When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
I went to the track again Sunday afternoon, this time with correct lev rimiter settings, to record some logs above 7,000 RPM. Ambient temp was 10F cooler than the previous outing, so it was close enough, methought. During the drive to the track, I noticed the engine was running a little rich everywhere, but especially in the cruising areas of the map, and thought that was odd, considering it was colder, and MAT correction is currently disabled. It should have been a little lean, if anything.
Anyway, I tracked it hard for 2 hours, enjoyed no spectacular spins, and frustrated my friend by refusing to be bested by his S2000.
Here is what's strange. I expected the algorithm to lean out everything above 7,000 RPM and lean out the cruising cells. What I did not expect is to see it go substantially richer everywhere else. I expected to see some minor changes, but nothing like this. I kept an eye on my AFRs on the track and never saw it go lean. Weird.
It is also worth noting that the areas it made richer were made leaner last time around, as I had made the whole map rich for safety.
This chart shows the VE table deltas between the map I drove on and the map created by MLV from 30 minutes of logging. The log includes driving to grid, a warmup lap, 13 hot laps with and without traffic, plenty of bouncing off the rev limiter in different gears, a cool-down lap, and driving back to the paddock. So, a little bit of everything and a lot of some things. A VEAL filter prevented use of records with a CLT below 175F.
Random thoughts:
1. This VEAL thing isn't as refined as it could be?
2. I have something wrong somewhere like a vacuum leak or something funny with my wideband?
3. I have some speed density settings that need adjustment somewhere or something?
4. I know just enough about this stuff to be dangerous?
5. I am a cotton headed ninny muggins?
Anyone have any input on this?
.
Last edited by Steve Dallas; 12-19-2017 at 12:01 PM.
Reason: Added tune and log
I have also experienced my car running rich in cold weather on a tune that had proper afr's at a warmer temperature. In for any insight people have.
This isn't the first time I have noticed the same "phenomenon" with the MS. I started tuning it in September and would spend a couple hours driving around with my laptop in mid 70s to low 80s weather. I would get it really close, then find it running rich the next [cooler] morning. Disabling MAT correction had some effect, but did not solve it.
I also noticed it would start to run lean, if I let it idle for a long time with the hood closed. Idle AFRs would climb from the upper 13s to the mid 16s. Once the intake tract cooled some, it would shift to being a little leaner than "normal". My early working theory was, the IAT sensor was heat soaking, due to its location right behind the radiator, and throwing everything off.
IAT differences may play a role, but IATs were similar during these 2 sessions of track tuning--within 10F, which is part of what makes this a head-scratcher. Something else is not constant. That has me looking at my wideband cross-eyed and wondering about other potential causes.
I looked over the tune again last night, and discovered Barometric Correction was still turned on. I have that off now, and will keep an eye on things to see if it makes any difference.
Still odd that VEAL wants to add fuel to much of the map by 5% or even more, when it had previously looked really close to being right.
^ I also have the same exact thing happen to me when I idle for a long time, I've always credited it to the return less fuel rail on our cars heat-soaking which is what many people on here have led me to believe. I wonder if there is just some setting that is set wrong when DIY ships these boxes, it wouldn't be the first thing. I spent an hour on the phone with them earlier this year trying to figure out how the fans were setup only to find they had the outputs for the two fans set wrong and labeled wrong in the manual, they revised it now. My barometric correction has been turned off.
^ The behavior is opposite of what is expected, therefore correction would have to be reversed. That is what is so perplexing.
In any case, if you tune in say, 70F weather, then drive a few days later in 60F weather, that 10F should not be enough to throw off the VE table by 5% or more. The difference should be very slight.
I went to the track again Sunday afternoon, this time with correct lev rimiter settings, to record some logs above 7,000 RPM. Ambient temp was 10F cooler than the previous outing, so it was close enough, methought. During the drive to the track, I noticed the engine was running a little rich everywhere, but especially in the cruising areas of the map, and thought that was odd, considering it was colder, and MAT correction is currently disabled. It should have been a little lean, if anything.
Anyway, I tracked it hard for 2 hours, enjoyed no spectacular spins, and frustrated my friend by refusing to be bested by his S2000.
Here is what's strange. I expected the algorithm to lean out everything above 7,000 RPM and lean out the cruising cells. What I did not expect is to see it go substantially richer everywhere else. I expected to see some minor changes, but nothing like this. I kept an eye on my AFRs on the track and never saw it go lean. Weird.
It is also worth noting that the areas it made richer were made leaner last time around, as I had made the whole map rich for safety.
This chart shows the VE table deltas between the map I drove on and the map created by MLV from 30 minutes of logging. The log includes driving to grid, a warmup lap, 13 hot laps with and without traffic, plenty of bouncing off the rev limiter in different gears, a cool-down lap, and driving back to the paddock. So, a little bit of everything and a lot of some things. A VEAL filter prevented use of records with a CLT below 175F.
Random thoughts:
1. This VEAL thing isn't as refined as it could be?
2. I have something wrong somewhere like a vacuum leak or something funny with my wideband?
3. I have some speed density settings that need adjustment somewhere or something?
4. I know just enough about this stuff to be dangerous?
5. I am a cotton headed ninny muggins?
Anyone have any input on this?
.
1) You have no EGO on, so that is not the issue. It would be OK if it were on, but then you could be reading perfect AFR's due to the EGO compensation, even though VEAL would want to adjust the EGO out of the VE table.
2) The real issue is that you are including all kinds of transients in your tuning run. When running VEAL, it is imperative that you drive like your grandmother. Very smooth with slow throttle changes. The transients cause large fluctuations in AFR.
3) The transients also mess up VEAL due to the time delay between fueling changes and when the O2 sensor sees and responds to those changes. Thus, even with a well-tuned EAE, there will be transient AFR_Error that will show in VEAL.
Strange behavior aside, what I think I am reading here is, contrary to many other resources I have consulted on tuning, the idea is to use VEAL to get it in the ballpark, then just rely on EGO correction to do the rest? My approach had been to get it really close and use EGO correction to make tiny adjustments.
Being a perfectionist software engineer, I want things to be predictable and repeatable. Perhaps I should adopt a different outlook on this one?
Also, I did not realize VEAL took EGO correction into account, but I guess I should have, considering it wants to know which field EGO correction is in the log. I was under the impression all corrections should be turned off. Constantly learning...
I am saying that I rarely use veal. I try to get my VE table nearly perfect, then I let EGO fix heat soak in daily driving conditions, like sitting in traffic.
Define "Ballpark". You would want the correction to be no more than 3-4%, and definitely zero on wot. From my experience, ego can't respond that quickly to big fluctuations in target.
I have 15% correction at idle. 4% up through cruise, but rarely is that needed, then tapering to 0% above 150kPa (Per memory). The 15% at idle gets fully utilized. -12 in cold weather, +15 after hot restart in the summertime. Yellow top injectors.
Tuning of the VE table was done initially with VEAL, until it started adding to some cells, and then subsequently subtracting. For higher loads, hills come in handy.
The next step is to get the MAT corrections correct. That also minimizes EGO.
Basically what Son said, but I used VEAL a lot initially.
Define "Ballpark". You would want the correction to be no more than 3-4%, and definitely zero on wot. From my experience, ego can't respond that quickly to big fluctuations in target.
Without thinking too hard about it, and without being convinced of the approach, I was supposing 5% to be in the ballpark.
Originally Posted by DNMakinson
I have 15% correction at idle. 4% up through cruise, but rarely is that needed, then tapering to 0% above 150kPa (Per memory). The 15% at idle gets fully utilized. -12 in cold weather, +15 after hot restart in the summertime. Yellow top injectors.
Tuning of the VE table was done initially with VEAL, until it started adding to some cells, and then subsequently subtracting. For higher loads, hills come in handy.
The next step is to get the MAT corrections correct. That also minimizes EGO.
Basically what Son said, but I used VEAL a lot initially.
My initial approach to tuning had been to use VEAL with my laptop in the car. I would drive around my neighborhood very carefully for a while, then move out to the side streets with higher speed limits in the general area, then move to the highway, with its hills and overpasses to help me hit different load regions. Using cruise control is a good method for cheating at smoothness. I would do that, until VEAL stopped making changes or made very minor changes. That seemed to work, except that doing the same thing the next day in similar conditions would result in significant changes to the VE table, which I find strange. I think it should be repeatable, but maybe I'm a maroon.
BUT, hitting the areas of the map I typically use on the track, which mostly involves flooring it, since a N/A Miata's throttle is more of an on-off switch, proved to be difficult to do on the street for any meaningful duration. So, I decided to try using the VEAL feature in MLV to analyze some log files from the track to tune for what I actually do out there. This worked pretty well the first time, but the results from the 2nd time were surprisingly different, as noted above.
Here is the VE table after applying that last set of changes. Interestingly, most of the areas of the map that are easy to hit are pretty smooth, without having applied any smoothing. The upper left and lower right corners need some work, but they seem nearly impossible to hit without the luxury of a dyno.
I'm going to burn this, turn on EGO correction, and play with MAT correction and see how things look. Unfortunately, a cold front just came through, and it might be weeks before I see good weather for comparison.
.
Last edited by Steve Dallas; 12-23-2017 at 09:27 AM.
Reason: I am, in fact, a maroon.
Had an unexpectedly nice afternoon here yesterday, with ambient temps in the low 50s, so took the Miata for a spin with EGO correction turned on. My AFR gauge now looks like a schizophrenic monkey on speed. Enabling EGO correction at idle causes RPM to hunt any time it is closed loop with a variety of settings, which it did not do before, but it is not as badly affected by MAT heat soak as it was before.
At least the VE table I posted above seems to be pretty decent, at least in yesterday's weather.
I have attached my current tune file and a log of the car oscillating at idle in my garage, should anyone want to have a look.
Now considering moving my wideband controller ground. It is currently grounded to the body in the footwell. I have read to ground it to the ECU ground, which is probably good advice I obviously ignored. The ground wire on my controller appears to be 18 gauge. The ECU ground wire appears to be 22 gauge. Splicing the controller ground into the ECU ground looks like it would overload the ECU ground wire and potentially cause all sorts of problems, which is why I didn't do it. The ECU ground wire appears to terminate at the front of the head. I suppose I could snake a wire through the engine bay to that point and ground it there (after testing to make sure the ECU ground really does terminate there, of course).
.
Last edited by Steve Dallas; 12-24-2017 at 12:59 PM.
12-18-2017, 03:47 PM
Now with Tune and Log!
0
