Tuning ITBs on a Racecar and what I learned.

[whittlebeast]

I wanted to try to document how I tuned my racecar in a hope that other people could learn from my journey. Here is what I came up with.

Sorry that it got so long.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

I have been working on an ITB build racecar this spring and want to share what I have learned. The build is a 1985 Honda CRX 1488 cc motor. It is running all the typical autocross prepared mods like torque cam, lightened crank and flywheel, Carrillo rods…. The ignition has been converted to 4 coils in full sequential mode with the stock dizzy used as a cam sensor.

Here is a few pics of the install

http://www.nbs-stl.com/CRX/CRX%20Engine.jpg

http://www.nbs-stl.com/CRX/CRX%20Ignition.jpg

http://www.nbs-stl.com/CRX/CRX%20Interior%20Wiring.jpg

Here is the wiring diagram that I used for the engine.

http://www.nbs-stl.com/CRX/Civic%20CRX%20Wiring.jpg

Tuning:

It car is not street legal, so all tuning had to be done in the driveway, on a dyno or on the autocross course. We started the process in the driveway. I took a wild guess at the timing. After verifying all the wiring with the help of the MS3 test modes, checking the base timing and I twisted the key. The motor started right up. At this point I was running pure Speed Density. I slowly brought up the RPM wile I let the VEAL process autotune the basic tuning. Everything went as expected and within a few warm ups and I was ready for a trip to the dyno.

At the dyno, I elected to start by having the dyno set at a very low load and rolling the throttle around letting VEAL do it’s magic. Then we moved on to letting the dyno hold a fixed RPM and rolled on the throttle. If a clear pattern developed in the fueling started to develop, I would take an educated guess if I needed to raise or lower the VE before we got there as a best guess.

We later bumped the target AFR and timing up and down doing WOT sweeps, looking for clear patterns at what the motor wanted for timing and AFR the motor really wanted to make power. This is what we ended up with.

http://www.nbs-stl.com/CRX/Dyno%20Pull%20AFR.png

http://www.nbs-stl.com/CRX/Dyno%20Pull%20Graph.png

and the dyno sheet

http://www.nbs-stl.com/CRX/CRX%20Dyno%20Graph.jpg

Notice the fairly distinct peak in torque at 5400 RPM. The power rolled off fairly fast at about 7500 RPM so I set the soft rev limit right about there.

The guy running the dyno knew from doing lots of this sort of thing that this motor would show about 142 HP on a more typical roller dyno. It was now time to move on the real world of racing and data logging.

This is the VE table I wound up with as I headed off to racing.

http://www.nbs-stl.com/CRX/VE%20Table%203d.PNG

Notice that there is a distinct point in the table where the amount of fuel the motor is wanting jumps. Also notice the bump in fueling that the motor needed at 5400 RPM and WOT.

On the left, is the Speed Density VE map we ended up with at the dyno.

http://www.nbs-stl.com/CRX/MAPxRPM%20IT ... Es%202.png

Moving on to the track, a new pattern started developing. At this point, I need to get into a tune verification method I found. On a Speed density system, the vast majority of your air flow is a function of RPM and MAP. Double the RPM and you generally get twice as much air processed. The same goes for MAP. If you double the pressure, you get twice the air thru the motor. This rule works on any motor that the intake and exhaust don’t have major ram tuning going on. Here are a couple of examples. The first is from my Harley Sportster. The second trace is off my personal 1500 cc, supercharged 300 HP jetski.

http://www.nbs-stl.com/HarleyTuning/MAP ... ty%202.png

http://www.nbs-stl.com/motec/MAPxRPM%20 ... 0Motec.png

Notice how on both of these motors, If you plot the optional calculated field MAPxRPM vs Duty Cycle, you get essentially a straight line. Try it on any typical (non ITB) fully tuned motor.

ITB tuning is a different animal. When tuning a Yamaha FZ1 motor a few years ago, I discovered that below about 15% throttle, that motor followed this rule just fine. But above this 15% throttle position, all hell broke loose and the HP Gods showed up. At the time I was not sure what I was looking at but I knew I had to go to throttle based tuning to get anything to make sense. When I got to tuning the CRX, This stuff all started to become more logical. What changed in that several year span was the abilities of MegaLogViewer and the HD version of MLV. I could now do filtering of data and find the patterns of what was going on.

Here is a screen shot of MAPxRPM vs Duty Cycle on the CRX.

http://www.nbs-stl.com/CRX/ITB%20MAPxRP ... Levels.png

This thing was breaking all of my rules. What happened to a straight line. Why does this motor want so much fuel up top? You can see this starting to develop on the early WOT screen shots from the dyno.

http://www.nbs-stl.com/CRX/Dyno%20Pull%20AFR.png

Look at the bump on the right graph, right at 5500 MAPxRPM. If I start looking for patterns in the MAPxRPM vs Duty Cycle scatter plots, it becomes very apparent that this motor really wants more fuel as a function of both RPM and Throttle Position. Way more than the MAP and RPM should imply.

http://www.nbs-stl.com/CRX/ITB%20MAPxRP ... 0Power.png

Here is what it looks like with filtering applied to separate the low power (Speed Density) stuff from the high power (Alpha N) stuff on the right plot.

http://www.nbs-stl.com/CRX/ITB%20MAPxMP ... 0Power.png

How do you actually tune a motor that behaves like this?

I then created filters to isolate each RPM into 300 rpm slices. Here is what each filter looks like in MLVHD.

As an example, for 6000 RPM, I used

[RPM] < 5850 or [RPM] >6150

Here is what developed in the MAPxRPM vs Duty Cycle plot.

http://www.nbs-stl.com/CRX/ITB%20MAPxRP ... 20RPMs.png

From left to right, these are 4500, 5500, 6500 and 7500 RPM. Notice how there is a fairly abrupt change in the fueling requirements right at 22 degrees TPS. As it happens, this transition also happened right at 85 KPA. This is the very essence of what lead the Megasquirt programmers to ITB tuning. They found the same patterns thru different methods. But motors are motors. They have patterns, MLVHD just makes it possible to visualize.

In looking at the racing data coming off my ITB Motor, It was apparent that the VE was changing very quickly at the upper MAP values. It was going to be almost impossible to have enough granularity to build a Speed Density MAP. But when I looked at the VE table in Alpha N presentation, the VE table was gradual. That is when I decided to go to having a Speed density table to control the basic fueling and have a TPS based second table that was a multiplier to the Speed Density table. As the tune developed with time, this is what the two table started looking like.

http://www.nbs-stl.com/CRX/MAPxRPM%20IT ... Es%202.png

Or as a surface map it looks like this.

http://www.nbs-stl.com/CRX/MAPxRPM%20IT ... %20VEs.png

At this point the motor is running amazing. I will post data when I get the final tune nailed down. Hereis the MSQ that I am running.

http://www.nbs-stl.com/CRX/CurrentTune.msq

Feel free to ask any questions and have fun tuning.

Andy

[whittlebeast]

Here is a view I bet very few people here realized was possible in MLVHD. This is what the Alpha-N AFR results are but you could also use this same feature to generate an Alpha N VE table. I then changes the data trace to 25000 points to see where the motor goes thru in a typical autocross. I find this feature very useful.

http://www.nbs-stl.com/CRX/CRX%20AFR%20 ... 20Path.png

Andy

[muythaibxr]

Andy, you basically just got to the point with 2 tables that we can now tune with 1 table in ITB mode.

I came up with this mode to solve this exact problem in 2009 IIRC.

[dontz125]

1 table in ITB mode.

Well - one table and two curves. Curious - what's the workload / effort like, tuning the switchpoint curves?

[whittlebeast]

I had very little effort at getting my head around my method of tuning. It was easy to understand. The way I use my motor makes most auto tuning almost impossible. Everything in the log is tranditions.

[whittlebeast]

This is what 3 min worth of racing data looks like in scatter plot view. Blue areas are rich. I make the adjustment for the next tune on the side that has a more clear pattern and square area. I adjust about 4% at a time to sneak up on things.

http://www.nbs-stl.com/CRX/CRX%20AFR%20SD%20vs%20AN.png

[jsmcortina]

I've used and dyno tuned a 4-cyl engine with ITBs using the ITB mode and it works fine.

I've told Andy to try this numerous times, but he doesn't want to.

James

[whittlebeast]

This is the other view I use to address the red or blue areas first. Note that you need to upgrade the the HD version of MLV to get access to some of these views.

http://www.nbs-stl.com/CRX/CRX%20AFR%20 ... togram.png

Andy

[whittlebeast]

Here is a video to get an idea of a typical run.

https://youtu.be/3iNdHm81lgs?t=95

This is very different from what you get on a dyno.

[muythaibxr]

1 table in ITB mode.

Well - one table and two curves. Curious - what's the workload / effort like, tuning the switchpoint curves?

Easy. Drive it a bit while datalogging, use MLV scatter plot to map the throttle position at 90 %baro, set the switchpoint curve to what you find, and leave the load at switchpoint curve alone with its defaults unless you have a really good reason not to.

[muythaibxr]

Andy, dyno or not, the way you are tuning with 2 tables, and that being difficult and time consuming to tune are the reason I designed ITB mode. It is even easier to get your head around... Set the switchpoint curve per the log, then tune like any other engine.

[KenK]

07/05/16

Very interesting.

The engine I have running and setup I am currently using is a straight SD tune method. I now have plenty of data to find the 90% switch point. A question I have that is not clear to me is "What does the 90% number refer to".

90% of the Baro at the current time which could move around a lot depending on elevation or 90% of standard ATM which would be about 90kPa.

If we are talking about 90kPa, the only way to find this would be or to run the engine at sea level.

I tend to run the engine at varying elevations and in fact the elevation could change 300 to 1200 meters in a single run.
Ken

[whittlebeast]

KenK, can you post a tune and a datalog. I would love to see where you are at on the tune. Keep in mind that most ITB motors can be tuned with the ITB mode that the software supports and is far more friendly to autotuning than what I did

[muythaibxr]

07/05/16

Very interesting.

The engine I have running and setup I am currently using is a straight SD tune method. I now have plenty of data to find the 90% switch point. A question I have that is not clear to me is "What does the 90% number refer to".

90% of the Baro at the current time which could move around a lot depending on elevation or 90% of standard ATM which would be about 90kPa.

If we are talking about 90kPa, the only way to find this would be or to run the engine at sea level.

90% is 90% of baro. If you're at sea-level this would be 90 kPa, If you're at elevation, it'd be lower, but the table lookup will always be hit at 90% of barometric pressure (unless you set it differently.)

I tend to run the engine at varying elevations and in fact the elevation could change 300 to 1200 meters in a single run.
Ken

As long as barometric correction is tuned this should not pose a problem I don't think.

Ken

[dontz125]

Given the dependance on % baro, are ITB tunes more sensitive to elevation changes than pure SD?

[whittlebeast]

Given the dependence on % baro, are ITB tunes more sensitive to elevation changes than pure SD?

Great question. I wish I knew the answers.

After looking at my data, I am leaning toward trying an Alpha-N base tune and then have a Speed Density correction table. I can post up what the my first guess Alpha-N would look like. I would simply let MLVHD generate it for me with the histogram table generator.

Andy

[suberimakuri]

Given the dependance on % baro, are ITB tunes more sensitive to elevation changes than pure SD?

Yes, because ITB motors VE is a logarithmic curve upwards with kpa on SD load method. Single throtles are pretty linear with SD load method.
As altitude rises, max load kpa readings reduce and thus you will not reach high load regions of the SD VE table and will miss the extra fueling of those regions. Even though the load on the motor warrants this high VE.

Dontz, for example:
Our example ITB motor VE at 90kpa might be 40, but at 95kpa it might be 70.
For the same motor on single throttle, VE at 90kpa might be 60 and at 95kpa it might be 70.
Drastic differences in fueling required when using SD only as a basis of load.

If you use %baro and not SD you will reach the high load VE regions, which is great.
However, you will still have the issue with logarithmic curve and the row spacing will be very small above 90kpa (~2kpa each) and any pulsing in the map signal will pull fuel from the 'wrong row'.

If you use AN you have a linear curve much like regular single throtle SD, so it's easy to tune, easy row spacing, but you miss out on SD advantages at low load.

ITB mode with baro = best of all.
Andy, please just try ITB mode, you have all the data. As Ken said, we have been down the other paths.

(I have tuned lots of 1600-2000cc 4 cylinders with ITB's in SD and AN, varying altitudes, now adding turbos.)

[whittlebeast]

Is there a way to derive an ITB map with the data I have in MLV? This car is not the slightest bit streetable and I would not be able to get on the dyno till after race season. I only have about 4 events left this season that I will be able to get to.

What is the formula for ITBLoad?

I know the MAP and TPS where my motor transitions. That part is easy to find.

Andy

[whittlebeast]

I just posted the data from 5 autocross runs. Just change the name back to a zip file to open it up.

http://www.nbs-stl.com/CRX/2016-06-26_1 ... tt.zippppp

Here is the tune that I tweaked after those runs.

http://www.nbs-stl.com/CRX/2016-06-28%2 ... 205000.msq

Here is what I wanted to try next. Pure SD below 3450 RPM and Pure AN for racing above 3450 RPM

http://www.nbs-stl.com/CRX/CRX%20SD%20t ... h%2000.msq

[whittlebeast]

Just for fun, here is what my motor would want if it was tuned Pure SD

http://www.nbs-stl.com/CRX/CRX%20if%20Pure%20SD.png

And if tuned Pure AN

http://www.nbs-stl.com/CRX/CRX%20if%20Pure%20AN.png