Intelligent question about DIY standalone tuning by one person
08-07-13, 01:27 AM
#1
Intelligent question about DIY standalone tuning by one person
I've done a search on this subject, but have come up with nada, so please excuse the n00bness of this question.
When tuning a standalone, I understand that getting the AFRs in the neighborhood can be done in neutral by a single person in the manner of: Rev to XXXXrpm, check AFR, enter correction in fuel map, re-check AFR, move to next RPM point, repeat.
After doing 3-4 rpm points in neutral, give the car time to cool down sufficiently. I personally run it to 210 degrees, then let it cool down to 130 degrees.
On my currently-N/A 20B, I've already gone through these motions and the fuel map is in the 12.5-13.5 range all the way to 8000rpm. Car idles and revs okay, but floods in under a mile so the mixture is still too rich. During my test drive, I had both a video and datalog running, but cannot pinpoint the areas in which the car is fouling out due to being behind the wheel in traffic when it occurs. Examining the datalog doesn't reveal any sort of 'trigger' for it that I can discern such as a certain cell on the fuel map being significantly too rich.
So my question is this: How does everyone actually do their tuning? Is a second person necessary, with one driving and giving feedback on the car's behavior while the other makes the adjustments on the laptop? Or is there a trick to doing this such as driving for a short while, then pulling over? This seems very unlikely to work due to one's inability to see the third axis (load) on the fuel map unless viewing it as the car is in motion. Plus, one can't exactly stop every 2 minutes to try this. And let's not even suggest attempting to drive and tune at the same time by yourself...
When tuning a standalone, I understand that getting the AFRs in the neighborhood can be done in neutral by a single person in the manner of: Rev to XXXXrpm, check AFR, enter correction in fuel map, re-check AFR, move to next RPM point, repeat.
After doing 3-4 rpm points in neutral, give the car time to cool down sufficiently. I personally run it to 210 degrees, then let it cool down to 130 degrees.
On my currently-N/A 20B, I've already gone through these motions and the fuel map is in the 12.5-13.5 range all the way to 8000rpm. Car idles and revs okay, but floods in under a mile so the mixture is still too rich. During my test drive, I had both a video and datalog running, but cannot pinpoint the areas in which the car is fouling out due to being behind the wheel in traffic when it occurs. Examining the datalog doesn't reveal any sort of 'trigger' for it that I can discern such as a certain cell on the fuel map being significantly too rich.
So my question is this: How does everyone actually do their tuning? Is a second person necessary, with one driving and giving feedback on the car's behavior while the other makes the adjustments on the laptop? Or is there a trick to doing this such as driving for a short while, then pulling over? This seems very unlikely to work due to one's inability to see the third axis (load) on the fuel map unless viewing it as the car is in motion. Plus, one can't exactly stop every 2 minutes to try this. And let's not even suggest attempting to drive and tune at the same time by yourself...
08-07-13, 02:25 AM
#2
talking head
get the wideband 5v aux out wired back as input to the ecu and then put the lappy in car for data log
.. or if your ecu doesnt log then go for external watchdog like combining plx wideband and exhaust temp and map units
and then watching trend using the software on pc .. or having several dm6 smart gauges on trend view .. and setting the watchdog warnings as you wish
its just observation of the problem rev and map point and catching the bad trend on the wideband feedback mindful of a little lag in the recording
.. having the ms pulse on the trend view in the pc datalog is possibly the most useful clue as to matching the bad cell and the delay in the afr trend
.. or if your ecu doesnt log then go for external watchdog like combining plx wideband and exhaust temp and map units
and then watching trend using the software on pc .. or having several dm6 smart gauges on trend view .. and setting the watchdog warnings as you wish
its just observation of the problem rev and map point and catching the bad trend on the wideband feedback mindful of a little lag in the recording
.. having the ms pulse on the trend view in the pc datalog is possibly the most useful clue as to matching the bad cell and the delay in the afr trend
Last edited by bumpstart; 08-07-13 at 02:28 AM.
08-07-13, 03:38 AM
#3
Yes, I have an Innovate LC1 Wideband and it is wired into the ECU to provide correct feedback. Its gauge is spliced into the harness so both the ECU & gauge receive the exact same information. This has been the case since I started tuning the car.
ECU is a Haltech Platinum Sport 2000, plenty capable of datalogging, but I don't think you quite understand the question I am asking. With the laptop connected & the engine running, the fuel map has a set of blue crosshairs that hover over the current cell in which the engine is currently in, based on RPM & Manifold Pressure. In neutral, this is easy enough to understand and correct: hold the engine at a set rpm, find the target, put in a correction and check the AFR again. Lather, rinse, repeat for the next data point.
However, the issue is that it doesn't have this targeting feature when replaying a data log in ECU Manager, so one can only guess which cell(s) are needing adjustment and by how much while hoping it doesn't disrupt the car's behavior in the surrounding cells. For the sake of clarity, let's refer to this log-and-change method as "Alternating Tuning" as it is alternating between driving and tuning.
I have attempted this and it works to make an engine go from flooding itself in 10 minutes to flooding in one mile. However, the problem is that one would have to make an identical run at least 3 times while having only one change in order to accurately pinpoint a trouble spot by the scientific method. From what I have seen, none of my datalogs have been anywhere near identical, even when made back to back. Between this and trying to smooth out AFRs that change 10 times per second, how could one possibly expect to approach the problem logically when you can't make your changes in real time to offset these fluctuations? Instead of one or two cells being affected, the little blue crosshairs could easily be dancing around through 7 cells in a circle as I have witnessed before. And with the datalog unable to reproduce the pattern on the fuel map, it becomes even more difficult to narrow it down to a specific issue...
If the issue were a bit more consistent in behavior, I can see how it could be corrected in this manner. However, given the rapidity of these fluctuations, I hope this illustrates the flaw and limitations in the Alternating Tuning strategy. If one is able to make the appropriate corrections based on such quickly changing data, I would love to see it and believe it should be required viewing for all standalone newcomers.
With all of this in mind, let me re-iterate the question:
How does one tune properly when it can't be done in real time without another person present?
ECU is a Haltech Platinum Sport 2000, plenty capable of datalogging, but I don't think you quite understand the question I am asking. With the laptop connected & the engine running, the fuel map has a set of blue crosshairs that hover over the current cell in which the engine is currently in, based on RPM & Manifold Pressure. In neutral, this is easy enough to understand and correct: hold the engine at a set rpm, find the target, put in a correction and check the AFR again. Lather, rinse, repeat for the next data point.
However, the issue is that it doesn't have this targeting feature when replaying a data log in ECU Manager, so one can only guess which cell(s) are needing adjustment and by how much while hoping it doesn't disrupt the car's behavior in the surrounding cells. For the sake of clarity, let's refer to this log-and-change method as "Alternating Tuning" as it is alternating between driving and tuning.
I have attempted this and it works to make an engine go from flooding itself in 10 minutes to flooding in one mile. However, the problem is that one would have to make an identical run at least 3 times while having only one change in order to accurately pinpoint a trouble spot by the scientific method. From what I have seen, none of my datalogs have been anywhere near identical, even when made back to back. Between this and trying to smooth out AFRs that change 10 times per second, how could one possibly expect to approach the problem logically when you can't make your changes in real time to offset these fluctuations? Instead of one or two cells being affected, the little blue crosshairs could easily be dancing around through 7 cells in a circle as I have witnessed before. And with the datalog unable to reproduce the pattern on the fuel map, it becomes even more difficult to narrow it down to a specific issue...
If the issue were a bit more consistent in behavior, I can see how it could be corrected in this manner. However, given the rapidity of these fluctuations, I hope this illustrates the flaw and limitations in the Alternating Tuning strategy. If one is able to make the appropriate corrections based on such quickly changing data, I would love to see it and believe it should be required viewing for all standalone newcomers.
With all of this in mind, let me re-iterate the question:
How does one tune properly when it can't be done in real time without another person present?
08-07-13, 04:38 AM
#4
talking head
observation of hesitation and course spots and mentally noting their rev and load points during the logging run is your best friend
have the data logging start on from a switch command
and set up so you can see from the TPS feed back when you go open throttle
you should also log the rev and map and ms plots and the afr plot
the little lag in the afr trends can be gauged from a difference in recording from when you go WOT on the TPS and when you see the reaction in the AFR plot
you can then use this to align any afr glitches back a little with the appropriate rev and map points
and then use this guesstimate,, and the displayed ms around or just before that time point to go back , find and adjust the appropriate tables
handy also during the run is tuner gauge page..
and or the corrections pages up to see if the change required is actually in the map.. or the correction factors at play
mindful misfires can indicate as lean ( but yet cold on EGT )
.. and you should always err the cell richer first .. to prove its already too rich..
and only make steps.. one at a time ..
have the data logging start on from a switch command
and set up so you can see from the TPS feed back when you go open throttle
you should also log the rev and map and ms plots and the afr plot
the little lag in the afr trends can be gauged from a difference in recording from when you go WOT on the TPS and when you see the reaction in the AFR plot
you can then use this to align any afr glitches back a little with the appropriate rev and map points
and then use this guesstimate,, and the displayed ms around or just before that time point to go back , find and adjust the appropriate tables
handy also during the run is tuner gauge page..
and or the corrections pages up to see if the change required is actually in the map.. or the correction factors at play
mindful misfires can indicate as lean ( but yet cold on EGT )
.. and you should always err the cell richer first .. to prove its already too rich..
and only make steps.. one at a time ..
08-07-13, 06:38 AM
#5
As you already found out revving in neutral will only get you on the very bottom part of the fuel map. Datalogging and making corrections after driving seems like an afwull way to tune it, your car will spend a lot of time running around with an incorrect fuel mixture, and the ending result won't be great. Is it really impossible for you to get someone to drive the car while you do the tuning? I tuned my own FD a few weeks ago, did most of it on the dyno but I finetuned the cruising areas by having someone drive my car while I messed with the ecu, which worked great! Got the whole fuel map covered on the dyno in about 20 minutes, and finetuning took about an hour or so. How is the autotuning feature on the haltech? Maybe you can use that to get a driveable map.
Last edited by John Huijben; 08-07-13 at 06:41 AM.
08-07-13, 02:00 PM
#6
As you already found out revving in neutral will only get you on the very bottom part of the fuel map. Datalogging and making corrections after driving seems like an afwull way to tune it, your car will spend a lot of time running around with an incorrect fuel mixture, and the ending result won't be great. Is it really impossible for you to get someone to drive the car while you do the tuning? I tuned my own FD a few weeks ago, did most of it on the dyno but I finetuned the cruising areas by having someone drive my car while I messed with the ecu, which worked great! Got the whole fuel map covered on the dyno in about 20 minutes, and finetuning took about an hour or so. How is the autotuning feature on the haltech? Maybe you can use that to get a driveable map.
This situation calls for someone with competence akin to a FD owner. It's not a slight toward FC owners as there are quite a few with FD-level or greater comprehension such as Aaron Cake, but the majority of FC owners are still trying to figure out how a ground works
As my build has more in common with a FD than a FC now, it just needs someone that understands behaviors more commonly attributed to a FD. Of course, if Aaron Cake showed up here, I'd be falling over myself to give him the keys
Currently have a request on Ohio Rotaries
08-07-13, 02:46 PM
#7
Moderator
iTrader: (3)
Join Date: Mar 2001
Location: https://www2.mazda.com/en/100th/
Posts: 30,761
Received 2,556 Likes
on
1,819 Posts
i've actually found you need almost no skill for the cruising parts of the map, you just have to find someone who can hold a steady speed and not hit stuff.
i tuned my 20B by myself, all you need on the haltech is the home key, and then the pg up and down. obviously once it got to WOT and like 4000rpm, things were happening too fast to really do this...
this was in 2001, when using a phone in the car was perfectly fine.
the data log and change method should work fine too, but you want lots of parts where RPM and load are stable, which might be a bit tricky. once you've sampled the RPM and load for a few seconds, then you should be able to go to that cell and make changes.
or method 3 is just to give the o2 sensor lots of authority and just run it in closed loop all the time, the Mitsubishi guys were running 9's in the quarter mile like this, and BMW* factory ecu's are like this as well.
*BMW ran a narrowband o2 sensor into an amplifier, to make it a 0-5v signal, and then just let the car (90's 5 series) run from that.
i tuned my 20B by myself, all you need on the haltech is the home key, and then the pg up and down. obviously once it got to WOT and like 4000rpm, things were happening too fast to really do this...
this was in 2001, when using a phone in the car was perfectly fine.
the data log and change method should work fine too, but you want lots of parts where RPM and load are stable, which might be a bit tricky. once you've sampled the RPM and load for a few seconds, then you should be able to go to that cell and make changes.
or method 3 is just to give the o2 sensor lots of authority and just run it in closed loop all the time, the Mitsubishi guys were running 9's in the quarter mile like this, and BMW* factory ecu's are like this as well.
*BMW ran a narrowband o2 sensor into an amplifier, to make it a 0-5v signal, and then just let the car (90's 5 series) run from that.
Trending Topics
08-19-13, 03:49 AM
#8
The "intelligent" solution is to put the car on a proper dyno. More precise and a whole lot safer than road tuning by any means.
Specifically speaking about closed loop, since I know you're already having issues with this; turn it off until you have the base calibration done. Otherwise, you're fighting the ECU the whole way while you're trying to calibrate the fuel map. The fuel map, once it's done, should be close enough that closed loop operation won't need to add or pull more than a couple percent. If it has to to achieve target mixtures, the base fuel map is wrong.
Specifically speaking about closed loop, since I know you're already having issues with this; turn it off until you have the base calibration done. Otherwise, you're fighting the ECU the whole way while you're trying to calibrate the fuel map. The fuel map, once it's done, should be close enough that closed loop operation won't need to add or pull more than a couple percent. If it has to to achieve target mixtures, the base fuel map is wrong.
08-19-13, 05:22 AM
#9
I'll also say there is a specific order in which you calibrate each map. The Haltech instructions hint at this, more or less, but there are better resources for the process in general. Hartman's How To Tune And Modify Engine Management Systems is one of the best I've read. http://www.amazon.com/gp/aw/d/076031...sr=8-1&pi=SL75. Greg Banish also has a pretty good book on the subject.
08-28-13, 12:12 PM
#10
Moderator
iTrader: (3)
Join Date: Mar 2001
Location: https://www2.mazda.com/en/100th/
Posts: 30,761
Received 2,556 Likes
on
1,819 Posts
I'll also say there is a specific order in which you calibrate each map. The Haltech instructions hint at this, more or less, but there are better resources for the process in general. Hartman's How To Tune And Modify Engine Management Systems is one of the best I've read. How to Tune and Modify Engine Management Systems (Motorbooks Workshop):Amazon:Books. Greg Banish also has a pretty good book on the subject.
with the haltech its even more important, the fuel map is 32x32, that is 1024 cells! when you add the ignition, another 1024 points, plus all the corrections, its just too much for the human mind to deal with, so you have to break it down into chunks.
i've also combed through the FSM's and training manuals, and have collected as much of the correction maps as i can, so it may not be 100% right, but it does save some brainpower
Thread
Thread Starter
Forum
Replies
Last Post
armans
3rd Generation Specific (1993-2002)
5
08-15-15 09:08 PM
