One Lap of Lime Rock through the eyes of an Oxygen Sensor

[304065]

This thread might also be titled-- why you want your speed switch to be working.

Setup is 1971 911E with MFI, O2 sensor in port side muffler inlet, LM-1 data logger with RPM converter.

Here is a map of the track so you can follow:



The analyis begins with the throttle blip at the end of the main straight into Big Bend. You can see I'm off the throttle and the AFR falls into the 11 range.

The reason why this happens is the RPM is fairly high but the throttle is closed-- the MFI pump is still putting out fuel because my speed switch is broken- ordinarily with RPM above 1500 and throttle closed, the solenoid would shut the fuel off entirely and you would see the AFR stabilize or go lean. Where the AFR trace dips and goes back up again, you can see I've picked the throttle back up before the apex.

Then it falls off again at the brake point for the left hander, a throwaway turn, because you get set up for the long pull through the right hander-- that's the next mountain on the RPM curve-- and you can see that AFR jumps around basically in between 13 and 14:1.

The uphill is interesting- this is the some of the biggest G loading on the track- and the AFR spikes way lean right after I come off the throttle-- have to investigate this, could be a fuel pressure problem, or a leak in the throttle shafts.

Back side of the track is fairly straightfoward-- WOT pulls in between the turns, with a big rich falloff on trailing throttle. Interestingly, the last pull before the blip, the mixture actually gets RICHER-- this is WOT in 5th Gear (you can see that from the shallower slope of the mountain) And then the AFR goes rich big time in the braking zone.



There are a few conclusions: first, I need to fit my rebuilt throttle bodies to the car and set the mixture. The ones I used for this race are probably leaking air around the throttle shafts, and I know correlation isn't perfect-- hence the jumpy AFR.

Second, the mixture could be leaned out a touch-- I was shooting for low 13s for best power mixture by adjusting the main rack-- which looks to be adjusted about right-- but the idle mixture is still WAY rich-- when the pump is turning fewer RPMs, the mixture drops into the 11:1 range. Of course the engine was running about 180F, and the thermostat system is in perfect order, so I don't think it was thermostat-related (or the main rack would have been rich too).

Finally, I need to fix the speed switch so the fuel cuts off when the throttle is chopped.

I'm open to any comments or suggestions (except that I'm too slow. . .or that the falloff of the RPM isnt' dramatic enough because my transition from full throttle to full brakes isn't crisp enough. . .these things I know and am working on . . . )

Enjoy!

[NY65912]

Build a 3.8

[911st]

It is hard to understand the above. Better to put in on a Dyno in 4th gear and plot the A/F's by rpm. Or, could try to just give a 2k to redline pull in 3rd or somthing.

I think the way MFI is designed, it should be at its ideal A/F at TQ peak. Then they will usally go Richer after TQ peak. This is because the quanity of fuel will stay the same from TQ peak to red line on every stroke. Then as the effecency of the motor falls off, there will be to much fuel per stroke.

You could try to set it up so the A/F is at ideal at HP peak but it may be to lean at TQ peak which is where the motor is running at its most effichent level and taking in its highest quanity of fuel and air per stroke.

I think leaking throtle bodies are mostly an issue at idle and just above. They can be adjusted around to a degree on a track car.

If one has good A/F's up top and is to rich down low, some times the rod to the pump can be tinkered with to lean it some. The goal being to make sure the ramp on the space cam is rotating all the way to the closed postion. The way the space cam is designed, as soon as the throtle is applied, there is a ramp falls of kind of quickely and adds fuel quickely. Thus, if the rod is adjusted short, it will richen it at idle to the point where you can not use the normal adjustment to get it close.

If attempted and there is a stumble, may have to live with the rich mix.

It has been a while sense I played with MFI so I may be a bit off here. I had to make my own tools to play with my pump and change the space cam out for a 2.8 twin plug I built once. I love MFI but space cams are a gross way to meter fuel.

[jcge]

Originally Posted by 911st

I love MFI but space cams are a gross way to meter fuel.

Here's what an MFI fuel map looks like.....(1960 MB 220er)

No surpirses in the similarity to aftermarket EFI maps (using alpha n control)



John

[911st]

Yes, a space cam looks like a carved fuel map in metal. No such thing for ignition.

Like ignition, I belive, once the torque peak it reached and the internal weights are extended, the MFI pump is not really able to pull back its fuel delivery.

You would think it would be doable to translate any motors needs using some type of computor probram to make a custom space cam using a 3d mill for any motor if one had the skills and knew the air flow of a motor at all points.

One of the issues with MFI is that it is a "referance based fuel system" that has very specific expecations. If any thing is out or air flow is not as expected, it will still send the same amount of tuel and can not compansate for changes except tempture and altitude to a small degree.

It also uses uses springs, weights, linkages, a pin in a very small 3d map. Many things to go out of spec. There are limits to its adjustability. For most of us, we can only make gross adjustments and richen or lean it at all points w some adjustment at idle. Add a race muffler and that dose not have as much back pressure and or flows better up top and things are out of wack.

I like how MFI squirts fuel at about 230 psi just as the intake valve opens up and with nothing in the way of air flow like a metering valve.

MFI is cool and responsive. However, anythime we modify a motor comprimises will be necessary.

Just my $02,

[304065]

Fortunately for me I am racing in a stock class where modifications aren't allowed. Hence, the stock equipment works fine. I agree that technology has moved on from MFI, but the rules are such that I can only switch to carbs (and give up 10 HP).

Let me explain the above a little better-- what you are looking at is a series of successive WOT "pulls" in fourth gear starting before the apex of each turn and ending at the braking point for the next. These are actual data from the car on the track, what better to analyze? While a dyno setup is good, nothing matches the transient response like track testing. That said, the next step is to install the Aux Box (from Innovate) so I can see actual acceleration data- then go run the tests to see quantiatively what the best settings are.

The rod is exactly 114mm long, I have heard of it being adjusted short to fix the richness issue. When I get the replacement TBs installed I'll run through the correlation steps again (I actually have a set of protractors from Porsche) to make sure everything comes off the stops properly. I agree with you that at WOT leaking throttle bodies shouldn't make a difference-- air leaks in the shafts shouldn't account for the mixture falling off like that-- I really need throttle position data also as perhaps what we're seeing is the throttle return prior to getting on the brakes.

Thanks for checking in, do you have any photos of those MFI special tools? A fellow Pelican wrote an extensive thread about tuning the internal adjustments to get the slope of the fuel curve correct-- it would be interesting to see what you came up with.

JCGE-- where do you GET such cool data? That looks original. Oustanding.

Of course in the MFI repair instructions there's tabular data on fuel quantity per 1000 revolutions of the pump at various throttle positions-- I'll see if I can dump it into Excel. . .

Cheers

[jcge]

John - fuel map came from a MB website - albeit in German

http://sterntwiete.mparschau.de/html/body_einspritzung_2.html

I think MFI was used more extensively by MB than by Porsche - MB community has a lot to offer enquiring MFI minds....

The MFI repair instructions (pump specs pages that you refer to with the fuel delivery volumes per 1000 rpms) also makes reference to the adjusting screws(white pair, black pair and CO) that Jeff Higgins found so useful in the open heart thread (look in the right hand column of the sheet)

John

[cmcfaul]

I think you are leaving a bunch of HP on the table. I have been driving around with my LM1 for a couple months now. Seems to make max HP at 14.5 - 15 WOT. Having made numerous adjustments during this time I have found a 10% increase in HP (gets to redline better than 10% faster) at this setting vs 13.0 - 13.5.

I get a miss at 15.5 +. I finally setteled on a setting that gives me 13.5 cruising (part throttle) and 14.7 at WOT.

Very cool tool. Finally takes the guess work out of MFI tuning

Chris
73 911 e

[304065]

Originally Posted by cmcfaul

I think you are leaving a bunch of HP on the table. I have been driving around with my LM1 for a couple months now. Seems to make max HP at 14.5 - 15 WOT. Having made numerous adjustments during this time I have found a 10% increase in HP (gets to redline better than 10% faster) at this setting vs 13.0 - 13.5.

I get a miss at 15.5 +. I finally setteled on a setting that gives me 13.5 cruising (part throttle) and 14.7 at WOT.

Very cool tool. Finally takes the guess work out of MFI tuning

Chris
73 911 e

Hard to tell- that car is in storage at the moment.

How do you measure 10% faster, from the RPM trace?

What gasoline are you using? E10 will require a much richer mixture than pure gasoline due to the Ethanol component. Ethanol stochiometry is 9.0 so the adjusted stochiometry of E10 is 14.1.

[cmcfaul]

Not the most scientific method but I count how long it takes to get from 3k rpm's to 7k rpms in third for a short test and 4th when I want to confirm my findings. 4th at 7k is 120 mph so I don't do that test too often but it provides the most accuracy.

The seat of the paints difference is apparent. At the leaner setting the engine really screams. The richer setting and it feels more bogged down (lethargic).

My readings might be a bit off. My sensor is mounted on the tail pipe as close to the muffler as I could get it. I have been told that outside air could be going into the tailpipe giving me slightly leaner readings then actual.

I suggest trying the leaner settings and see what effect it has.

I am running 94 octane e10. Engine is brand new from rebuild. 9.5 cr from JE pistons and a dc30 cam grind (e cam with more lift) otherwise stock 2.4 E.

Chris

[Eagledriver]

I wouldn't change a thing (except to get the speed switch working). Theoretically 12.5 is max power for gas. In reality 13-13.5 seems to be just as good. You are running right around 13-13.5. These traces are very good for a non-efi set-up. You should see what it looks like with carbs (terrible).

-Andy

[KTL]

I'm thinking the same as Andy. While the AFR is a bit jumpy, the results are consistent, except for the unexpected lean spikes. Just turn the graph upside down and you've fixed your problem.........

Agreed the track plot gives real world indications of how the car behaves dynamically. But still don't rule out the dyno for fine tuning in a more controlled enviro- especially for tracking where your engine's power lives. I found with some of my tinkering that my car's power fell off dramatically beyond 6500 rpm. Had I not put it on a dyno, I wouldn't have realized that because the car feels like it pulls strongly to 7000 or more. Maybe my butt dyno needs calibration.......

[Jeff Higgins]

John, your A/F vs. rpm graph looks just fine to me. Fantastic, as a matter of fact, in MFI terms. I've gone hog wild and have run every car I can get access to with my LM1 hanging out its pipe, and they all look almost identical to yours. With just one caveat, I wouldn't touch a thing.

That caveat would be with regards to your comment on leaning out the main rack a tad to bring the A/F up a bit leaner on part throttle. My simple advice would be "don't". As a matter of fact, I would back it out a bit until the leanest it ever runs is in the upper 12's. I went through this exercise a few years back on my bastard 3.0 project, looking for an acceptable lean cruise. I sacrificed WOT, high rpm (allowing it to go leaner than desired) to achieve this. Here is the result:



This was with the maximum rpm, WOT running as lean as 14:1. Many would think that is "safe", but I have come to disagree - for obvious reasons. Granted, my motor lived for three years, 30,000 miles, and dozens of track days with that pump setting. But... we typically run in the manner shown in your graph - on the gas, off the gas, repeat - in very short intervals. This piston failed after several minutes at WOT redline in fifth gear (kind of like a run down the Mulsanne, only in Northern Oregon...). It did not have the chance to cool by virtue of chopping the throttle, going a bit rich again on the throttle, and so on. Granted, again, kind of unusual circumstances that most of us will never encounter (or be lucky enough to stay out of jail if we do...), but I have to believe the possibility exists to achieve this result even on the track. I've come to feel the high 12:1 range is far safer.

With my pump set to never allow it over about 12.7-12.8:1, my part throttle mid range A/F - the "freeway cruise" mode - runs in the high 10:1 range. But, guess what - it still gets 17 mpg on the freeway. Yes, the raw number says the mix is exceedingly rich, but the motor is actually using very little of it, hence the decent gas milage.

I've come to believe the only "cure" for this is to regrind a portion of the space cam. That's my next mission. It seems if we take down that portion of it that controls WOT / high rpm (which actually starts to ramp up on the cam, thereby leaning the mix), we can adjust the main rack to remain "safely" rich under those conditions, while leaning the part throttle cruise. In other words, we can lean out the main rack without going into the "danger zone" of A/F on the top end.

Oh, and Chris - 14.7:1 is going to cost you in the long run. Avoid my pain (and cost) by fattening that thing up. Considerably. One of the issues in air cooled heads is that they are particularly susceptable to over heating. It's a viscious downward spiral once they do, with overheating leading to detonation leading to more overheating leading to more.... without the capacity to recover and cool down like water cooled heads. Live with the rich part throttle cruise. It's cheaper in the long run...

[cmcfaul]

Hey Jeff,

I hear ya. The reason for my rebuild was a burnt piston. Not as bad as yours but still broke the ring landing on the same / similar JE pistons as yours. The problem is that it is so much faster when lean.

A big contributor to my piston demise was heat. I let my car get way too hot (well north of 250 on the track). I will not do that again. I am adding another oil cooler or upgrading my little Mocal cooler to a Carerra cooler before it does another DE.

These AFR meters are the coolest things ever. MFI has been a mystery forever until now. And I am actually getting good gas milage.

I am going to keep the settings where they are for now for street driving. It only sees WOT high RPM's once a day ;-). Since the rebuild I have not seen temps over 210 and only use the highest octane fuel I can get.

Chris

73 911 E

[txhokie4life]

How much power do you lose when you go richer?

I've got an RSR track car I am rebuilding the engine on
and want to favor longevity vs peak power (within reason :-)

thanks,

mike

[Walt Fricke]

OK

I have some frustrations dealing with my LM-2. While the RPM recording works fine (shows my bad downshift matching, and some overshoots on upshifting which I have pretty much gotten rid of), that's not why I bought it.

Here are two screens from a race last fall at Hallett. The first is "raw." The second has had pretty heavy smoothing applied to it in order to be able to make some minimal sense out of it. It says I am nicely rich at WOT, but it leans when off the gas. The rich part is what we aimed for on the chassis dyno setting up the Electromotive EFI, looking for 10-12 AFRs up in the 6-8,000 range.

But why is the "raw" part so jumpy? I can't really tell anything from it.




Might this be the difference between an MFI trace versus an EFI trace?

My other peeve is that at the moment, when I try to log a session, it says I don't have an SD card inserted. But I do - same card on which I previously recorded some stuff, plenty of space for more. I can't possibly watch my display while on the track, beyond maybe a quick glance on a straight, so without the recording it is useless. Also, in the past it has been OK with some cards, not others. Baffling. I've formatted them (FAT 32 seems to be the option), too, though I suspect they come formatted.

Puzzled

[Eagledriver]

Walt,

Your graph is poorly scaled. The x axis needs to be stretched out more to see better. You are interested in the "dyno pulls" on each straight. What it does when you lift is not of much importance. If it's too lean or rich as you open the throttle you'll notice it with a stumble or bog. If you stretch out the graph in the places where you have a steady WOT pull you'll get a better idea of your AFRs. It may also be that you have electrical interference causing the jumping. I don't have data logging with my oxygen sensor but my camera catches the display so I can get a good look at it after a session.

-Andy

[Jeff Higgins]

Every "raw" LM1 map I have ever seen looks like that, Walt. I'm not sure what the problem is, not being the least bit electronically inclined. It does it regardless of motor, regardless of induction. I've checked quite a few Porsches with MFI, CIS, Motronic, and carbs. Harleys, one Ducati, two old Beetles, and even the wife's new Subaru. I have had to apply smoothing to all of them to get a legible result.

Chris, your oil temps don't tell the whole story. It's cylinder head temp that does the damage, and that does not transfer all that well into oil temperature. You will get far better data from a head temp sensor. Mount one of the cheap under-the-spark plug variety and prepare to be amazed. Or frightened... Head temps spike far more quickly, and drop far more quickly, than oil temps. Oil "buffers" the temperature gain and loss rates so you don't see the spikes, only the "average". It in no way captures or displays localized spikes such as what the heads see.

Mike, I have seen minimal loss of power with the richer mix, if any at all. Unless you are racing at the highest level, for real money, I believe the trade-off in longevity vs. a possible five horsepower (if that) just isn't worth the risk, and certainly not worth the cost when it all gets over taxed.

[304065]

Walt, not bad at all really. A few tips.

Go into MS paint and put a nice thick line across the 13 AFR point. That makes it easier to see. Likewise, put a different color line across the RPM at the torque peak. (I forgot to do that).

Next. Here is Hallett.



The smooth trace is pretty helpful. See the sharp slopes? Those are the straight entries in lower gear- the RPM picks up quickly due to lower aero drag and torque multiplication through the gearbox. Then there's a shift and the trace falls off, followed by a shallower slope in the higher gear to the redline (maybe a little over Walt!) then a big tumble when you get on the brakes. The broad valleys are the big sweeping turns, with a couple blips in between where you are obviously modulating throttle to stay at the limit. You can actually see that better in the unsmoothed trace.

Anyway, EFI has a throttle position sensor, so when you snap off the throttle, the EFI system cuts the fuel. That is why you see the whammo lean spike right after the RPM peaks- when you are on the brakes, fuel flow is basically zero. Contrast that with my malfunctioning Bosch MFI trace above- when I snap off the throttle at the end of a straight the mixture goes WAY rich which is usually followed by fire shooting out of the tailpipe. Dephlogistic, yes: optimum, no. Like I said I gotta get that fixed.

Other than that I agree with Andy- the data are all there, just too tightly compressed to analyze with the naked eye. Stretch out the X axis and you should be all set.

In-car graph of the AFR is pretty much useless for either track or street. It's OK at idle I guess, but the measurements we care about are under load and usually wide-open-throttle. And under those conditions we can't be staring at a blinking light, or worse yet, a moving needle. But once you have three or four basic tweaks of analysis down, it's a simple matter to come back into the paddock, whip out the SD card, slap it into the laptop and start looking at the data.

It's probably a full time job for one person at the track. Hell, Porsche has about seven dedicated guys, all engineers, on the RS spyder team. One guy's job is to do nothing but analyze the suspension data during the race to optimize it. Wonder if he forgot to pack the beer in the MAN transporter and ended up with that job? Anyway, the data are overwhelming, which is why a few basic tricks must be deployed in order to make it managable by mortals such as we.

[304065]

I would be just as afraid of too rich, for that matter. The GAMI folks, famous for igniting the "lean of peak" technique for turbocharged aircraft engines in cruise, have all kinds of instrumented test data on the evils of too rich. (Before I re-ignite that controversy here, lean of peak egt (which occurs at approximately the stochiometric point) does reduce power and lower EGT/CHT, and so pilots practicing it will increase manifold pressure to achieve the same target cruise speed at the point of minimum brake specific fuel consumption). We want it the other way around. But too rich washes the cylinder walls and creates all kinds of nasty crap in the engine and environment with no appreciable gain. We want a best power mixture for E10.