Ve starts falling as the revs keep growing, NA of course).

Yes, I spent some more time playing in the logs and once the AFM maxes out, it becomes just about the number of pulses and the Temp I/Temp II mods - full throttle switch does impact as well, but the main thing is the revs and the pre "mapped" requirements of the engine.

Greg

marton

I just checked, both the Bosch books mentioned by quick carl are available from Amazon.

Marton

Carl Fausett

Greg and John -

There is no doubt that you two guys are further down the road of MAF L-Jet mods than anybody has ever been...

Greg - given your last post - we came to the determination that L-Jet would fuel 6 psi of boost at 800 CFM without mods. Do you agree?