Lizard928

Greg,

When looking at the map there are massive spikes in the tune, up and down.
when datalogging, especially in a lower gear, they will not show up as dips/spikes.
But there will be split second changes in the AFR going from a good AFR to a lean AFR. You can see in the WOT area of the map it has jumps from a cell value of 25 to 97.
The maps really must be smoothed out (preferably in a 3D mapping software) to achieve the most consistant results, and the safest values for the engine.
Additionally, the zero'd area, should anything change (altitude is enough), it will be hitting this different part of the map, and BAM tune is out, and out enough to blow up the engine.



Attached are two files showing how the tune will wind up without smoothing, and how it "should" look to achieve a smooth consistant AFR, which will auto compensate for altitude and other changes.

victor25

Colin
Your right about the 25 to 97 jump, and that should be smoothed out, and was missed. Many of the numbers were set by a couple long runs and sharkplotted. Yes this map needs some smoothing and an area or two of missed numbers like that 25. But for you to say it's dangerous and could blow up the engine is a little much hey! He has the O2 sensor enabled and also adaptation. And did you not read where he says the car is running quite good... here I will quote it

""After using the Sharktuner and Sharkplotter software I have my AFR pretty nice and close to 14.7 all the time at idle and appropriately enriching as you put the coals to it.""

Jeff is a smart guy and knows what it should be like

Lizard928

Victor,

All it takes is 1 really bad knock to take out a head gasket.
And all it takes is a series of small knocks to do the same.

When you have peaks and valleys like are shown, they are dangerous to be running around on.

Adaption is turned off once you get too high throttle input. As well, my point stands, he takes the car for a drive, altitude changes, he then hits the area of the map with zero's, even with 6-9 deg of ignition being removed by the EZK, it will NOT save the engine. So yes, I stand by that it is dangerous.
You have to have the map smoothed, as I have shown or the moment you change your altitude, all the tuning is out the window as it will be running in a different area of the map!

jeff spahn

I recorded a video on the way to dinner. I'll upload eye I get home on my Mac. I'll data logo the way home too. I'll do it with O2 sensor off.
Then I'll shark plot and post

I am still a noob to tuning so I'll take any help.

Also note in video i post later I have it set to 30 # injectors to smooth surge. Any help there would be good.

Rob Edwards

Just for grins, here's Jeff's LH map plotted in Excel:





And the factory 90GT LH map for comparison. Obviously an apples to oranges comparison between two engines differing in injectors, forced induction, etc. Every LH map I've ever seen tails up at high MAF/hi RPM values, they don't drop off to zero. Jeff's map just seems backwards of what I'm used to seeing- Is this a typical centrifugal S/C fuel map?

victor25

Jeff

Your injectors Should be set at 30# as that is what Jim and I tuned it to.

Attached is that same map with circled areas that could use some attention. Blue rich go down a little, red lean go up a little to smooth out the map. This will literately make no difference if the drive-ability of your car, as all of these adjustments are outside your normal driving range in the map. Hence the reason the weren't adjusted by the plotter or us.

Colin, please correct me if I am wrong... But if your map is all zeros, that means you are making no adjustments to the cars internal baseline map, YES? When you have a positive or negative number in a cell that means your adding or subtracting a SMALL amount of fuel to the cars baseline map for that cell. With 127 being a maximum of +/- 25%. So a zero means that we are accepting the base map as being good. This is a good thing, not a bad one!! That also means that an area of zeros in the map is as SMOOTH as it can possible get in that area. If I am wrong I hope that Jim or John will speak up and set me straight . This car has a supercharger and no air straightener at the moment. The numbers are what they are for a reason. Yes they need to be polished, but are where they are for correct fueling. I will be the first to stand up and say I am wrong if I am... Hopefully I am not

That GT map isn't much smoother

victor25

Here is the map on a GT with the same stage 1 kit as Jeffs. Same injectors, pulley size, same everything except black hoses instead of red.
this car runs 11.5 at wot, when maf referance is 500 at 6000 rpms

GregBBRD

Interesting. And confusing.

Most all of the correction maps that I've ever seen are full of peak and valleys.....even the very high tech maps from Porsche Racing. I've highlighted correction, because, as I understand it, we are looking at is actually a very small percentage of the actual total map. And like looking at any graph that is blown up, little changes seem very large (advertising people love to show blown up graphs, so that little changes look huge.) I'm guessing that if combined with the total map, the corrections will not seem nearly as radical.

Universally, the professional tuners that I've talked to, over the years, refer to the super smooth correction maps as "throwing a blanket" over the engine, and almost universally consider the engine to be poorly tuned.

My understanding is that the requirements of an internal combustion engine are quite different at minor changes in load and rpm.....which is why correction maps need to have all those peaks and valleys.

Once again, I believe that when these corrections are combined into the "total" map, the radical looking peaks and valleys are actually going to hardly be noticed.

jcorenman

There is some good info here, let me just add a couple of comments.

Certainly smooth AFR's are helpful in terms of driveability. But a smooth map does not necessarily translate to smooth AFR's-- intakes and exhaust systems have resonances, and the engine needs more or less fuel at certain RPM and load points in order to maintain a target AFR.

Certainly you can get peaks and valleys with Sharkplotter or any other method of tuning, particularly in areas of the map where there isn't much data. But as long as those maps values are based on valid data of some sort, it is hard to call them "wrong".

In fact a map which has been arbitrarily smoothed can be worse than one which is lumpy-- because the smoothing process ignores what the engine actually wants, fuel-wise. If the "peaks" are errors then cool, smooth them out. But if the engine needs more fuel at that point then don't take it away.

In working with smoothing algorithms for Sharkplotter, I've generated lots of nice-looking maps which produced worse AFR's than before I started. It is something that needs to be done cautiously, followed by more testing.

One thing that is not showing on Jeff's maps is the WOT map, which adds fuel when when the WOT switch is actuated. The WOT map is useful if more fuel is needed at high loads, beyond the main fuel map. But it makes tuning more difficult, because two different map cells are contributing to the fuel calculation. Whenever possible I think it is a good idea to set the WOT map to zero and just tune the main map.

The GT (and S4) maps have big numbers in the last two rows (the "bleachers" in Rob's plot), where you can't get to with anything resembling a stock NA motor. I think the factory figured that if you ever got there, pour in lots of fuel. Those cells are clearly wrong (but "safe"), but is a cell which can't be reached really "wrong"?? (If a tree falls in the forest, ... oh never mind).

Not true. O2-adjust gets turned off, but "Adapation" is the long-term fuel correction which is active all the time. Adaptation is the accumulation of the fuel adjustments that the LH makes in response to the O2 sensor-- the short-term "O2-adjust" correction.

For example, suppose the MAF is aging and reading low, then the AFR will always be a bit lean. If O2-adjust is always adding 10% more fuel, on average, then this will be accumulated as a +10% adaptation value, which will add 10% more fuel all the time-- whether idle or WOT, cold or warm.

Sharktuner allows Adaptation to be disabled for tuning, which is important-- otherwise you have one more variable adding fuel to the equation. Turn adaptation off for tuning, and back on when finished so that the LH can compensate for long-term changes.

But the point is correct-- O2-adjust will indeed try to compensate for a "lumpy map", and correct any cells that aren't right. But it can only do that at light load, and even that is a losing game because adaptation is storing all that bad data.

jcorenman

Hey! Don't drag me into this, I only looked at a couple of maps and then you guys went off and did something...

Cells which aren't reached during normal (or even abnormal) driving are hard to tune... sometimes you just have to guess.

This is all correct.

Well, that all depends on what the engine really wants. The baseline map is a theoretical baseline which ignores all of the reality stuff. The goal is smooth AFR's, not a smooth map.

Red hoses are faster. . The problem I have with maps that are artificially smooth is that they are the result of arbitrary decisions, not data. But OK, in spite of all that, the AFR plot you posted looks pretty reasonable. It needs more data, there is a lot of throttle between cruising along and WOT..

This is correct. The map is +/- 25% fueling, so the difference between a map value of 25 and 97 is around 12% more fuel.

If it is based on what the engine wants, then that is right. But it is easy to get fooled, and translate a couple of anomalous data points into a bogus map value. But you are correct, that even that would hardly be noticed, unless it lands where you spend a lot of time driving. The good news is that the map cells where you spend a lot of time, are easy to tune.

I wonder if we can talk Jeff into coming out for Sharktoberfest?

victor25

Here is a plot of raw data from Jeffs car tonight.
looks pretty good to me in the meat of the driving and power zones.
It just needs a little more fuel in the circled area. Low rpms, middle maf. But that also depends on your target AF

jeff spahn

I'd love to come out but no vacation time left this year. Going to Disney Oct 20-28th or so and that kills my 4 weeks this year.

Oh, here's video about 15 min. 30# settings. no O2 adaptation turned on if I have it right.

GregBBRD

Jim:

As you know, from my stupid questions, I'm not an "expert" tuner...nor do I pretend to be one.

Help me understand this process, if you can.

My understanding is that the fuel injection system is actually taking into consideration information from the cells that "surround" any particular cell (up and down) and coming up with an "average" of those cells....as the engine transitions through individual cells.

Is this true?

If it is, this would point to a correction map with peaks and valleys being "more correct". If the engine needs to be leaner or richer to achieve proper mixture, at any one particular point, it might require a radical change to one cell, to compensate for the previous cell.

Let me explain what I'm thinking....in a changing rpm cell direction only....no change in load (single plane.) And I'm not only not having any changes in load, but there are only zeros above and below these "corrected" cells (in the rpm plane).

So here's what is in my theoretical cells:

2000 rpm=0
2500 rpm=20
3000 rpm=-20
3500 rpm=20
4000 rpm=0

And let's assume that those are correct fuel requirements, not glitches.

In my example (and my understanding) the actual fuel delivered will be (approximately).

2000=0
2250=10
2500=20
2750=0
3000=-20
3250=0
3500=20
3750=10
4000=0

If the original "map" was smoothed, everything would end up being "O", right?

So the engine would be lean at 2500, rich at 3,000, and then lean again, at 3,500....even though it got tuned.

The "smoothing" destroyed the tuning. right?

Here's what I do....right or wrong.....and I'd love it if you would be critical about what I do.....on my "big" engines.

I "smooth" once....usually after the first drive. There are generally negative and positive numbers all over the entire range. Most times I'll have numbers all the way from -127 to +127. Smoothing will "remove" all of these -127 and +127 numbers, generally.

From then on...I do multiple test drives.....I "allow" the correction numbers to be whatever the system determines that number to be....no more smoothing.

If it needs to be leaner somewhere, it can do that.

If it needs to be richer at one particular point, it can do that.

But I don't smooth it again, which I believe "destroys" the fine tuning, between cells.

jeff spahn

My brain is full tonight. Battery in tuning laptop is nearly dead. Going to sleep.

jcorenman

Yes, that is correct-- the LH interpolates (averages) map values from adjacent cells, depending on the actual load and RPM.

Correct, but that "radical" value stands alone when the RPM & load are exactly centered on that cell... This is why the RPM scales for the LH are uneven, areas where larger changes are expected have smaller cells.

This is correct, no "approximate" needed. If the RPM's are at 2250 then the LH averages the values for 2000 and 2500 (for this hypothetical map) and comes up with 10.

That depends on how strong the smoothing was, but yes-- any rational smoothing algorithm would look at the -20 surrounded by +20's and say "Hmmm, that ain't right... surely he meant for all those cells to be closer together. Depending on the amount of smoothing, it might be 0,0,0,0,0 or 0,+5,-3,+5,0 or something like that. But if the +20 and -20 is what the engine wanted, then correct-- that's gone.

Well I'm no expert either, I get the computer stuff but the whole combustion/rotating machinery thing is a big mystery. But that's pretty much what I've been doing-- do the smoothing early, then more testing and tuning to fix what I messed up.

The dilemma is always the areas where there is little data-- 3/4 throttle at high RPM's, low RPM and higher load, where you have to make an effort to get data.

But yes, I think that makes a lot of sense.