How do you measure WOT performance while street tuning?
08-27-2010, 04:08 PM
#1
How do you measure WOT performance while street tuning? I get my wideband hopefully next week, and I know that I can do most of the tuning on the street, but how do you get any kind of input on actual performance at WOT? Is there a way to measure it? Obviously that is what dynos are for, but it seems like there has to be a way get some kind of feed back other than "seat of the pants"
Input?
Input?
08-27-2010, 04:28 PM
#2
Moderator
iTrader: (11)
Join Date: Mar 2002
Location: East Central Florida
Posts: 12,604
Likes: 0
Received 6 Likes
on
6 Posts
I like using an accelerometer, which I can log via the
EIO cable. I have made them using eval boards from
Analog Devices and FreeScale, and also hacked an old
(red LED style) G-Tech to pigtail out the accelerometer
voltage (those use one of the older ADI parts).
You'd want to lock the converter, perhaps using a
modified "tuning tune" that locks above a safe minimum
road speed for stalling and unlocks at near-idle throttle
positions. Doing it by the keyboard is too distracting.
EIO cable. I have made them using eval boards from
Analog Devices and FreeScale, and also hacked an old
(red LED style) G-Tech to pigtail out the accelerometer
voltage (those use one of the older ADI parts).
You'd want to lock the converter, perhaps using a
modified "tuning tune" that locks above a safe minimum
road speed for stalling and unlocks at near-idle throttle
positions. Doing it by the keyboard is too distracting.
08-27-2010, 05:21 PM
#3
9 Second Club
iTrader: (2)
Join Date: Jul 2002
Location: on the dyno tuning in MD
Posts: 2,583
Likes: 0
Received 1 Like
on
1 Post
That is the hard part about street tuning... If I do it, I go based on experience for the setup and use the wideband to get me there. In other words, I know what to target AFR to, and then use the wideband to get there. Same for timing - just go with what I know from dyno experience. It isn't the way to max out a setup... but you can get a good safe tune that way. The other alternative I use is track tuning - then you have timeslips to go by...
08-27-2010, 07:34 PM
#4
If you are HPT, log the Torque Delivered.
Use it like a dyno run, 2000 rpm to 6000 rpm eg., in what ever gear.
Just do it consistently, I have a nice up hill grade I like to use in third gear.
Works very well, when I went to the dyno I picked maybe 5hp, I new why & were from the street tune.
Use it like a dyno run, 2000 rpm to 6000 rpm eg., in what ever gear.
Just do it consistently, I have a nice up hill grade I like to use in third gear.
Works very well, when I went to the dyno I picked maybe 5hp, I new why & were from the street tune.
08-27-2010, 07:50 PM
#5
Moderator
iTrader: (11)
Join Date: Mar 2002
Location: East Central Florida
Posts: 12,604
Likes: 0
Received 6 Likes
on
6 Posts
The problem with Delivered Torque is that it's calculated
based on the same things you are likely changing - spark
and air mass / airflow - and assumptions that may have
been changed by your mods (CR, cam, exhaust, etc.).
based on the same things you are likely changing - spark
and air mass / airflow - and assumptions that may have
been changed by your mods (CR, cam, exhaust, etc.).
08-27-2010, 11:00 PM
#7
I just need a nice simple way of something that measures performance and then see if it increases or decrease.. I dont care what it is.. somebody has an answer, I don't know that I'm smart enough to rewire a accelerometer and log it and quantify it somehow.. I'm a finance student not an engineer
Trending Topics
08-28-2010, 06:54 AM
#8
9 Second Club
iTrader: (2)
Join Date: Jul 2002
Location: on the dyno tuning in MD
Posts: 2,583
Likes: 0
Received 1 Like
on
1 Post
Just use a G-tech or similar - like Jimmy is talking about - only don't do all the complicated hooking it in to log it. I believe they record each run... However, they are not all that accurate if you ask me... Honestly, by the time you do all this, you would be better off just getting it safe on the street and then paying for an hour on a dyno to fine tune it...
08-28-2010, 07:55 AM
#9
Just use a G-tech or similar - like Jimmy is talking about - only don't do all the complicated hooking it in to log it. I believe they record each run... However, they are not all that accurate if you ask me... Honestly, by the time you do all this, you would be better off just getting it safe on the street and then paying for an hour on a dyno to fine tune it...
08-28-2010, 07:59 AM
#10
Without a dyno, WOT tuning is basically getting your A/F where you wnat it and taking an educated guess on timing.
I just shoot for .87 Lambda and 24-28degrees depending on the engine. Itll get you in the ballpark. If you must have every last HP, you have to rent a dyno.
I just shoot for .87 Lambda and 24-28degrees depending on the engine. Itll get you in the ballpark. If you must have every last HP, you have to rent a dyno.
08-28-2010, 08:49 AM
#11
Here is what I do know, I've log this consistently for 1.5 years. In this time frame with the same vehicle it has seen at least three different dyno's. The actual numbers may not agree, the gains & losses are extremely consistent. The gains & losses are the most important of all.
I'm extremely trusting of Delivered Torque, I use it to verify the Dyno as I use Airflow to verify both.
So I agree with the many factor's you mention, keep in mind the one you cannot change is Airflow. Airflow is a result of a change you have made such as timing & fuel.
HPT gives people the impression they are changing airflow, I dislike that part. I understand it is only a definition.
So to the OP use the PID as another tool, you will find it consistent. At least then you have measurement to work with & a very good one. To improve you have to be able measure.
We are having a problem right now with a particular dyno & how it is reporting. The Torque Delivered & MAF logs are giving us valuable data to confirm the error with dyno. If we did not have those two PID's I'm not sure how we could have identified the problem with this dyno.
They are all just tools, it's up to the individual how useful the tool's are..
Last edited by ctd; 08-28-2010 at 09:08 PM. Reason: Clarified
08-28-2010, 08:58 AM
#12
Maybe, I'm not near wise enough to argue.
Here is what I do know, I've log this consistently for 1.5 years. In this time frame with the same vehicle it has seen at least three different dyno's. The actual numbers may not agree, the gains & losses are extremely consistent. The gains & losses are the most important of all.
I'm extremely trusting of Delivered Torque, I use it to verify the Dyno as I use Airflow to verify both.
So I agree with the many factor's you mention, keep in mind the one cannot change is Airflow. Airflow is a result of a change.
HPT gives people the impression they are changing airflow, I dislike that part. I understand it is only a definition.
So to the OP use the PID as another tool, you will find it consistent. At least then you have measurement to work with & a very good one. To improve you have to be able measure.
We are having a problem right now with a particular dyno & how it is reporting. The Torque Delivered & MAF logs are giving us valuable data to confirm the error with dyno. If we did not have those two PID's I'm not sure how we could have identified the problem with this dyno.
They are all just tools, it's up to the individual how useful the tool is.
Here is what I do know, I've log this consistently for 1.5 years. In this time frame with the same vehicle it has seen at least three different dyno's. The actual numbers may not agree, the gains & losses are extremely consistent. The gains & losses are the most important of all.
I'm extremely trusting of Delivered Torque, I use it to verify the Dyno as I use Airflow to verify both.
So I agree with the many factor's you mention, keep in mind the one cannot change is Airflow. Airflow is a result of a change.
HPT gives people the impression they are changing airflow, I dislike that part. I understand it is only a definition.
So to the OP use the PID as another tool, you will find it consistent. At least then you have measurement to work with & a very good one. To improve you have to be able measure.
We are having a problem right now with a particular dyno & how it is reporting. The Torque Delivered & MAF logs are giving us valuable data to confirm the error with dyno. If we did not have those two PID's I'm not sure how we could have identified the problem with this dyno.
They are all just tools, it's up to the individual how useful the tool is.
08-28-2010, 09:44 AM
#13
So you are saying you believe by logging delivered torque, it is consistent enough that if I make a change and it increases, that i know i just improved? I don't care about if the actual number is correct, I just want something to show if a change moves me forward or backwards.. this is assuming similar conditions..
You can target a certain rpm range, such as a dip in a dyno run. You now can measure a specific area, make changes & verify loss or gain.
08-30-2010, 06:52 PM
#15
"The Tuning School - Advanced Course" talks about using the delivered torque as CTD described. I played with it last week and it was pretty interesting. I setup a histogram (go to my web site and download it Engine Delivered Torque.hst) and logged like they described, 1000 rpms for 10 seconds, 1500 for 10, etc, up to 4000 rpms at first and later up to WOT.
They have you do timing changes, no fueling. This is all done after you tuned the MAF and VE tables so the only change is timing.
I did this 3 times, all cells logged within a couple lb ft of torque. I made a timing adjustment to small ranges at a time and relogged, if the torque increased I went up again if it went down I took away timing. I have a section in the mid-range that I added 4 degrees in total and I showed a 10 lb ft gain. Around my peak torque, 4800 or so, I add two degrees and showed a 20 lb ft increase.
It may not be perfect or be a dyno, I call it a poor mans dyno, but it looks like if done correctly it can show you if you are going in the right direction.
Do not believe the numbers (they seem high) but just look for increases or decreases in delivered torque.
They have you do timing changes, no fueling. This is all done after you tuned the MAF and VE tables so the only change is timing.
I did this 3 times, all cells logged within a couple lb ft of torque. I made a timing adjustment to small ranges at a time and relogged, if the torque increased I went up again if it went down I took away timing. I have a section in the mid-range that I added 4 degrees in total and I showed a 10 lb ft gain. Around my peak torque, 4800 or so, I add two degrees and showed a 20 lb ft increase.
It may not be perfect or be a dyno, I call it a poor mans dyno, but it looks like if done correctly it can show you if you are going in the right direction.
Do not believe the numbers (they seem high) but just look for increases or decreases in delivered torque.
08-30-2010, 07:00 PM
#16
You can only get so far optimizing your tune on the street.
The most obvious method is to look at the time slips from the drag strip, or look at the times of achieving a certain speed in the logs. I personally like a 3rd gear 40-100mph pulls, most cars can do it and cover a good range of RPM.
But that method has one huge flaw: you're looking an aggregate result of all the changes, so a loss of power at lower RPM might numerically 'hide' the gains at higher RPM. So to separate these out, you start partitioning data out to separate the high and the low ranges. But that's still not enough, because you can encounter the same sort of gain/loss scenario within even a smaller intervals of speed, rpm, or time, whatever you use for the basis of the comparison. So you keep sub-segregating the intervals, until you realize your data doesn't have enough resolution/samples to be really useful.
So then you try to figure out acceleration using derivative based measurements of speed and time, and get severely disappointed as the lack of resolution on the speed PID combined with large time intervals create a horribly noisy graph beyond the point of being useful. There's been multiple people over the last 5-6yrs that have created various tools for it and they all falter due to the same low-res data inputs. If you're the curious type, I highly recommend finding them, they're not very useful, but they are highly educational.
The next thing you try is eliminating wheel-speed data, and relying on GPS to get precise speed/time data. And you get disappointed again, as the timing intervals between samples, combined with lousy GPS precision will also result in meaningless results.
So then you try to measure airmass or torque PID, and use them as instantaneous power figures. That will get you closer, but it's still problematic, as it doesn't normalize for atmospheric conditions, so while you can do comparisons when tweaking in one session, it's a lot less precise to compare data from different days. They're also not precise enough to really see a result of changing timing by a degree or two.
So after all these experiments you end up finding a real load-bearing dyno, holding a given load and measuring instantaneous power. Sometimes it's just easier to do it right from the get-go
The most obvious method is to look at the time slips from the drag strip, or look at the times of achieving a certain speed in the logs. I personally like a 3rd gear 40-100mph pulls, most cars can do it and cover a good range of RPM.
But that method has one huge flaw: you're looking an aggregate result of all the changes, so a loss of power at lower RPM might numerically 'hide' the gains at higher RPM. So to separate these out, you start partitioning data out to separate the high and the low ranges. But that's still not enough, because you can encounter the same sort of gain/loss scenario within even a smaller intervals of speed, rpm, or time, whatever you use for the basis of the comparison. So you keep sub-segregating the intervals, until you realize your data doesn't have enough resolution/samples to be really useful.
So then you try to figure out acceleration using derivative based measurements of speed and time, and get severely disappointed as the lack of resolution on the speed PID combined with large time intervals create a horribly noisy graph beyond the point of being useful. There's been multiple people over the last 5-6yrs that have created various tools for it and they all falter due to the same low-res data inputs. If you're the curious type, I highly recommend finding them, they're not very useful, but they are highly educational.
The next thing you try is eliminating wheel-speed data, and relying on GPS to get precise speed/time data. And you get disappointed again, as the timing intervals between samples, combined with lousy GPS precision will also result in meaningless results.
So then you try to measure airmass or torque PID, and use them as instantaneous power figures. That will get you closer, but it's still problematic, as it doesn't normalize for atmospheric conditions, so while you can do comparisons when tweaking in one session, it's a lot less precise to compare data from different days. They're also not precise enough to really see a result of changing timing by a degree or two.
So after all these experiments you end up finding a real load-bearing dyno, holding a given load and measuring instantaneous power. Sometimes it's just easier to do it right from the get-go
08-30-2010, 07:10 PM
#17
Dang Marcin, your being a party pooper
I don't have any load bearing dynos in my area and a regular inertia one just doing WOT pulls won't do much for 2400 rpms and .32 g/cyl.
Come up with something with math that we can use that we only need to plug in some logged data.
I don't have any load bearing dynos in my area and a regular inertia one just doing WOT pulls won't do much for 2400 rpms and .32 g/cyl.
Come up with something with math that we can use that we only need to plug in some logged data.
08-30-2010, 08:49 PM
#18
i'm no party pooper, i'm saving you time and money by telling you how i've wasted years of time and thousands of money
And yes, I've tried to 'fix it with math' but ultimately the data just isn't precise enough. Garbage in, garbage out, as the old adage says.
And yes, I've tried to 'fix it with math' but ultimately the data just isn't precise enough. Garbage in, garbage out, as the old adage says.
08-30-2010, 09:43 PM
#20
You can only get so far optimizing your tune on the street.
The most obvious method is to look at the time slips from the drag strip, or look at the times of achieving a certain speed in the logs. I personally like a 3rd gear 40-100mph pulls, most cars can do it and cover a good range of RPM.
But that method has one huge flaw: you're looking an aggregate result of all the changes, so a loss of power at lower RPM might numerically 'hide' the gains at higher RPM. So to separate these out, you start partitioning data out to separate the high and the low ranges. But that's still not enough, because you can encounter the same sort of gain/loss scenario within even a smaller intervals of speed, rpm, or time, whatever you use for the basis of the comparison. So you keep sub-segregating the intervals, until you realize your data doesn't have enough resolution/samples to be really useful.
So then you try to figure out acceleration using derivative based measurements of speed and time, and get severely disappointed as the lack of resolution on the speed PID combined with large time intervals create a horribly noisy graph beyond the point of being useful. There's been multiple people over the last 5-6yrs that have created various tools for it and they all falter due to the same low-res data inputs. If you're the curious type, I highly recommend finding them, they're not very useful, but they are highly educational.
The next thing you try is eliminating wheel-speed data, and relying on GPS to get precise speed/time data. And you get disappointed again, as the timing intervals between samples, combined with lousy GPS precision will also result in meaningless results.
So then you try to measure airmass or torque PID, and use them as instantaneous power figures. That will get you closer, but it's still problematic, as it doesn't normalize for atmospheric conditions, so while you can do comparisons when tweaking in one session, it's a lot less precise to compare data from different days. They're also not precise enough to really see a result of changing timing by a degree or two.
So after all these experiments you end up finding a real load-bearing dyno, holding a given load and measuring instantaneous power. Sometimes it's just easier to do it right from the get-go
The most obvious method is to look at the time slips from the drag strip, or look at the times of achieving a certain speed in the logs. I personally like a 3rd gear 40-100mph pulls, most cars can do it and cover a good range of RPM.
But that method has one huge flaw: you're looking an aggregate result of all the changes, so a loss of power at lower RPM might numerically 'hide' the gains at higher RPM. So to separate these out, you start partitioning data out to separate the high and the low ranges. But that's still not enough, because you can encounter the same sort of gain/loss scenario within even a smaller intervals of speed, rpm, or time, whatever you use for the basis of the comparison. So you keep sub-segregating the intervals, until you realize your data doesn't have enough resolution/samples to be really useful.
So then you try to figure out acceleration using derivative based measurements of speed and time, and get severely disappointed as the lack of resolution on the speed PID combined with large time intervals create a horribly noisy graph beyond the point of being useful. There's been multiple people over the last 5-6yrs that have created various tools for it and they all falter due to the same low-res data inputs. If you're the curious type, I highly recommend finding them, they're not very useful, but they are highly educational.
The next thing you try is eliminating wheel-speed data, and relying on GPS to get precise speed/time data. And you get disappointed again, as the timing intervals between samples, combined with lousy GPS precision will also result in meaningless results.
So then you try to measure airmass or torque PID, and use them as instantaneous power figures. That will get you closer, but it's still problematic, as it doesn't normalize for atmospheric conditions, so while you can do comparisons when tweaking in one session, it's a lot less precise to compare data from different days. They're also not precise enough to really see a result of changing timing by a degree or two.
So after all these experiments you end up finding a real load-bearing dyno, holding a given load and measuring instantaneous power. Sometimes it's just easier to do it right from the get-go
Dang, you just owned this whole thread... lol
Although i think i may still use some of the delivered engine torque data to seek improvements.. its better than flying completely blind..