SD-8

For tuning does it make a difference if you use CanScan or Hymee's Scanalyser? After my RB intake get her I am finally going to tune my CZ unit and I was wondering what software I need to buy so I know what data to feed into the Greddy Emanage.

Thanks!!!

Nemesis8

iTrader: (3)

I'm using both currently. Be careful which Canscan you get. You need the one that was co-engineered by Maurice. If you buy a used Canscan, get Maurice to upgrade it for you.

r0tor

iTrader: (1)

thats actually a tricky question...

last time I checked the a/f ratio on the CZ version of the Canscan doesn't match the a/f ratio of the regular Canscan because of a correction curve - and then those two don't match what you would get out of the Scanalyzer as its giving people normal a/f ratio's in the 11's and the Canscan bottoms out at 12.0. Yet, they all should be looking at the same value.


I used the cz edition canscan, set my target a/f ratio for around 13.4:1 and have no problems on premium gas - which seems right as UNiChip supposedly targets 13.5:1.

With the regular canscan i recall people have pinged their engine at an indicated 13.0:1 - which makes no sense.

I'm not even going to try to guess what will happen tuning with the Scanalyzer as i've seen people post their a/f ratio stock is 11.6:1 while my cz canscan said it was 12.6:1.

:o

SD-8

Thanks for the responses.....Canscan it is.

Hymee

iTrader: (1)

Guys, you should be tuning using Lamba readings (called Equivalence Ratio in the OBD specs and sCANalyser), not AFR. To get AFR you multiply the Lambda value reported by the O2 sensor by the stoich value for the actual fuel being used. If you use AFR you must know what your stoichiometric ratio is otherwise your readings will be inaccurate. If you just use "14.7", you will not be spot on, and perhaps get the types of discrepencies noted above. If you tune to a target LAMBDA value, it doesn't matter what the actual stoich value of the fuel is. Tuning units such as Motecs unit use a LAMBDA METER, not an AFR METER.

I've done a comparison with the stock O2 sensor with two other sensors and found it to be totally accurate... see my post here regarding the "calibration" of the stock O2 sensor, as I posted a long, long time ago... https://www.rx8club.com/series-i-tech-garage-22/sneek-peak-new-hymee-enhanced-product-39583/#post863154

Cheers,
Hymee.

BTW1: I don't know how many different "stoich" values are programmed into canscan, and what other "fudge factors" have been applied. There is no such thing in sCANalyser because it isn't necessary. If you want to use AFR in sCANalyser the stoich value is user configurable, in one setting.

BTW2: The O2 sensor can't be crap, otherwise how would it be CARB legal?

Nemesis8

iTrader: (3)

Hymee is right on this one. Once I receieved my sCANalyser, I have been very happy with it. In fact, I will most likely be selling my Canscan soon.

r0tor

iTrader: (1)

with all due respect, i see no difference between using lambda or AFR (lambda x stoich)... you can't put anything in the tank except gasoline, so stoich is a constant 14.65 no matter what. It is so much more convienient to tune while looking at AFR and comparing it with the target afr you wanted then to figure out your targeted a/f ratio and then use the stoich factor to convert that into a lambda so you can comprehend what your lambda guage is trying to tell you. I really fail to comprehend why one way is more "accurate" then another.

As for tuning a Motec - last time I tuned one they tell you to aim for a certain AFR, then give you the corresponding lambda so you can compare it with whatever O2 sensor you are using.


Also, as far as I have ever been able to figure out supposedly the cz canscan correction curves were based on a graph of lambda vs voltage as the curve becomes very non linear under a lambda of .85, but the ecu only has a linear function in it because the car is only using it during closed loop conditions where lambda is ~1. It is not different stoich values.

this is a typical O2 sensor output...

sco

With all due respect, you can't be sure that your gasoline is the stoichiometric ratio you assume unless you test it. The stoichiometric ratio will be affected by the blend. The point is that if you use lamba (aka equivalence ratio) then you don't have to know what the stoich ratio is.

The graph you posted is of a switching type sensor, which the Wideband O2 sensor in the RX-8 is not. The raw OBD II data reported by the PCM also reports a current reading, not a voltage reading for the O2 sensor.

r0tor

iTrader: (1)

1) So your telling me there is a difference between looking at lambda or looking at lambda x a constant?? Thats like saying there is a difference in my cz by tuning it using the rpms/2 setup it comes with rather then rewiring to read full scale rpms. The AFR displayed is a scaled up lamda - its always going to display on the display as lambda x 14.65, and there is no difference if you say you are looking to target 13.5 afr and do so by reading the calculated afr of lambad x 14.65 or devide 13.5 by 14.65 and look at your lambda guage.

13.5= (13.5/14.65)*14.65 !!!


2) If you truely believe that our O2 sensor behaves entirely different then the simple example I posted then by all means please post the output of our O2 sensor then. I have the calibration records of 2,000+ sensors of probably at least 30 types from my workplace and I haven't come across a single one that is linear across the entire range - thats actually how you size a sensor in the first place by looking at the sensor's curve and finding its linear zone and then finding the one that will fit you operating zone. In the case of an automobile, the operating zone is going to be ~14.65.

I can point to several posts where people using your lambda are claiming lambdas at WOT that correspond to an 11.6 AFR and I can also point to every single dyno in here and not a one has an 11.6 AFR recorded as they are all above 12... something is not right about that picture, I have argued this before and got no answers and I will now stop because i will still get nowhere this time either and its completely not worth my time or energy.

sco

r0tor,

Before I start - this is not personal. We're having a discussion here based on facts and I am as keen as you to understand exactly what is going on here.

With regards to point 1... I said nothing of the sort.

You said, and I quote
I replied and said that you can't assume stoich is any value, because you don't know what the blend of your fuel is. The only way you can determine stoich is to test the fuel.

The point still stands... if you don't know what the stoich ratio is for your fuel, eliminate from the equation to maximise accuracy.

With regards to your second point...
I never said O2 sensors are linear. I simply said the graph you posted appears to be that of a switching type O2 sensor. I understand that Wideband O2 sensors are typically more linear across a wider range which makes them more accurate than switching types outside the area close to stoich.

I don't know if the RX-8's current to lambda conversion is linear. I'll find out. I'm currently in the middle of some changes to sCANalyser that will allow us to log both current and lambda at the same time (currently you can only log one or the other not both). As soon as they're done (hopefully in the next few days), I'll log current and lambda, graph and post the results.

Now with regards to your final paragraph. It is not "our lambda" -- we are reporting the data as reported by the RX-8 PCM converted as per the OBD II specifications.

Hope that clarifies a bunch of things... will follow up as soon as I can with the promised graph.

Cheers

sco

two rotors

Well according to the Mazda Publication Part No 9999-95-102F-04,the front O2 sensor in the RX-8 has sort of linear mA output over a range from0.7 lambda to about 1.4 lambda and is a pump cell type.The rear O2 sensor is a zirconium element which exhibits significant emf change about stoichiometric A/F.(this is shown in rOtor's post above.)What non Mazda people do not appear to know is what the ecu does with the raw data from the sensors--at least I do not.
By definition lambda =1 when A/F is stoichiometric.Where does fuel quality come into the equation?The sensors measure only O2 cocentrations.Fuel blend may affect how the fuel burns,energy released etc.

sco

Any chance you could post a scan of that page?

Again my point (and Hymees) as that you should be using lambda not an AFR based on an assumed stoich ratio... so we're in violent agreement :p

I'll get some data on sensor data versus reported lambda as soon as I can.

two rotors

My point is that effectively it doesn't matter whether one uses Lambda or A/F since the only ACTUAL measurement made is O2 concentration in the exhaust gas.

sco

two rotors - Thanks heaps for that scan

Nemesis8

iTrader: (3)

So, for 13.6 AFR I should tune for .9295967 Lambda on the sCANalyser assuming I use 14.63 stoich, right?

sco

That looks right to me.

r0tor

iTrader: (1)

thanks for that info two rotor


and look at that, a linear approximation doesn't work under ~13.0 - so unless Mazda out of the goodness of their hearts included a correction curve in the ecu for ranges they don't need for the O2 sensor it looks like Maurice might have been on to something about a year ago.

and if you look closely the voltage for 11.6 on the linear curve corresponds to around 12.4 on the real curve - which would explain why people with the hymee software are claiming their a/f ratio's are so low

sco

OK, curiosity got the better of me. I did a little hack to log current vs lambda... see graph below for the results.


This is straight data from the PCM... no lookup tables in sCANalyser . just conversion from binary to numeric format according to the OBD-II specifications. The graph looks to me to be very close to what was in the page posted by two rotors - i.e. the conversion is not linear

P.S. You will be able to do this yourself in the next major release of sCANalyser.

Edit: I changed the crossover point of the X axis, changed grid lines to dotted and removed the legend to make it easier to read

Hymee

iTrader: (1)

I sometimes wonder why I try explaining stuff here. People argue the un-arguable and try to make an argument about newer current based wideband O2 sensors using data about old voltage based narrow-band O2 sensors.

The PCM reports both CURRENT and LAMBDA, as Sco has plotted above. And low and behold, it looks near exactly the same as the diagram in the Mazda manual. Let me say I am not surprised.

Do I need to shout it out load? "THE PCM CONVERTS THE CURRENT READING FROM THE SENSOR TO THE LAMBDA VALUE, AUTOMATICALLY".

And, with all due respect, r0tor, all gasoline is not 14.65 stoichiometric. It does actually depend on the exact blend you are using.

All I am trying to say is if you want to tune for 13.6 AFR, that is fine, as long as you know what the exact stoich value for your fuel is, otherwise you will be out.

If you want to tune for a target lambda (say 0.91 for arguments sake), you can measure that accurately with a lambda meter / gauge.

two rotor is spot on - the sensor is measuring O2 concentrations. That is why it is called a wideband / widerange O2 sensor. The "lambda" is the "units" it measures it in. AFR can then be derived ONLY IF YOU KNOW THE EXACT STOICH VALUE FOR THE FUEL. No wonder other software has fudge factors built in to try to get it right.

So, no matter what blend of petrol or gasoline you have in your car, LAMBDA DOESN'T LIE.

I think (hope!) sco's post hopefully puts to rest some FUD / myths about O2 sensor "calibrations"!

Cheers,
Hymee.

r0tor

iTrader: (1)

I can really not comprehend how dense you are in thinking that looking at a constant x lambda is completely different then looking at lambda. I do not care what fuel i am using because the CANScan is only going to give me a constant multiplied by lamda - which for the last time is in all reality is just lambda but in units that means something to me!!!

and I have still yet to hear how every dyno shows open loop afr ratio in the mid to low 12's and the sCANalyzer is giving people 11.6's - if open loop truely was in the 11's then there would be practically no need for fuel control with a forced induction setup!

r0tor

iTrader: (1)

that is good news then... thank you

sco

r0tor, you'll have to forgive me -- I haven't read a lot of the threads on Greddy tuning... so feel free to point me to some relevant threads.

Let's go back to basics here and try and work out what is happening here. So we know:
1. The data reported by sCANalyser is very close to what is published in the service highlights manual.
2. The graph from both goes below AFR of 12.0 (using theoretical stoich of 14.7)
3. My testing has been on an Australian car (?fuel map differences in the US?)
4. I've triple checked the conversion in sCANalyser against the OBD specs - it is correct (i.e. I haven't mistyped a number in a constant used in the calculation). The correlation between my results and the service highlights graph seems to confirm this.

So can you fill me in on testing that has shown AFRs only going down to 12.0. Has that testing been done only with CanScan (CZ edition)? Has anyone tried the "standard" edition? Have similar results been seen with wideband O2 sensors inserted into the exhaust pipe pre cat?

r0tor

iTrader: (1)

my main concern is posts like this (and the next page) https://www.rx8club.com/showthread.p...&page=41&pp=15 of people saying that their a/f ratio's aren't even making it to .75 lambda (~11.0 afr) during normal driving conditions - ie not overheating the cat. I've seen this mentioned a few other places as well.

If you look at any dyno done on this forum, you typically do not get anything lower then ~12.2 at WOT. My personal car was around 12.5. My comment about the greddy was to put things into perspective since they need to add a bunch of fuel just to get under 12.0 to be safe.

Something just does not make sense about these readings or I am missing a something.

sco

Hmm... Rasputin has a european car, I have an Australian car. I really do wonder if it's due to fuel map differences in the programming of US cars. Given the current heat related problems in the US (engine replacements, recall) I wonder the same also.

sco

Here is a graph posted by Nemesis8 in this here: https://www.rx8club.com/showpost.php...&postcount=669


It's a third gear run on the stock fuel maps and a Revi intake and captured by sCANalyser. Notice his lambda gets down to about .78 (so approx 11.4 AFR). So that kills my idea that no readings below 12:1 is US fuel map related.