.::Lee::.

Well, if curiosity killed the cat than I am doomed to a curious death!

For quite some time now, I have wanted to gain a better understanding of just how my Honda works. I'm a firm believer that the Honda engineers were and are very intelligent people and they designed our cars to run how they do for very important reasons. However, without coming up with a great deal of bribe money, I may never be able to hear the proprietary design concepts straight from the horse's mouth.

So I decided the next best thing would be to gather my own data and share it with others so that together as a community we can analyze the significance of it and hopefully be able to more fully understand our cars. I do not intend to delve further than is necessary to be able to implement the data. However, I do intend to keep the information as scientific, unbiased and very importantly unfettered by opinion if it in no way assists in this project.

I have a few basic projects planned out, and as time goes on I would like to extend my studies out into other areas, but I don't want to get in over my head to start with

So here's what I plan to start with:

Air/Fuel readings of the factory Honda:

I am going to datalog the A/F ratios of both a stock 3G Acura Integra LS and a near stock Honda Civic DX (Ex intake mani, exhaust) in order to understand how these cars act in vacuum utilizing factory injectors and factory ecu's unmodified other than to provide datalogging capabilities.

Both of these cars will be tested using PLX M300 widebands with the O2 sensor located in the factory location (primary o2 location in the case of the Civic). Datalogging will be done using CromePRO for the simple reason of availability and also the data layout of this software. Data is recorded into individual cells using CromePRO and hopefully will be more useable in this state. It will be impossible to fill all cells whereas some of them are theoretically impossible to fill. I will do my best, however, to fill as many as I can on the street (unless some local wants to donate some dyno-time

Both of these cars will be turboed in the near future and additional data will be available as that takes place.

Under-hood temps and EGT's of the factory Honda:

Utilizing the same cars, I am going to datalog under-hood temps in a variety of locations in conjunction with ambient air temps and other engine data (most likely RPM, IAT, and MAP data). This will be done using a datalogging unit and K type thermocouples. I am undecided as to where to place the thermocouples but will keep locations consistent throughout testing. Tentatively I had planned to place one within the fins of the radiator, one offset slightly from the exhaust manifold, one somewhere near the intake manifold, and also another in the intake tract (to compare with IAT data from the ECU). I am very much open to suggestions on thermocouple placement.

This is the datalogger I will be using. Eight analog channels and four digital.



I am still in search of some good thermocouple amplifiers, so if anyone has any ideas (or even better hookups!) I am open to them.

DataQ has a nice 8-channel unit specifically for thermocouples; however, I am not in a position to pay nearly $900 for one.

In conjunction with the under-hood temperature testing, I will be placing an EGT probe (same thing as a K type thermocouple) in each of the runners of the exhaust manifold. I will be looking for any variances that may exist out of the factory, but more just to get a general idea of the EGT's from the factory.

So why is this in the FI forum?

All of the temperature and EGT testing will be done again as both of these cars are turboed. The same test will be done on a variety of cars local to me running a variety of different FI products and cooling system setups. I hope to standardize sensor placement as much as possible to make the data reliable.

I have plenty of other ideas of future testing to do (most dealing with FI) but I will save those for another day. Please give me any feedback that you may have and suggestions for future projects or even how to set up those at hand.

Thanks,

Lee

mrx

great idea... curious about the results... go on

.::Lee::.

I'm going to begin A/F testing today. Will report later!

RACEPAK

keep up the good work....

Bailhatch

Sounds interesting

M.A.R.C.

Awesome.. good job.

N20civicB18

well the DX car with EX parts will most likly run a little leaner then the factory intended... and are you adding in a seperate o2 sensor bung and running stock pcm's because if you rememove the stock o2 sensor im sure that will toss the readings off? no?

mrbsponge

i would like to see the results as well.

from our testing of a stock civic, it ran pretty lean in high gear part throttle like in the 16:s

i know i idle my car in the 15s it seems to like it there plus i know there is no load on it

ToasterOven

thats why i love Honda right there....people love them so much that they will datalog worthless ( to everyone else ) information, just for the knowledge of how the car runs. And these kind of people choose Honda just like the others.


I'm so jealous!


I would also like to install wideband sensors in each cylinder a few inches off the head. Or if that is unreasonable maybe just 4 EGT probes and 4 readouts....


my hat's off to you

Muckman

iTrader: (2)

I am very interested in this information. I hope it doesnt become unfinnished like so many other comparison and testing threads started here.

.::Lee::.

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by N20civicB18 &raquo;</TD></TR><TR><TD CLASS="quote">well the DX car with EX parts will most likly run a little leaner then the factory intended... and are you adding in a seperate o2 sensor bung and running stock pcm's because if you rememove the stock o2 sensor im sure that will toss the readings off? no?</TD></TR></TABLE>

Interesting point. I have been going back and forth in my mind of whether that would make it lean or not. That's why I'm so excited about all of this is because I will be able to stop speculating.

As far as the OBDII civic, with my limitid resources what I will have to do is to run an OBDI ecu so that I can datalog the other sensors to make the A/F info useful. It the case of this car I will run a stock DX map in two different conditions. One in full open loop so I only get straight information from the cells of the map. Secondly, I will run closed loop but using the narrowband output to monitor how the ECU adjusts. This will not be a perfect test in this instance because of outside factors, but that's where the Integra comes in.

The Integra (which I will begin datalogging in about two hours ) will be with the emissions apropriate ecu, stock program, monitored in both open and closed loop situations.

Anyways...I'll keep everone updated as I gather the information and it is in a complete enough state to present.

V-SPEC 1

subscribe.

shortyz21

yes. dont let us down, prof

AbitAvenger

sweet idea

DENCIO

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by V-SPEC 1 &raquo;</TD></TR><TR><TD CLASS="quote">subscribe. </TD></TR></TABLE>

.::Lee::.

OK, this is gonna be a long one, but I want to be as thorough as possible. I know a lot of ppl aren't expecting any of this to actually happen, and I question myself whether I will ever be able to do all that I would like to with this project. However, I feel really good about accomplishing at least what I have listed out up to now.

The good news is I have already begun gathering data. Also it looks like I will be able to use a dyno in order to solidify some of the information and to hopefully create a 2D lambda chart. For now I will post up what I have.

Tonight we (Brendan aka Veggiemaster and I) began by datalogging the stock Integra. We used a PLX M300 and had it calibrated within .01 Lambda or around .1 A/F point. We used Crome Pro to log into cells. All of this was done at aprox. 4500 ft. altitude (Utah). Because of this you'll notice that the ECU never reads from the last three columns. However, using the atmospheric pressure sensor, the ECU is able to compensate, thus you may move all of the data to the right aproximately 2 columns for sea level equivalent.

This first screen shot shows the relevant A/F reading that were recorded. The cells reflect an average of all readings done within the given cell. This means that there are some variations due to load differences and slight gear correction however this was all done with closed loop disabled so as to avoid any o2 correction. I will point out some of the more substantial differences a little later on.



Crome Pro allows you to select a target A/F ratio for any given cell. I made a table that reflects values that a popular chart on the internet shows to shoot for (shown below).



Now, this next shot shows any substantial variances between the actual A/F's and the target A/F's.



I think this shows something very interesting. Honda created fuel maps with varying A/F's throughout the RPM band. In other words, tuning any particular column to one set A/F will make the engine run differently than Honda intended.

Crome Pro also has an interesting feature that filters out statistically impossible data using an algorythim written into the software. This picks out data that occurs from wideband latency, unburned fuel that gets dumped, decel conditions etc. I post this only as an addition to the raw data, but it should give a better idea of what exactly is going on.



Then the A/F differences based on filtered data.



I hope Brendan will post up any insight he had while datalogging tonight and maybe we can get some discussion going about the current info.





Modified by Ibiza at 11:33 PM 10/1/2005

.::Lee::.

Some things worth noting:

In columns 2-5 in the 7000 RPM row we noticed A/F's of as low as 10.8~9 on the wideband while driving on a level road with very little load. However, whenever we were climbing a hill the same cells produced ~12.0 A/F's. This was tested several times and shown to be consistent.


Column 3 at the 2248 RPM row is acurate data. We tried over and over to see if it was a fluke that the car ran over a half a point richer in between two other cells within the column. I have no explanation for this whereas the fuel maps themselves are consistent in this region.

The 4000 RPM row seemed to consistently run richer than the surrounding row. I would guess that this is close to maximum volumetric efficiency and the torque peak. This may be a measure to quench any signs of knock in this region.

I beleive that the 6000 and 7000 RPM cells of the 4th columns are inacurracies of the averaging process. I beleive they actually fall right in line with pattern that otherwise exists in those two rows. (i.e. 6000 RPM should read ~12.7 and 7000 RPM should read ~12.35)

If someone would like to help out it would be awesome to makes some 3D graphs of this data in excel. It might help to visualize it better.

.::Lee::.

Does anybody know a good place to get thermocouple amplifiers or know of a good unit to use? That's my big hang up right now with getting some of the other testing going.

0xenocron

I think your logic is a little flawed. I dont think P30 code with P75 maps is anywhere near stock. I dont care if you match the rpms and map heading up perfectly it will still run differently, import, etc...

You need to log a STOCK unmodified ECU with a wideband if you hope to get anywhere near the results you are looking for. And then you need to somehow access the data that the ECU stores for fuel trim top analyze that as every engine (even if its stock) while trim differently.

I myself have thrown P75 maps into P30 code and got very different results from you on my swapped B18b. However...my setup is far from stock besides the injectors so I cant say with any certainty what my results show or dont show.

I do applaud you on the initiative to do this however. Keep with it...

sharkcohen

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by Ibiza &raquo;</TD></TR><TR><TD CLASS="quote">Now, this next shot shows any substantial variances between the actual A/F's and the target A/F's.



I think this shows something very interesting. Honda created fuel maps with varying A/F's throughout the RPM band. In other words, tuning any particular column to one set A/F will make the engine run differently than Honda intended.</TD></TR></TABLE>
What it shows is the difference between running in factory closed loop and open loop vs your target a/f setup. There's nothing remarkable there.


Modified by sharkcohen at 1:00 PM 9/30/2005

.::Lee::.

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by sharkcohen &raquo;</TD></TR><TR><TD CLASS="quote">
What it shows is the difference between running in factory closed loop and open loop vs your target a/f setup. There's nothing remarkable there.
Modified by sharkcohen at 1:00 PM 9/30/2005</TD></TR></TABLE>

I'm not sure that it is pertinent to the purpose of this thread to defend my understanding of the fundamentals of EFI per your deleted comment (Thanks for pointing me to where I can learn the basics ) but suffice it to say that no matter how much I know or understand there will always be more to learn.

I think you're missing the point altogether of why I'm doing this. This is not research in the sense that I am going to discover anything new. I don't hope to stumble upon some amazing discovery that will change the world of Honda or EFI forever. I am merely trying to qualify and quantify information that will help both myself and others understand more fully the operation of the engine and clear up some of the confusion and misunderstanding of such.

So was there anything remarkable in the data shown in that picture? Did I find out something that I did not already know? Not exactly. However, rather than to just say that it is incorrect to tune for a single target lambda throughout a whole column as many people mistakenly do, I prefer to show why not through solid information rather than clouded opinion and hopefully provide a better way of doing it while I'm at it.

Lee

sharkcohen

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by Ibiza &raquo;</TD></TR><TR><TD CLASS="quote">However, rather than to just say that it is incorrect to tune for a single target lambda throughout a whole column as many people mistakenly do, I prefer to show why not through solid information rather than clouded opinion and hopefully provide a better way of doing it while I'm at it.

Lee</TD></TR></TABLE>
Much of the variation in the columns you are seeing is from oscillating fuel mixtures during closed loop as the ECU hunts for .5 V from the O2 sensor. So it's a fallacy to conclude that one's fueling should vary like that throughout a column from your observation. You're trying to reverse engineer something that has already been reverse engineered, and you are drawing inferences that are not quite correct. Just because it does what you see does not mean that is the way it should be. What you are seeing is the best anyone could design for 1 V O2 sensor feedback.

.::Lee::.

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by sharkcohen &raquo;</TD></TR><TR><TD CLASS="quote">
Much of the variation in the columns you are seeing is from oscillating fuel mixtures during closed loop as the ECU hunts for .5 V from the O2 sensor. So it's a fallacy to conclude that one's fueling should vary like that throughout a column from your observation. You're trying to reverse engineer something that has already been reverse engineered, and you are drawing inferences that are not quite correct.</TD></TR></TABLE>

Not the case. All these measurements were done with closed loop disabled. The variations that were recorded within individual cells were to due to load and gear variations. These will be isolated when I continue my testing on the dyno where I can lock in these factors easier. The figures that are shown thus far were averages based on a great deal of driving in every day street conditions. Each cell was visited repeatedly in order to produce the broadest variety of readings possible and then averaged.

While the data presented up to this point is not meant to give exact figures for a particular load or gear set, it is meant to show where on the lambda scale the car tends to run in a given cell.

I think you're looking for something that I never intended. I am currently preparing for a little bit more comprehensive testing, however, you start with the bread and move up to the meat. However, as simple as this data might be up to this point it may not be quite as transparent as you think it is.