couple ve tuning questions,ls1 and hptuners.
11-14-2013, 02:49 PM
#1
10 Second Club
Thread Starter
iTrader: (22)
Join Date: Mar 2006
Location: Northwest side of Chicago
Posts: 3,677
Likes: 0
Received 2 Likes
on
2 Posts
couple ve tuning questions,ls1 and hptuners. I'm doing the ve portion of my tune on the street. Im logging afr error, afr and stft.
How can a cell populate in afr error and not have that same cell populate in the afr histo?
When logging normal driving there are random cells that say either plus or minus a large number like 15, 20, 25. Do I literaly add or subract that large of a number in the ve table for just that one random cell?
Why can't the histogram populate every cell? Im going for a hour cruise and trying to hit every possible load ,rpm, boost level.
Alex
How can a cell populate in afr error and not have that same cell populate in the afr histo?
When logging normal driving there are random cells that say either plus or minus a large number like 15, 20, 25. Do I literaly add or subract that large of a number in the ve table for just that one random cell?
Why can't the histogram populate every cell? Im going for a hour cruise and trying to hit every possible load ,rpm, boost level.
Alex
11-14-2013, 08:08 PM
#3
Launching!
Join Date: Jan 2012
Location: Davisburg, MI
Posts: 205
Likes: 0
Received 0 Likes
on
0 Posts
I'm using EFI-L so I can't specifically answer your first question. There is a lot of filtering of data, I assume the same in HPT. The cells are filtered out in one chart and not the other.
You should throw out the large numbers. I did better looking at the trends and extrapolating the data. In EFI-L I can look at the high and low for each cell and see there can be a significant range. That is why you need a lot of hits to average this out.
It is impossible to hit every cell. The car just doesn't operate in a mathematical square. Normal driving you will notice more of a fat ellipse where the engine operates. I found crazy driving on the street trying to hit more cells just gave me bad data. Again, better to extrapolate from good data.
You should throw out the large numbers. I did better looking at the trends and extrapolating the data. In EFI-L I can look at the high and low for each cell and see there can be a significant range. That is why you need a lot of hits to average this out.
It is impossible to hit every cell. The car just doesn't operate in a mathematical square. Normal driving you will notice more of a fat ellipse where the engine operates. I found crazy driving on the street trying to hit more cells just gave me bad data. Again, better to extrapolate from good data.
11-15-2013, 12:12 PM
#4
TECH Addict
You can't hit every cell but when you go for a drive try to find a hill so you can hit those high load areas. If i remember correctly you multiply by %. So if the cell was 50 and you have a +5 you would end up multiplying that cell by 105% to get 52.5 . If it was -5 you would multiply by .95 to get 47.5. I had better luck multiplying by the 1/2 percent then the full % because i has over shooting multiplying by the full %.
11-15-2013, 03:01 PM
#5
10 Second Club
Thread Starter
iTrader: (22)
Join Date: Mar 2006
Location: Northwest side of Chicago
Posts: 3,677
Likes: 0
Received 2 Likes
on
2 Posts
Whats weird is there are some cells I've added or subtracted a lot from and they still show up in the histogram with large numbers in afr error.
they are odd cells like 30kpa at 2800 says to add like 28. Then 90kpa at 3600 says to subtract 22. But the cell right next to it is either close to 0 or doesn't populate.
they are odd cells like 30kpa at 2800 says to add like 28. Then 90kpa at 3600 says to subtract 22. But the cell right next to it is either close to 0 or doesn't populate.
11-15-2013, 03:59 PM
#6
Moderator
iTrader: (11)
Join Date: Mar 2002
Location: East Central Florida
Posts: 12,604
Likes: 0
Received 6 Likes
on
6 Posts
There is filtering but there's a problem with that,
which is that only the in-the-moment data is used.
You can't, for example, reject a datapoint that is
from an unsteady MAP or unsteady RPM period of
the log. Any data like that is full of transient crud,
some sensors are fast and some are slow and the
relationship sort of falls apart so it's bad for tuning
feedback, to use it.
As a result I've never gotten on board the paste-histo
love train. I take my data by eye from the logs, where
I believe it's most trustworthy, I paste it in by hand
where it belongs and then hand-dress the surrounding
cells in a way that makes sense to me, looking at the
surface (3D or 2D view) and using interpolation function
if it seems sensible, hand-extrapolating or -interpolating
if it doesn't look like a linear blend is the best idea.
Maybe you do or maybe you don't have a gut level
sense for this. But beware the garbage-in, garbage-out
nature of tuning and the untrustworthiness of the
sensors you've got (or, the limited conditions under
which they can be trusted). Garbage is going to come
in, that's unavoidable in a street log (and just about
anywhere except a slow changing, load-holding dyno).
What you want is the best way of sifting for stuff
worth acting upon, and all I have to offer is shade
tree type advice.
which is that only the in-the-moment data is used.
You can't, for example, reject a datapoint that is
from an unsteady MAP or unsteady RPM period of
the log. Any data like that is full of transient crud,
some sensors are fast and some are slow and the
relationship sort of falls apart so it's bad for tuning
feedback, to use it.
As a result I've never gotten on board the paste-histo
love train. I take my data by eye from the logs, where
I believe it's most trustworthy, I paste it in by hand
where it belongs and then hand-dress the surrounding
cells in a way that makes sense to me, looking at the
surface (3D or 2D view) and using interpolation function
if it seems sensible, hand-extrapolating or -interpolating
if it doesn't look like a linear blend is the best idea.
Maybe you do or maybe you don't have a gut level
sense for this. But beware the garbage-in, garbage-out
nature of tuning and the untrustworthiness of the
sensors you've got (or, the limited conditions under
which they can be trusted). Garbage is going to come
in, that's unavoidable in a street log (and just about
anywhere except a slow changing, load-holding dyno).
What you want is the best way of sifting for stuff
worth acting upon, and all I have to offer is shade
tree type advice.
Trending Topics
11-15-2013, 08:40 PM
#8
11 Second Club
iTrader: (12)
Join Date: Oct 2005
Location: Springhill, FL
Posts: 195
Likes: 0
Received 0 Likes
on
0 Posts
What I like to do is do the 50% multiply and then "save as". Then use the compare feature to compare the new tune with changes to before the changes. Using the combined compare table makes it easy to see excessive spikes and smooth them out. 
11-15-2013, 10:55 PM
#9
10 Second Club
Thread Starter
iTrader: (22)
Join Date: Mar 2006
Location: Northwest side of Chicago
Posts: 3,677
Likes: 0
Received 2 Likes
on
2 Posts
When you guys say to multiple by 50% I'm not sure what you mean.
Is that when let's say at 90kpa and 2800 afr error says to add 6 I just add 50% of that? So I add 3 in they cell?
Is that when let's say at 90kpa and 2800 afr error says to add 6 I just add 50% of that? So I add 3 in they cell?
11-16-2013, 01:52 PM
#10
11 Second Club
iTrader: (12)
Join Date: Oct 2005
Location: Springhill, FL
Posts: 195
Likes: 0
Received 0 Likes
on
0 Posts
The number your AFR error table outputs is the % it is off. Not the direct number you should add/subtract to your VE values(though it may almost work that was with VE values near 100). HPTuners has a copy and paste feature that allows you to copy the entire table from your histogram and then in your editing software multiply the entire table by that % or by half the %.
Here is a decent walkthrough:
http://www.4moores.com/280z/files/HP...structions.pdf
The particular info about using the VE histogram is on page 10-11
Here is a decent walkthrough:
http://www.4moores.com/280z/files/HP...structions.pdf
The particular info about using the VE histogram is on page 10-11
11-19-2013, 06:45 PM
#12
10 Second Club
Thread Starter
iTrader: (22)
Join Date: Mar 2006
Location: Northwest side of Chicago
Posts: 3,677
Likes: 0
Received 2 Likes
on
2 Posts
For the guys that do the copy and paste of the ve data from the histogram. What are you doing about the large number cells that are probably not accurate data? If you copy and paste those are going in to.
I noticed that my large number cells are in areas that are right before it goes into pe mode. Its like the car has reached pe mode but the wb is still at 14.7 or so for a second before it can
I noticed that my large number cells are in areas that are right before it goes into pe mode. Its like the car has reached pe mode but the wb is still at 14.7 or so for a second before it can
11-20-2013, 12:16 PM
#16
10 Second Club
Thread Starter
iTrader: (22)
Join Date: Mar 2006
Location: Northwest side of Chicago
Posts: 3,677
Likes: 0
Received 2 Likes
on
2 Posts
Ok just did some logging and ve tuning using the copy and paste special method and using the compare feature to weed out any large numbers. I left pe enable tps alone and just stayed out of pe. I did a couple 30 minute logs highway and street and copied and pasted. My logs now still look pretty similar as when I started and Im wondering if I havent disabled everything Im supposed to before ve tuning.
I have stft in open loop disabled and ltft disabled. One question here is why do we need to disable stft in open loop? I was under the impression you would never have stft in open because in open your not reading o2 sensor data.
I left my open loop f/a vs coolant vs map alone, should I have put it to 1? I think there might be a stock file in it now where cells are all different. Not sure what to do here, my command afr is still showing 14.6x so does this even matter?
Also the closed loop disable is still at 93 should I put it to 285? Or does it even matter since I disabled stft in open loop?
Im almost there thanks for all the help
I have stft in open loop disabled and ltft disabled. One question here is why do we need to disable stft in open loop? I was under the impression you would never have stft in open because in open your not reading o2 sensor data.
I left my open loop f/a vs coolant vs map alone, should I have put it to 1? I think there might be a stock file in it now where cells are all different. Not sure what to do here, my command afr is still showing 14.6x so does this even matter?
Also the closed loop disable is still at 93 should I put it to 285? Or does it even matter since I disabled stft in open loop?
Im almost there thanks for all the help
11-21-2013, 06:28 PM
#17
Moderator
iTrader: (11)
Join Date: Mar 2002
Location: East Central Florida
Posts: 12,604
Likes: 0
Received 6 Likes
on
6 Posts
The STFT in open loop will try to trim whenever your open loop
commanded AFR is the same as stoich. Which it ought to be, in
a lot of areas. Then it could be useful; in fact this is how I run,
"open loop" with STFTs enabled. Don't like LTFTs much, I tune for
slightly rich and let the STFTs pull to center. You get to skip the
history effects that way.
I think the advice is about, you don't want to use an AFR-error-paste
when there's something (STFT) messing with that error meanwhile
(trying to zero it, and taking some time to do it, means a lot of stuff
in the data-set with questionable basis),
commanded AFR is the same as stoich. Which it ought to be, in
a lot of areas. Then it could be useful; in fact this is how I run,
"open loop" with STFTs enabled. Don't like LTFTs much, I tune for
slightly rich and let the STFTs pull to center. You get to skip the
history effects that way.
I think the advice is about, you don't want to use an AFR-error-paste
when there's something (STFT) messing with that error meanwhile
(trying to zero it, and taking some time to do it, means a lot of stuff
in the data-set with questionable basis),
