Sharktuning Fine Tuning
09-27-2013, 12:49 AM
#33
Race Car
Every time I look at those zeros at the bottom of the posted fuel map, I cringe. Stock maps do not have this, and you do not have access to modify the base map. I can just see someone upgrading to a smaller upper S/C pulley and blowing the engine up almost instantly the first time they hit WOT. It doesn't matter if in its current configuration the car can never hit those lower cells, you should still make them safe so that if something ever changes and those cells suddenly become accessible, the head gaskets won't explode. I'm sure the factory tuners were pretty good at their job and they did exactly that with the stock maps. If you want to pick a number to delineate where the engine can't go, I'd rather use '+127' instead of '0'.
Otherwise, a very good discussion. I also tune like Greg. I will start off with some base values, then smooth everything out once, then fine-tune. I also believe that smoothing excessively will lead to a less optimal map.
Dan
'91 928GT S/C
475hp/460lb.ft
Otherwise, a very good discussion. I also tune like Greg. I will start off with some base values, then smooth everything out once, then fine-tune. I also believe that smoothing excessively will lead to a less optimal map.
Dan
'91 928GT S/C
475hp/460lb.ft
09-27-2013, 02:09 AM
#34
Rennlist Member
Join Date: Oct 2009
Location: Michigan... Grand Rapids
Posts: 758
Likes: 0
Received 2 Likes
on
1 Post
Dan, you like some of the others are missing the point. If we put positive numbers in there, the air fuel will be in the 9's or low 10's. We may end up putting negitive numbers in there to get to the 11 to 12 range by the time we are done. The factory puts high numbers in as a safety net assuming they won't be reached. The GTS only hits about 260 maf reference with 345hp. Look at the gt map and plot I displayed. That car is like 11.5 with the cell numbers at 20, and he can get to 500 maf in the right conditions
09-27-2013, 02:18 AM
#35
Rennlist Member
Do you have a boost gauge? Anything going on there during those periods? Light throttle, low RPM-- I wouldn't expect there is much pressure if any.
You also indicated that it wasn't real smooth under load-- anything similar or related there?
09-27-2013, 02:20 AM
#36
Rennlist Member
Dan, you like some of the others are missing the point. If we put positive numbers in there, the air fuel will be in the 9's or low 10's. We may end up putting negitive numbers in there to get to the 11 to 12 range by the time we are done. The factory puts high numbers in as a safety net assuming they won't be reached. The GTS only hits about 260 maf reference with 345hp. Look at the gt map and plot I displayed. That car is like 11.5 with the cell numbers at 20, and he can get to 500 maf in the right conditions
09-27-2013, 02:39 AM
#37
Rennlist Member
Join Date: Oct 2009
Location: Michigan... Grand Rapids
Posts: 758
Likes: 0
Received 2 Likes
on
1 Post
The WOT I believe is zero'd out
If you look at his plot he was in the 324 maf range a few times and was at 12 af (yellow). You should also notice that the 324 cells for that rpm range are zeros!! Actually if you look closely, the shark plotter is recommending that he go less than zero for 324 by 3300 rpm cell
If you look at his plot he was in the 324 maf range a few times and was at 12 af (yellow). You should also notice that the 324 cells for that rpm range are zeros!! Actually if you look closely, the shark plotter is recommending that he go less than zero for 324 by 3300 rpm cell
09-27-2013, 03:06 AM
#38
Drifting
I assume that there are a few SC out there and I haven't heard of anyone braking any rings yet.
Even the frankenstein that Greg had delivered to him ran, though bearly by the sounds of it.
When I see big numbers at the top I increase my top fuel allowed on my X/Y chart
I run the car and let it fill the map and then I save it.
Reprogram as the new saved file and then when I re open it, it starts with all zeros and I fine tune again
but I suppose the electromotive system is different and Colin may correct me.
I would guess that the hunting would go away if it was richer, at least I would try that first.
If it does not help then I would start leaning it out and monitor for knock.
Either way I started from scratch,
made sure I started rich with min timing, decreasing fuel and increasing timing and I guess for 2-3 years I have been Lucky. But I'm not blown, I only wish
Even the frankenstein that Greg had delivered to him ran, though bearly by the sounds of it.
When I see big numbers at the top I increase my top fuel allowed on my X/Y chart
I run the car and let it fill the map and then I save it.
Reprogram as the new saved file and then when I re open it, it starts with all zeros and I fine tune again
but I suppose the electromotive system is different and Colin may correct me.
I would guess that the hunting would go away if it was richer, at least I would try that first.
If it does not help then I would start leaning it out and monitor for knock.
Either way I started from scratch,
made sure I started rich with min timing, decreasing fuel and increasing timing and I guess for 2-3 years I have been Lucky. But I'm not blown, I only wish
09-27-2013, 04:55 AM
#39
Rennlist Member
Yet another interesting discussion on this subject matter. I have no idea about the mathematical algorithm being used for smoothing but I can tell it is embedded and doing something- be it good or bad. Thus I find the "best idle" is achieved by looking for 14:1 and setting the cells around the operating point to the same correction value. Nio idea whether this is good or bad practice but it seems to work for me.
Greg alluded to a point I have been trying to come to terms with since I started trying to play around with ST2 and that is that there is a base tune which as I understand varies fuel based on rpm and load. The rpm input logically has to be directly proportional- more strokes needs more fuel type of thing but for load that is more of a puzzle to me but clearly the deeper you bury the throttle the more work you are trying to get out of the motor. Thus if you want a meaningful graphic you need to plot it based on the premise that the operating point is 100 and the corrections are quantities adding or subtracting up to 25% of the base value. Thus to get a true picture of what your mapping is doing you really need to plot a measure of the absolute amount of fuel being added to understand what your map looks like. If you do this I suspect the graphic will look way more smoother than plotting correction values which I suspect does not mean much.
At the end of the day the map is simply an approach to breaking down the tuning into a number of manageable steps. I have never understood why the array has so many cells that in effect are never visited and relatively few in the areas we are generally "most interested in" [see Ken's approach].
Clearly, where there is a cell boundary, there is change and step function change cannot be correct by definition so the model perhaps tries to make a [progressive?] change to assimilate the correction values as the boundary is crossed thus smoothing the transition boundary. Less difference less to be smoothed but frankly, if as I do, I look at the distribution of values recorded in a given cell to help determine the correction to be applied, then "articificial smoothing" does not seem to make too much sense.
Clearly one is trying to avoid too much cliff hanging here but I am intrigued as to what consititues sensible tuning and what is in effect trying to "paint a pretty picture".
Thus if you break the map down into what really matters- on an auto [and in effect with a manual for that matter] what happens below 2k rpm is largely irrelevant other than idle. The cruise map is controlled by the cells bounded by 2k rpm to about 4k rpm [for the most part] and load values in the range of 100 to 200 and the go **** is controlled by what happens in the range 200 to 350 for N/A engines.
Thus the fuelling cruise map is controlled by an array of 42 cells and full load fuelling by a snake type array of cells three deep over 9 rpm ranges [27 cells]. The rest appear mostly irrelevant. Thus about 70 cells are relevant in an array of 256 cells.
When you analyse AFR points in given cells we typically see a variance of plus or minus about 0.3 AFR. Presumably with boundary condition smoothing and resonant pulses floating around this is the best level of repeatability that can be expected.
Even the stock maps seem to jump around and good as it may be, I often wonder whether dyno tuning repeats on the open road when you have things like air blast to factor in.
It is not difficult to understand why tuning is quite an art- just wish I was a bit better at it. Above all else I try to tune out knocks or at least keep them to a minimum even if it means pulling back the timing a bit as I had to at full load/high rpm's.
Whether or not tuning a F/I motor is more difficult would be an interesting challenge but it seems as though pulling timing is inevitable unless you have access to rocket fuel.
just a few thoughts- not sure they add value to this thread but it does indicate what we are up against and no shame if one struggles a bit with it.
Even if folks like Victor come up with a decent broad brush tune, it should be seenas a starting point- a building block if you like but then you need a certain amount of skill to be able to add further value. Stock maps by definition have to be conservative to be safe for all examples and this logic has to apply to what folks like Victor offer. At the end of the day he cannot control what gasoline you are using or what temperature it is used at in your car.
Regards
Fred
Greg alluded to a point I have been trying to come to terms with since I started trying to play around with ST2 and that is that there is a base tune which as I understand varies fuel based on rpm and load. The rpm input logically has to be directly proportional- more strokes needs more fuel type of thing but for load that is more of a puzzle to me but clearly the deeper you bury the throttle the more work you are trying to get out of the motor. Thus if you want a meaningful graphic you need to plot it based on the premise that the operating point is 100 and the corrections are quantities adding or subtracting up to 25% of the base value. Thus to get a true picture of what your mapping is doing you really need to plot a measure of the absolute amount of fuel being added to understand what your map looks like. If you do this I suspect the graphic will look way more smoother than plotting correction values which I suspect does not mean much.
At the end of the day the map is simply an approach to breaking down the tuning into a number of manageable steps. I have never understood why the array has so many cells that in effect are never visited and relatively few in the areas we are generally "most interested in" [see Ken's approach].
Clearly, where there is a cell boundary, there is change and step function change cannot be correct by definition so the model perhaps tries to make a [progressive?] change to assimilate the correction values as the boundary is crossed thus smoothing the transition boundary. Less difference less to be smoothed but frankly, if as I do, I look at the distribution of values recorded in a given cell to help determine the correction to be applied, then "articificial smoothing" does not seem to make too much sense.
Clearly one is trying to avoid too much cliff hanging here but I am intrigued as to what consititues sensible tuning and what is in effect trying to "paint a pretty picture".
Thus if you break the map down into what really matters- on an auto [and in effect with a manual for that matter] what happens below 2k rpm is largely irrelevant other than idle. The cruise map is controlled by the cells bounded by 2k rpm to about 4k rpm [for the most part] and load values in the range of 100 to 200 and the go **** is controlled by what happens in the range 200 to 350 for N/A engines.
Thus the fuelling cruise map is controlled by an array of 42 cells and full load fuelling by a snake type array of cells three deep over 9 rpm ranges [27 cells]. The rest appear mostly irrelevant. Thus about 70 cells are relevant in an array of 256 cells.
When you analyse AFR points in given cells we typically see a variance of plus or minus about 0.3 AFR. Presumably with boundary condition smoothing and resonant pulses floating around this is the best level of repeatability that can be expected.
Even the stock maps seem to jump around and good as it may be, I often wonder whether dyno tuning repeats on the open road when you have things like air blast to factor in.
It is not difficult to understand why tuning is quite an art- just wish I was a bit better at it. Above all else I try to tune out knocks or at least keep them to a minimum even if it means pulling back the timing a bit as I had to at full load/high rpm's.
Whether or not tuning a F/I motor is more difficult would be an interesting challenge but it seems as though pulling timing is inevitable unless you have access to rocket fuel.
just a few thoughts- not sure they add value to this thread but it does indicate what we are up against and no shame if one struggles a bit with it.
Even if folks like Victor come up with a decent broad brush tune, it should be seenas a starting point- a building block if you like but then you need a certain amount of skill to be able to add further value. Stock maps by definition have to be conservative to be safe for all examples and this logic has to apply to what folks like Victor offer. At the end of the day he cannot control what gasoline you are using or what temperature it is used at in your car.
Regards
Fred
09-27-2013, 07:36 AM
#40
Supercharged
Rennlist Member
Rennlist Member
Join Date: May 2002
Location: Back in Michigan - Full time!
Posts: 18,925
Likes: 0
Received 60 Likes
on
34 Posts
I've got two supercharged cars, both twin screw, and I've never had any problem like the one discribed. I've had lots of other problems... Just not this one. Hmmmmm....
09-27-2013, 08:05 AM
#41
Rennlist Member
Thread Starter
Sorry, Disneyworld, not land. Went to that coast last year. Switch coasts every year.
09-27-2013, 08:10 AM
#42
Rennlist Member
Thread Starter
Something else to bring up. Before we did the Supercharger it had a smoothish idle. Now it hunts a bit. Not smooth like my Acura, my car has never been that smooth. Had idled a little "lumpy" since I got it. Greg never said anything about that while it was at his shop during PPI so I took it as, that's how it idles.
Now it idles a little more lumpy. Not head gasket. Idle didn't change when I replaced those. All gaskets and sensors are new when I pulled engine two years ago. It's not rough idle, just not silky smooth but worse that stock was. I think it has something to do with our current tune myself because it showed up right after we did the install. Almost a stumble lumpy every now and again. Hard to describe. Look at the video I posted and see the values move around.
Sean will see what I mean when he is up here.
Don't get me wrong here. We are talking stuff that is princess and the pea level.
Now it idles a little more lumpy. Not head gasket. Idle didn't change when I replaced those. All gaskets and sensors are new when I pulled engine two years ago. It's not rough idle, just not silky smooth but worse that stock was. I think it has something to do with our current tune myself because it showed up right after we did the install. Almost a stumble lumpy every now and again. Hard to describe. Look at the video I posted and see the values move around.
Sean will see what I mean when he is up here.
Don't get me wrong here. We are talking stuff that is princess and the pea level.
Last edited by jeff spahn; 09-27-2013 at 09:44 AM.
09-27-2013, 09:55 AM
#44
Rennlist Member
Thread Starter
Here is the drive to work this morning data logging.
In the "Sharktune Data Files" use date and parameters as the name.
so 9271330lbnoo2.txt means:
date of 9/27/13
30# tune
no O2 enabled.
https://www.cubby.com/pl/My+Cubby/_d...28639db9deda3a
In the "Sharktune Data Files" use date and parameters as the name.
so 9271330lbnoo2.txt means:
date of 9/27/13
30# tune
no O2 enabled.
https://www.cubby.com/pl/My+Cubby/_d...28639db9deda3a
09-27-2013, 10:55 AM
#45
Rennlist Member
...
Clearly, where there is a cell boundary, there is change and step function change cannot be correct by definition so the model perhaps tries to make a [progressive?] change to assimilate the correction values as the boundary is crossed thus smoothing the transition boundary....
Clearly, where there is a cell boundary, there is change and step function change cannot be correct by definition so the model perhaps tries to make a [progressive?] change to assimilate the correction values as the boundary is crossed thus smoothing the transition boundary....
Thus if you break the map down into what really matters- on an auto [and in effect with a manual for that matter] what happens below 2k rpm is largely irrelevant other than idle. The cruise map is controlled by the cells bounded by 2k rpm to about 4k rpm [for the most part] and load values in the range of 100 to 200 and the go **** is controlled by what happens in the range 200 to 350 for N/A engines.
Thus the fuelling cruise map is controlled by an array of 42 cells and full load fuelling by a snake type array of cells three deep over 9 rpm ranges [27 cells]. The rest appear mostly irrelevant. Thus about 70 cells are relevant in an array of 256 cells.
Thus the fuelling cruise map is controlled by an array of 42 cells and full load fuelling by a snake type array of cells three deep over 9 rpm ranges [27 cells]. The rest appear mostly irrelevant. Thus about 70 cells are relevant in an array of 256 cells.
The "fine-tuning" issue that Jeff raised here is not one of proper fueling-- that seems OK based on the plots that he posted, whatever the maps look like. He is chasing relatively small driveability issues, a surging or hunting in the light-throttle / lower-RPM "cruise", areas and at idle. From his video, I don't think it is an issue with the fuel map-- something else is going on.
Cheers, Jim