Tom2769 wrote:
I'm trying to dial in my MAF transfer on an 80mm LMAF. Right now I'm using a "known" MAF transfer that I pulled from this site, but I'm using a custom intake tract (3.5" pipe) so I assume the known curve is not dead on for my setup...
It may, it may not. But it should be close to the curve you have assuming the curve you have really is the curve for your MAF. But a 5% variance in a MAF value is well within acceptable range to expect to need to tweak a point on the MAF curve by. But if it were me, before I went messing with the MAF curve, I'd test to make sure my Injector Slope/BP values looked good in EA. If they are off, this will throw your MAF curve as well. Some people just throw numbers into the Injector parameters and call it good. I tend to prefer to make sure the parameters are good as any variance in those parameters relative to their actual behavior throws off any other calculation related to fuel...and being injector parameters affect fueling in a non-intuitive way (variance is over PW, not over RPM or Load directly). Verify your Injector Parameters and you may find your MAF curve a little more accurate than what you are seeing now.
Tom2769 wrote:
...I recently ran a datalog with forced OL, reset KAMS and learning disabled. I tried using EA's MAF tool to evaluate my datalog, but was getting some funky results (likely operator error),...
Your funky results were probably due to not having enough info for EA to make a determination about every point on the curve. Just as your XLS file indicates areas of "no data", EA will simply not update datapoints on the MAF curve in those same areas leaving you with a MAF curve that can look awfully jagged. As the person running the software, you have to be aware of this and not take whatever EA spits out verbatim, and instead apply those changes to your MAF curve with some thinking and eyeballing. Ideally, you want a diverse enough log OR a number of different logs concatenated together to cover all possible datapoints so EA doesn't ever have a point on the curve with "no data". If you ever do load multiple logs into EA, they must all be logs produced from the same tune. Changes in tune will affect datalogs...as you would expect. Well you don't want an old tune's data to get in the mix and taint the data that is intended to be the results of your latest tune. EA could have you making adjustments that are NOT correct for your latest tune.
Tom2769 wrote:
...so I resorted to just tweaking an excel sheet to calculate average AFR (from WBO2) for equal ranges around each voltage point on my known MAF transfer. I initially just charted the avg AFR (in lambda) as reported by the WBO2 and calculated the % that it was off from lambda 1.0 and then converted that into a % of "error" at each voltage point for which I had data. I calculated that I was between 5 and 10% rich at lower voltages (and not a consistent or smooth amount), and then jumped to 15% rich at voltages over 2.88. Initially I did not understand why I went so rich at high voltages/high loads - WOT.
Then I did more reading on here about the base fuel table and understand that at higher loads we need to go rich (~.86)...
While in Open Loop, the LAMBSEs represent the target AFR the EEC expects to get based on its calculations. Compare those to the Wideband's feedback. The difference between commanded and actual indicates the error. The error is what the spreadsheet and EA should be using to calculate a change in your tune. But just to be clear though, there is another function in GUFx (FN320A) that does additional enrichment on top of the fuel tables. To make tuning easier, I highly recommend you null this function out by entering all 1s in there. Otherwise the LAMBSEs you type into your table and the LAMBSEs you see while in those same RPM/Loads will not be the same.
Tom2769 wrote:
... For my datalog I had used a stock fuel table, but based on research here, I think I will be changing my fuel table to 1.00 at loads of 55 and under and 0.86 at loads over 55 with WOT enrichment at 1. If I understand the fuel table correctly, this means my LAMBSEs in OL are exactly that: 1.00 at low loads and .86 at high loads, right?...
For forced OL, this is what I would expect. Now are you using FN1306/FN1307 or do you have the FN1360 patch applied to your tune? If you don't have the patch, that's OK. I just want to make clear you don't want to set the entire FN1307 table this way as that's not RPM/Load based. That's RPM/ECT based. If this doesn't make sense to you, go search within BE for FN1306, FN1307, and FN1360 and read their comments for more details about them. For additional info, read relevant threads here containing discussions about them.
Tom2769 wrote:
...So based on that, I selected all my datalog WOT data rows (these all had loads over 55, BTW) and recalculated how far they were off from .86 lambda (instead of 1.00). I came up with between about 2 and 7% rich. So far so good? Am I thinking about this correctly?...
Based on what I'm hearing so far, I think you got it. But just to be clear, when you came up with your 2-7% rich, your LAMBSEs were 2-7% higher in value than your average WB for those same flow voltages? If so, then you got the idea. Although doing this through EA, once you figure out what EA is doing is much easier than what you are doing with the spreadsheet. Although if it took you doing the spreadsheet to get your head wrapped around what's going on, then the spreadsheet exercise was worth the effort.
Tom2769 wrote:
...Assuming this is all correct so far, I'm between 2-7% rich at WOT and 5-10% rich at part throttle. Once I'm done tuning, I'll allow CL, so I'm less concerned with part throttle AFRs at the moment, correct? I'm inclined to lean out my MAF curve by about 4-5% across the board and move on to injector settings, etc. Thoughts?...
As I said above, I would stop and do Injector Slope/BP/Offset analysis right now instead. An across-the-board rich tends to tell me you have slope/bp issues. My bet is if you load your logs into EA, find some better slope/bp values, then rerun all your MAF analysis, your MAF will be much closer. Even if it isn't, it shouldn't be "across the board" rich. Any time you have an across-the-board rich or lean condition, I immediately think there's something off with injector parameters. Even if the MAF curve is dead-on, the across-the-board problem can be reduced via the injector parameters thus minimizing the amount of adjustment required when you do get back to MAF adjustments. Again, this is my way of doing things. There are others here that just put whatever value they trust into the Injector parameters and tweak everything else in the MAF curve and let the KAMRFs work out the kinks and error at drive-time. There's no single right way to do things as long as the engine is getting the amount of fuel and spark it wants when it needs it...this method is just mine.
Tom2769 wrote:
...DO I need to be concerned that my OL AFRS are a bit erratic over teh voltage range at this point?...
What are you calling erratic? My AFRs jump around some but they average in a range that I'm targeting. The jumping around is just differences in combustion from cylinder to cylinder and even differences in exhaust mix within the same cylinder. We all like to think that the entire cylinder of air & fuel burns exactly the same. But it doesn't for some engines. Depending on how well your injectors atomize the fuel into the air, how much raw fuel sticks to the walls& piston top, and how much swirl & tumble there is going on as the mix is being aspirated in and compressed will determine just how homogenous the burn will be. Point is, if you could take ultra-high accuracy and ultra-high speed samples of the exhaust from each cylinder, you'd likely find consistent differences from the exhaust that leaves early on vs the exhaust that leaves in the middle vs the exhaust that gets out just before the exhaust stops flowing...and even those samples would be mixed averages of air/fuel that burned in the center of the cylinder vs air/fuel that burned at the edges. The best you can do is go with rough averages based on all gasses coming from all cylinders in the bank you are sampling from. The average will be close enough for you to tune by even if it does give you some erratic "jitter" when you watch the WB. But the jitter shouldn't be wild swings that vary multiple points. For instance when you go WOT, if you are targeting .85 lambda, you should expect logs that jitter with occasional flashes as wide as .82-.89 range but with the vast majority of the samples much closer to target and ideally averaging very near the LAMBSE the EEC is targeting. Well built engines with consistent injectors and a steady MAF can get you readings that won't vary more than +/-.01 lambdas from target.
_________________
...Always Somethin'
89 Ranger Supercab, 331 w/GT40p heads, ported Explorer lower, Crane Powermax 2020 cam, 1.6RRs, FMS Explorer (GT40p) headers, Slot Style MAF, aftermarket T5 'Z-Spec', 8.8" rear w/3.27s, Powertrax Locker, Innovate LC-1, GUFB, Moates QuarterHorse tuned using BE&EA
Member
V8-Ranger.com