I am having difficulties getting a smooth VE table while tuning my turbo
01-06-2013, 10:02 PM
#1
I am having difficulties getting a smooth VE table while tuning my turbo
Well, i have been fighting my damn tune for the last few days. Previously i was able to get a pretty damn smooth VE table when i was NA. Now it seems very rocky and the values are not really making sense... I know this can not be normal...
Thoughts?
Thoughts?
01-06-2013, 11:36 PM
#5
TECH Apprentice
Join Date: Dec 2010
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
I usually don't worry about the spikes until I get the AFR where it needs to be and then I go back and smooth things out. Are you tuning Speed Density or still tuning with MAF? If you can post up your VE and info and I can help out if you like.
01-06-2013, 11:48 PM
#7
TECH Apprentice
Join Date: Dec 2010
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
That makes sense. What size injectors are you running and what brand?
Trending Topics
01-07-2013, 12:20 AM
#9
Really the only important part of the injector data in non-idle areas is the flow rate table. Your regulator and pumps should keep the pressure constant (assuming you have a return system), so that table should be flat. From that, any errors in fueling is directly resulting from the ve table being off.
Go on the interstate at some non-busy hour and put it in 2nd gear and concentrate on holding each cell for at least a few seconds to get a solid reading. Obviously you wont be able to do the high-load, low-rpm ones (unless you use the breaks...), but this should get you a consistent reading. The key is to make slow, smooth transitions while moving between cells.
Go on the interstate at some non-busy hour and put it in 2nd gear and concentrate on holding each cell for at least a few seconds to get a solid reading. Obviously you wont be able to do the high-load, low-rpm ones (unless you use the breaks...), but this should get you a consistent reading. The key is to make slow, smooth transitions while moving between cells.
01-07-2013, 12:22 PM
#10
Well part of the issue is that the provided injector data is much more precise than the software/PCM can handle. I ran across this issue before, the software/PCM will round up or down the data to fit... For example see below.
I have started over with a smooth VE table to do some adjustments. I found previously when i was NA that if i started with a smooth table and kept it smooth the results were much more consistent... I will take a log on my way home tonight to see how it is doing....
The LS1 PCM (and most* vehicle controllers) are calibrated using integers (i.e. whole numbers) to reduce the cost and processing load on the CPU.
High precision floating-point values require dedicated hardware (more expensive) or software floating-point routines which slow down calculations.
* Note: Later model controllers use CPUs with built in floating point hardware and they do make limited use of high precision floating point data.
Storing calibration data as integers like the LS1 PCM does, allows only a discreet set of scaled floating point values.
For example if the PCM requires a table to contain data from 0.00-100.00 (for example throttle percentage) using only the set of integers from 0..255 (i.e. a single byte) then it would scale that set of integers by dividing each one by 2.55.
That would mean the following conversion would be in effect:
0 = 0.0
1 = 0.392157
2 = 0.784314
...
245 = 99.607843
255 = 100.0
As you can see, if the PCM can only store the values from 0..255, then when it has a 1 stored the PCM will interpret it as 0.392157, when it has a 2 stored the PCM will interpret it as 0.784314.
Since it is impossible to store an integer value between 1 and 2, which means it is also impossible to represent any value between the two discreet values: 0.392157 and 0.784314.
What you are seeing in the B3701 table is the effect of the limits of the integer to floating point scaling that I described above - just with different scaling.
The EFILive tuning software allows the user to enter full precision floating point values. It has to do that to preserve as much precision as possible while the user is editing/manipulating the data. Because of that you can enter any floating point value that you want into the tables. When the data is saved back to the tune file, the software must convert the final floating point values back to their nearest scaled integer values that the PCM requires.
There is no way to calibrate any values between the two values (0.061867 and 0.077333) that you are seeing. Its either one or the other.
Regards
Paul
High precision floating-point values require dedicated hardware (more expensive) or software floating-point routines which slow down calculations.
* Note: Later model controllers use CPUs with built in floating point hardware and they do make limited use of high precision floating point data.
Storing calibration data as integers like the LS1 PCM does, allows only a discreet set of scaled floating point values.
For example if the PCM requires a table to contain data from 0.00-100.00 (for example throttle percentage) using only the set of integers from 0..255 (i.e. a single byte) then it would scale that set of integers by dividing each one by 2.55.
That would mean the following conversion would be in effect:
0 = 0.0
1 = 0.392157
2 = 0.784314
...
245 = 99.607843
255 = 100.0
As you can see, if the PCM can only store the values from 0..255, then when it has a 1 stored the PCM will interpret it as 0.392157, when it has a 2 stored the PCM will interpret it as 0.784314.
Since it is impossible to store an integer value between 1 and 2, which means it is also impossible to represent any value between the two discreet values: 0.392157 and 0.784314.
What you are seeing in the B3701 table is the effect of the limits of the integer to floating point scaling that I described above - just with different scaling.
The EFILive tuning software allows the user to enter full precision floating point values. It has to do that to preserve as much precision as possible while the user is editing/manipulating the data. Because of that you can enter any floating point value that you want into the tables. When the data is saved back to the tune file, the software must convert the final floating point values back to their nearest scaled integer values that the PCM requires.
There is no way to calibrate any values between the two values (0.061867 and 0.077333) that you are seeing. Its either one or the other.
Regards
Paul