DME Motronic - O2 sensor questions
 

I hope this is the right forum for these questions or do I need to send them to Steve Wong? Anyway here goes....

O2 Sensor - I assume the way this works is that the O2 sensor reads zero at 14.7AF and the DME adjusts the PWM for the injectors up and down to get the O2 voltage to read around zero. This happens on most of the map until the WOT switch goes closed and then ?? happens?

As I understand it the O2 sensor is then (when WOT is positive) disregarded and the PWM is calculated from the map and at some point in the rpm range the PWM goes to full rich (the actual PWM for that is ?)

When the DME box gas quality adjustment switch is set to say position 2 (3% rich) or position 3 (6% rich), how does the O2 sensor still work? It would then be running at say 13AF so it would not be able to swing back and forth across zero to control the PWM because its reading would be off scale. How does this work?

Thanks,

-Henry

What a great question!
-Chris

I hope some guys will weigh in on this...not the usual rod bolt upgrade discussion.... that's why I suggested we have a CSI/Motronic section on this board....we'll see...

I went and re read descriptions in 911Chips (Steve Wong).
It looks like the voltage "trim" is applied to the O2 sensor to make the sensor act like it is trying to "control" to a new or different AF. E.g. add 1 volt of trim and the "zero" for the O2 is 13AF and not 14.7AF.

When you activate the WOT switch the DME goes to a completely different map and doesn't use the O2 signal and some other sensors.

What is not clear is that Steve states that while on this WOT map at some point no more fuel is added and you can run lean. What I want to know is when does this happen? Do you run out of PWM or ? Why would the map stop being in play?

-h

This is a good discussion! Am I correct in thinking the Euro-RoW 84-89 engines do not use the O2 sensor?? What changes did they require to legalize them when brought to the U.S.??

The O2 sensor reads roughly 450mv at stoich (lambda 1). The sensor output scales from 0-1V with 1V being as far as they'll read rich (lambda 0.70 I think) and 0 being full lean (lambda 1.25 from memory).

Here's the catch. Around lambda 1, the output changes very suddenly for every point of lambda. So the effect is that it outputs a continuosly swinging voltage of between 600 and 300 mv. The ECU reads this as being lambda 1 and uses it to correct idle mixtures and sometimes steady state cruise mixtures. It's pretty inaccurate but it works close enough to fool the authorities.

Bear in mind that most smog tests are only at idle or steady state cruise, and very,very rarely in a transient (accelerating) or WOT mode.

The ECU normally kicks itself out of closed loop control and into open loop and some preset amount of throttle opening, or a switch by the sound of it in this case.

"Full rich" injeciton pulse width would be around 80-85% duty cycle which could be as much as 32msec depending on the rpm your at. Injector control is infinite between the low end being as small as the ECU can control and the high end whichis dictated by total cycle time available (i.e. 20ms @ 6K rpm, etc)

Review the following chart on how a healthy narrow band O2 sensor operates:

http://forums.pelicanparts.com/uploa...1150420738.gif

As you can see, a narrow band sensor is most sensitive at the stoichiometric ratio of 14.7:1 (lambda=1), ratio for best overall emissions. This is a voltage between 0.2 and 0.7 volts and if the voltage exceeds this range, the DME will attempt to adjust the fuel to bring the voltage back into the correct range. The sensor only has influence at idle and part throttle, it is ignored at full throttle.

Now, depending on the program, the O2 sensor has a significant, but not complete influence on the air/fuel ratio at part throttle (no matter what LorenFb says - who is an idiot btw ;) ) Transient enrichment can be momentarily provided where the O2 sensor's influence is overridden for 2-2.5 seconds on a throttle transient, by increasing the quantity/pressure of the fuel injected either by software or hardware. This provides a richer mixture momentarily for power and throttle response. If you disconnect the O2 sensor, the lambda mixture correction will not function, same as open loop, thereby keeping the injector pulses to where they are mapped to. The following chart is an AFR vs. power curve for a standard internal combustion motor – although not necessarily Porsche:

http://forums.pelicanparts.com/uploa...1150421915.gif

At full throttle, as the O2 sensor is ignored, fuel quantity is referenced directly from the full throttle fuel map, multiplied by a factor of the output voltage from the air flow sensor. However, after 5000 rpm the flapper door has bottomed out and can supply no further voltage increase, so the DME’s injector pulses are solely determined by the map after this point. They maps are always in play and the injectors never stop fueling, no matter how high an rpm you go. So to clarify the statement you read, if you need more (or less) fuel at WOT for modifications or displacement increases, it must be done by reworking the WOT maps, as there’s no other way for the DME to know what the additional air flow increases and fuel requirements are.

Good discussion, I am going to need Steve's help with my 3.8 conversion once it is back together and broken in. I've been suspect as to whether the O2 sensor on my install is even working, I have made sure the connection from the DME to the O2 sensor is intact, but I still seem to run rich (as a 3.6 with a stock map). If the WOT performance doesn't use the O2 (which is what I have always read) then tuning this on the chassis dyno seems quite feasible for a man of Steve's talents.

Steve,
Thank you very much for your explaination.
This is all making much better sense now.
I need some time to digest your writings and then attempt to produce some intelligent questions if needed.
My focus right now is to confirm that my 89 3.2 is all working properly to start with and then look at making modifications, e.g. one of your chips.
Thanks again for you time and thoughts.
-Henry

Another thought now after rereading what Steve explains....

After 5000 rpm when on the WOT map or when the air flow pins the door the DME is operating "blind" as there is no direct way for the DME to know the amount of air going in the engine.

In these cases the DME uses the maps developed at Porsche with I guess the primary input being RPM to come up with the PWM values.

This seems sort of scary, because this is where the engine is running much of the time on the track. Steve, I guess this issue is eliminated with your air flow assembly using a hot wire type technology which I assume can still measure the air flow up to at least the rpm limit and if modifications are installed that allow pumping more air it will read that also.

Correct?

Thanks again,

-Henry

Here is a simple diagram of inputs and outputs to the DME for off idle operation.
I contend that the DME doesn't know anything else than this at a given time and possibly past history. Steve, is this correct?

This will be my starting diagram.
Next will be how I think the maps work and when the DME is calculating the PWM without enough information as Steve suggests. So we will all understand and see the benefits of a different chip. I will appriciate very much if Steve can check as I go.

Also.. Steve, Do I understand that the voltage trim you talk about is a "reverse" EMF applied to the O2 sensor so it will read a lamda of 1 at different than 14.72AF ?

http://forums.pelicanparts.com/uploa...1150480420.jpg

So far, that is correct, except that the AFM voltage and O2 voltage have no correlation to each other. The standard Motronic system, is only accurate compensating within a narrow boundary of where it is mapped. Modifications such as exhausts and displacement increases will throw it off at both high and low rpms. Our HFM system continues to supply metered airflow increases to the DME beyond the rpm and flow capacity of the most demanding normally aspirated motor, and does it so accurately that almost no changes are needed in the fuel maps when the engine is modified, whether the system is placed on a stock 3.2 motor, or installed a displacement modified 3.4 or 3.5.

The fact is, both part throttle and WOT of the factory chip is not programmed for best power, but for best emissions, and since the AFM can't meter past 5000, the air fuel ratios in the upper rpm zone are more of a WAG for the lowest common denominator. This means low and mid rpm is on the lean side for lowest emissions, and the upper rpm range is way fat to keep the ceramic monoliths of the catalytic converter from melting down under hours of sustained WOT operation on the autobahn at 150 mph, and for the factory to cover their butt. The key here is not as simple as just turning up the fuel, but to provide the correct amount of fuel where it is needed to achieve the optimum AFR. Look at the following remap of a 3.2 motor at WOT as an illustrated example. The engine is a fresh built 3.2 on Superflow engine dyno measured with an industrial grade wideband system, so the results are very accurate:

http://forums.pelicanparts.com/uploa...1150484953.gif

http://forums.pelicanparts.com/uploa...1150485162.gif

<i>So to clarify the statement you read, if you need more (xx) fuel at WOT for modifications or displacement increases, it must be done by reworking the WOT maps, as there�s no other way for the DME to know what the additional air flow increases and fuel requirements are.

The key here is not as simple as just turning up the fuel, but to provide the correct amount of fuel where it is needed to achieve the optimum AFR</i>

Ok. Sounds logical.

So if you wanted to adjust your AFR richer at WOT .....
you can monitor your AFR, say with a wide band display, and if you have enough fuel psi, and a rising rate fuel regulator, you might be able to achieve your objective provided the fuel demand is within the limits of the present system ?

Steve?

Yes.

Thank you Steve.

Does anyone know the limitations of the stock 3.2's fuel system in terms of maximum flow (per cylinder).

Steve,
Thanks very much.
What is disturbing and appaling is that the air flow meter "pegs" at considerably below the air processing capability of the stock motor.

IMHO - this is an unforgiveable product development decision on the part of Porsche - certainly they knew the AFM was limited.

What "saves" the situation is that the situation is "fixed" by setting the maps to run rich so that the motors don't self destruct.

Likely, I guess, is that for minor mods... e.g. cat out and 964 cam the engine will still be fat above 5000 and while leaving hp on the table I have a motor that still doesn't run lean causing the balance in my account transfering to the PelicanParts account.

Another thing that occured to me, trying to get more power by adjusting the fuel quality switch to a richer setting may in fact cause the reverse affect by going from rich to way too rich. I think I will check today to see that I'm at the zero setting.

Steve, also...what brand model or attributes should I look for if I want to get a wide band AF reader considering not wanting to spend a lot and also I don't think I need laboratory accuracy. I would be curious to find out how rich/lean I am at different operating points.

Thanks again for your time.

-Henry

Steve,
Just looking at your AFR graph shown in the bottom of your
dyno results for the 3.2 87:

I have a 3.2 87 that is basically stock we will say for the moment, and observed my wide band readings (from WMS). My readings seem to be different than yours. For instance, in loop it is typically 14.5 nominal and out of loop never below 12.
Any comments?

For a wideband AF reader, I'd recommend the Innovate Motorsports LM-1, as it is very easy to use, accurate, and expandable if you ever desire more data capabilities. There is also a less expensive LC-1, but for a display, you would either need a separate gauge or laptop. Wideband systems aren't usually inexpensive because they use a special type of sensor that requires a separate controller to operate it and provide a output.

There can be several reasons for AFR variations among different engines, even of the same configuration. First is the accuracy of the wideband you are using, which we shall assume for purposes of this post to be accurate. Having logged and examined dyno charts of many 911 engines, it never ceases to surprise me the amount of variation I see, even among stock to almost stock motors. It is usually the better maintained motors that are more consistent and read a richer AFR curve, while the poorly maintained ones read leaner (such as some Motormeister rebuilds I experienced). The standard deviation is as much as 2 full AFR points. This can be attrubuted to many factors:

1. Wear and inaccuracy of the AFM sensor. Wear on the resistance track of the AFM will cause dropouts and a reduced output voltage, leaning out the overall mixture. A loose spring tension from heat or age on the other hand will cause a richer low and mid rpm mixture. Renewing the meter by moving the wiper arm up or down will restore the resistance track accuracy.

2. Reduced fuel output from clogged injectors, fuel filter, or weak fuel pump.

3. Low system voltage thus not supplying enough power to open the injectors properly and drive the fuel pump

4. Deviations in the fuel pressure. While the standard fuel pressure of a 3.2 is supposed to be 2.5 bar, they can vary as much as 3 psi from norm. A new one a customer purchased from the dealer measured out at 2.72 bar, resulting in a richer than expected mixture.

5. Intake system air leaks: air leaks past the manifold to head gaskets, injector O-rings, vacuum hose leaks, and rubber manifold sleeves will all contribute to air not passing through the AFM sensor, thus the barn door not opening up as far as it should, resulting in a lean mixture.

6. variations in the intake air temp sensor or head temp sensor from spec

7. quality and amount of alcohol in the fuel used: oxygenated gasolines today have as much as 10% ethanol in them - mixture varies by season and region of the U.S. or world you are in. As alcohol has only about half of the specific output as pure gasoline, it requires roughly about twice the amount alcohol per c.f. of air to produce the same lambda. Thus a 10% ethanol mix is only supplying 95% of the fuel of pure gasoline, with the result being much leaner mixture than intended. Cars using race fuel or from European countries such as the U.K. often log richer than American cars because I believe their fuel does not contain oxygenates.

8. Cam timing

9. Engine manufacturing tolerance that account for flow deviations from motor to motor. The greatest deviation on Porsche motors are in the 5000+ rpm region. There are many duds out there, and occasionally you get a ringer that just jets.

10. Fuel quality switch setting

11. WOT switch activation. I see a high percentage of cars where their WOT switch does not activate either because the throttle linkage is not pulling it open all the way, or the switch is misadjusted - even on cars that are owned by Porsche mechanics that should know better. I used to get dyno runs where as many as a third showed the car never got out of part throttle mode, result being a 20+ hp loss.

12. Engine and exhaust modifications obviously.

It is because of all of this that for peak system performance, all of the above must be maintained properly. They are also part of the reason the factory fuel ratios are programmed as they are - insurance for all the potential deviations. That is why for maximum performance, tuners choose to have their chip or EFI systems custom tuned to their dyno and AFR plots, as it is the only equalizer for all of the above and ensures maximum power and reliability.

Steve,
I appreciate your detail of what can raise or lower AFR.
I wonder what Porsche's spec is for AFR in the loop? I would presume something to keep "everyone" happy: the fuel miser's, the liberals and naturally their performance minded enginears.

Closed loop AFR is always 14.7:1, or Lambda=1

Reference your AFR graph shown in the bottom of your
dyno results for the 3.2 87:

was from closed loop or WOT?

Dyno runs to measure for full power are always done at WOT, and so is open loop.

Oh. You were doing pedal to the metal.

I think I would would want to see what the readings (for various items) would be in closed loop, a full transition up to WOT, and then WOT.

OK, so your graph was entirely WOT.
Thank you.

Henry,
Chris is right, Great question. From Steve Wong, even better answers.
Thanks Guy's,

Steve,
just to clarify...

you said a few posts back that the lamda was always 1 and the AF was 14.7... this I guess assumes the O2 sensor is in play and we are not on the WOT map.

can the AFM door hit the stop and still be on the "regular" map...and then it could use the O2 sensor to know if it is getting too rich... does this happen? Maybe this would be a good reason to disconnect the WOT switch and make some use of the O2 signal to not run way rich or way lean... why doesn't this work?

I thought that the AF could be adjusted up and down a bit using the "trim" of voltage against the O2 EMF...I think you confirmed that this is how it works but now not sure again.

you also say that the AF can't be altered to run very far off the map.. so I guess that is what you mean by the AF is always 14.7 when not on the WOT map? or it is close to that.

That may not be a very clear set of questions. Sorry, I'm trying to understand a bit better and then make another diagram.

When you install your high tech. air flow meter do you use the O2 sensor and trimming to get the motor to run at 12.6 or there abouts on the new "Steve" WOT map?

My breakout connector and box should be here in a week or so. I'm planning to hook it up and view the behavior of some signals e.g. O2 voltage and I have a connector set coming to make a breakout for the injector ground wire. I'll attempt to see the PWM changing on an oscilloscope while driving around. I looked for a chip that can look at a signal and out put PWM value but didn't find one yet... would be a handy gage to look at to confirm how all this is working. I find chips to generate PWM but not report PWM.

Thanks,
-Henry

So hcoles and I just played around with aigel's LM-1. Last time I borrowed it, the best I could get with the Wong chip in my '86 Carrera was a peak low AFR of 13.5 at WOT Looked at Steve's recent list of things that can cause this .... decided to replace the two simplest things that I knew needed replacement - fuel pump and filter.

Today, after replacing those items, we did some more testing .. good results. On my car, I got a consistent 13.0 at WOT from 2500 RPM up to about 5200 RPM. At that point it leaned out a bit to between 13.5 and 13.8 all the way to 6000. MUCH better!! I'd like to get 13.0 all the way to redline - but based on what Henry thinks, I don't know if that is possible. I could send the injectors - they've never been serviced after 20 years and 160K miles. Should I even be concerned about this variance?

On hcoles '89 Carrera with the stock chip, the AFR graph looks just like it should. At WOT, AFR starts off at 13.0 at 3000 RPM and peaks down to 10.89 at 6000 RPM!

JB

another thing we noticed...on my car with the stock O2 sensor out and the wide band sensor in and just connected to the reader...the car runs very good and the AF is very close to 14.7 without the use of O2 sensor. I think it was even running a bit richer which made the idle smoother. Steve Wong's reports of the stock chip running very rich at WOT and above 5k, or maybe starting a bit lower, are correct.... I do sort of like the feeling that at least I'm not running lean....but I'm throwing away power and gas. I would like to see the car run at about 12.5 or so...just a bit rich... I think Steve says max. power on these cars is at 13AF.

At steady state loads, the O2 sensor tries to keep the mixture at a lambda of 1. Stock part throttle mapping for the most part is designed to keep the O2 sensor happy at lambda 1. At higher loads and rpms, it is not, nor do you necessarily want the mixture at lambda 1, as it would be too lean, potentially causing damage, and not the optimum ratio for maximum power. Thus after a certain rpm and load point in part throttle operation, the O2 sensor is ignored, however the part throttle mapping still remains in effect. So disconnecting the WOT switch and using O2 trimming will not work. You will also loose at least 20 hp, as part throttle timing maps are not optimized for maximum power, and the part throttle mapping under high load is optimized for acceleration enrichment, and not full power. Part throttle mapping also looses resolution after 6000 rpm, so the AFRs after 6000 rpm will not be optimum and power will fall off a cliff.

On the new carbon fiber HFM5 system, just as the stock system, the O2 sensor is ignored at WOT. WOT mapping is determined either by past statistical data, or ideally from fine tuning off a subsequent dyno or LM-1 AFR chart.

Jim, the 13.0 WOT you achieved is very much where I target for, however the 13.5-13.8 you're getting from 5200-6000 is leaner than what it should be. There is probably another factor or two limiting your fuel output. Whatever you try, if you're not able to correct this, you can always have your chip retuned to provide the additional fuel where you need it.

Steve,

Thanks for the explanations.
With your new carbon fiber assembly assuming it can read air flow as high as needed...doesn't the DME then have all the info. it needs to know the mass of air coming in and therefore not need "tuning" as it could have e.g. 12.8AF on the WOT map for a given rpm. ?

-Henry

Hi,

Just as additional info about O2 sensor operation:
The sharp "jump" in the voltage output of an O2 sensor is not actually a property of the sensor itself, but is an effect of the chemistry of the combustion process.
The voltage output of an O2 sensor is inversely proportional to the partial pressure of oxygen in the exhaust and to the sensor temperature. At stoichiometric (Lambda 1.0 or 14.7 AFR for standard gasoline) this oxygen partial pressure jumps from a few pascals to something as low as 10^-15 pascal over a span few hundredth of AFR (going from lean to rich). This causes the jump. The middle of the "jump" is at about 0.45V O2 sensor output (about 10^-5 pa O2 partial pressure). The ECU just looks at that voltage in closed loop and gradually leans out when higher, and riches up when lower. The result is that the measured AFR oscillates around 14.7 2-4 times/second. This oscillation is also needed for correct cat operation. The cat oxydizes CO and HCs to CO2, H2O and H2. For that oxydation it needs additional O2. This additional O2 comes from the reduction of NOx to N2 and O2 and from the excess O2 in the lean phase of the oscillations.
Older 2 way cats (Thermal reactor cats, no NOx reduction) required additional O2 to be pumped into the exhaust via an air pump as on 80-83 CIS.
Because of the sensitivity of O2 sensors to exhaust Oxygen, even a small, pinprick exhaust leak before the O2 sensor can cause the ECU to run the engine way too rich in closed loop, resulting in failed smog tests and lousy milage. The additional O2 from the leak causes the O2 sensor to read excess O2 (lean). This causes the ECU to richen up, trying to overwhelm the O2 leakage. I have seen engines run actual 9:1 AFR in idle closed loop (measured before the leak), smoking black, while the O2 sensor (behind leak) indicated stoichiometric.
That's why many ECUs have a limit to how much they correct by O2 sensor, otherwise even small exhaust leaks could possibly cause engine flooding.

Regards,
Klaus

Henry, yes the HFM system does a good job of measuring for changes and variations in your motor beyond the flow capacity of a 911 motor. The HFM sensor is much more consistent than the mechanical flow meter which have lots of variation from car to car. However it can't compensate to variations in the system such as fuel pressure or injector flow variances because of the open loop zero feedback condition, so optionally fine tuning from AFR data is the cherry on the pie to ensure that everything is running in as optimal tune as possible.

Klaus - thanks for that very valuable bit of information. Just before your post, I could not figure out why a customer's car idled at what was around 13:1 closed loop according to the LM-1 no matter how much I attempted to lean out the base idle mixture in the flow meter. Open loop, the car's O2 sensor measured out at 0.9 volts and after my feeble attempt to lean it out, it read around 15.7 at idle, open loop. Under load, it appeared to be fine. It never smelled rich and all sensors were checked. I assume that any exhaust leaks will also affect the wideband sensor of the LM-1 and fool it into reading a richer than actual mixture, would that be correct?

It will affect the LM-1 wideband for sure. An exhaust leak will make the wideband read lean (it detects higher partial pressure of O2 introduced by the leak). In closed loop the leak will cause the ECU to richen the mixture. The net result is a rich-running car...

I have been playing with my 84 Carrera DME and tuning it. While learning about how it works I decided to log (real time) the AFM signal VS RPM at WOT.

The run was done in 2nd gear starting at 2000RPM and going WOT up to 6000RPMs. I also know I'm in WOT mode because I also log and record the WOT switch and it showed closed/grounded durning the entire run.

In this chart I plotted the RPMs against AirFlow, air flow was measured by recording the AFM barn door position (0-5volts) and then converting the value to a % door opened. 0-100% 0 being closed and 100 being fully opened.

http://forums.pelicanparts.com/uploa...1189016286.jpg

I wanted to share this because this thread (as well as other threads) say that the AFM bottoms out at 4000RPM while in WOT mode. I mean no disrespect to any of the many smart folks on this board but my realtime testing shows diffrent.

So, is my car not working correctly? or maybe we don't understand how the AFM works 100%?

Just thought I'd share my findings.

EDIT: new information learned - it seems that some basic Motronic manuals state that AFM bottoms out at 4000RPM and this may be the case for some Motronic cars but not our 84-89 Carreras. For these cars the bottom out point for the AFM is around 5500RPMs as you can see from the graph above. This means that after 5500RPMs the DME will no longer utilize the AFM signal to calculate fuel, it will rely on RPM, WOT switch and intake air temp to calculate fuel.

the door doesn' look like it is opening to much more above 5500
I think this is what Steve Wong has said all along - when on WOT and above 5500 the injectors go full rich...I think that is what has been said and is the reason we run too rich
also the volts from the door to actual mass airflow....the algorithm for this is somewhere in the EMU.. likely it is not % of 5 volts.
the interesting thing is if you can record the pwm percentage for the injectors along with this...but I think that is hard to do.. I looked for a chip that would "listen" to a pwm signal and output the percent open time.. but couldn't find that animal

Great interchange of information! Thanks to Steve and all for sharing your experience with us.

If the VAF (AFM/Barn Door) is maxing out over X RPM, why not upgrade to a MAF system with the proper software that understands Air Mass?

yes, by all means... I think Steve W. sells a system that has all that figured out. Actually measures the maf up to the limits of what the engine can eat and adjusts the AF to what you want... there's a concept... that apparently was missed back in 1983 or so by Porsche.

I don't think I understand what you mean by "measures the maf up to the limits of "?

If the transfer function inside the DME software is rewritten to understand the MAF signal there should be no limit on the MAF signal.. We have used the same MAF sensor with the appropriate DME software, on a 400rwhp turbo engine, as well as a stock 911.
I'm not a regular on this board, feel free to email me directly if you have any questions.

on e.g. a 3.2 with Motronic... after 5500 rpm at WOT there is no new maf information from the maf device (the maf device is pegged)...it can only be infered using rpm and other factors, e.g. air temp. This is what I mean... and I think what Steve's product does is properly measure the maf up to the limits of what the engine can eat using the new and improved maf device. To put it in other words the system is not a great design at the higher rpm and power levels. This is why the Steve Wong chips work.. he tunes out the richness, amoung other things but I think that is the major one.

when I say Steve's product I refer to
http://www.911chips.com/HFM1.htm
This is what the factory should have done IMHO but the technology may not have been there quite yet.... If I had an extrie $2,700 I would buy this in a second.