saabin

So I admit it, I’m a geek for data. As an engineer by education, I no longer do technical work in my day to day work.. So this allows me to satisfy that craving

One of the things that I liked about the UMW tune and logger is the data provided and the full explanation (if one cares to know) behind the data. I’ve learned a ton from analyzing the datalogs and talking to Kevin.


I’ve been using datazap.me to graph data along the x axis for the different parameter.. useful for looking at load, timing, boost, IATs, etc..
Heres an example of injector duty cycle..




I wanted to take it a bit further and started playing with a new graphing tool – MegaLogViewer HD.. I forget what I paid for it, I think about $30..

All these graphs were from datalogs generated from multiple 3rd gear pulls from ~50-100+ in similar conditions.


In addition to the normal data graphing that datazap.me can do, it also can create histograms..

One of the things I learned from Kevin is that the ECU uses a torque model; engine load is the primary element.

Here’s one, showing engine speed on the X axis, Engine Load on the Y axis. The load values are plotted on the table; the darker the shading the more “hits” of that data value. I have filtering turned on so I only see data at WOT (Throttle >90) and RPM in a range I want to see,,




Here’s one I’ve been looking at that shows ignition timing vs rpm & load.. When Kevin sends me a new tune I can compare to see what timing has changed vs the other parameters. See how the timing increases as RPMs increase?





Now look at the timing of my old OTS FVD tune.. looks much different, with some random values scattered about, not smooth and linear like Kevin’s. Yikes!




As you can imagine, there’s a wealth of info to be gleaned.. How about boost vs load?





The tool supports a Z-axis delta, which means it will subtract two values for you and plot them on the “Z” axis.. What would that be useful for, you ask?

Well here’s a log of lambda deltas (Lam1 – Lam2) which is an indicator of O2 sensor responsiveness.. The closer to zero, the better.

This log was taken before I did my wideband O2 sensor change.. check out the rather random dispersion of values – as RPM picks up, I get some decent sized deltas.. again, this is filtered to only look at WOT. There were some variations while load was building.





Now here’s a log I just did a few days ago.. notice how the deltas are all bunched together nice and tight?

This was from a 2007 with 44K miles and no CEL.. No wonder Kevin tells people to change their O2 sensors..







Anyway, fun with data!

bondjockey

Great info, thanks for sharing. With a new UMW tune installed two months ago and new plugs, coils, exhaust and O2 sensors installed last week I am most definitely subscribed to this thread. Lots to learn...

saabin

Ok, have you ever wondered how higher octane fuel effects the engine? Lets look at the data..

My car was almost empty so I put in 10 gallons of 100 octane unleaded.. Normally I run 93. Ugh this is an expensive test..

Same method for both runs, 3rd gear pulls from 50ish to 100. Plots only look at data at WOT.

I have 3 pairs of plots, KNOCK, timing, and IATs. KNOCK is just a simple SUM of ALL 6 cylinder timing retards at a given instant. 93 octane first, 100 octane 2nd..

Notice how the 100 octane plot has much lower KNOCK values across the rev range..

93 KNOCK


100 KNOCK



Next is the timing, how does this affect timing? Well, because of the lower KNOCK, we have a slight increase in timing across the board..

93 Timing


100 Timing


So this translates to more power, correct? Well, not tonight as I noticed my LOAD was not quite as high with the 100 octane runs.. turns out I had slightly higher IATs this evening (70F vs. 60F OAT) that kept the load down just slightly.. My vbox verified there was no discernible change in my 60-100 times.. I need to repeat the test again when the weather is cooler...

93 IAT


100 IAT



I know a lot of people are not interested in this but hopefully some are

CincyScott

As a part of the NON engineer crowd, what constitutes a cell going green vs yellow or white? Optimal value?

saabin

The differnt shadings/ colors represent the frequency (or number of occurances) of a given value.

Each log contained multiple pulls at WOT, and each pull would have data on the plot. If I plotted just a single pull, all the cells would be white or at a low frequency. Each pull would then add to the "counts" for each cell, and it's shading would become darker..

white - less occurances (single event(s))
yellow - more occurances
lighter green - even more occurances
darker green - most occurances

Hope that makes sense..

DocHW

Very interesting, I've been logging for a short time and probably understand less than half of it, but I thought IAT above 45-50 C started to hurt power?

Knock retard translates to about 5 hp loss per degree right?
With a value of 12 in your graph that would be 60 hp, although very momentarily. In my logs I only see knock retards for 0.2 - 0.5 sec with a peak of (sum of all 6 cyl) -12 degrees like you for perhaps 0.1 sec. You generally would want to get rid of knock retard for engine safety?

TB993tt

unfortunately logs of 50-100mph in 3rd gear don't tell you very much about the tune of your engine, particularly turbo engines which get nice and hot in higher gears.

Kevin

The descriptions for KNOCK should READ, Retarded Timing for Knock Control (Knock Retard) The engine is adjusting Timing in each cylinder to eliminate detonation. Worded another way, the engine has removed "X" timing to run safely. This platform can remove up to -15 of knock retard.

For reference, a STOCK turbo running 91 octane will have -7 to - 10 degrees of knock retard at 75 degrees outdoor temps.

CincyScott

Yes, but fortunately, it helps us from going to prison to test a little

saabin

Based on what I'm seeing, the higher the IAT the more effect it will have. Maybe once it gets to 45C it really starts to dial it back, but from what I've seen even going from 18C to 28C IAT will effect timing retard, albeit slightly.

This KNOCK retard chart is identical to the one posted before but with IATs in the 18-22 range (as opposed to 27-30)




Heh, yes that is true.. I'm not looking to build a 1 mile airstrip car or anything crazy like that
I'm still pretty new to this so I'm still learning.. I've logged some 60-100 times (3.29 vboxed) and now I will probably work my way toward 60-130.
Ya gotta walk before you can run

Bskyoon

Apologize for the long post but I feel there's some very important information here I'm glad I was able to come across prior to my tune...

I'm in the process of tuning with Kevin, the datalogger alone has proved invaluable. As Saabin mentioned there's a gold mine of information to be found and I'm just focusing on a couple things for now. Attached are two pdfs using "93" octane from my convenient down the street 711 and Shell V power inconveniently located a little further in town but according to Kevin a much better quality controlled fuel. So the take home message is 93 octane is not 93 octane just because it says so although that is actually the case when you're not nearing 100% throttle application (more on that in a second). The yellow highlights are the throttle input, the other 6 columns show retard in cylinders 1-6, respectively. As you can see the highest retard I got with Shell was -6 with a much lower average across the cylinders vs a fairly dangerous but still wnl -9 with the 711 93 octane due to the less efficient burning fuel. This is just a snap shot of the 3rd to 4th gear pulls but the numbers were fairly consistent across a couple other runs. I won't bore you with the majority of the other spreadsheets but basically when you're driving around town, even getting 80-90% throttle in the 3500-4000 rpm range the knock sensors can keep up and you really don't see any significant retard, if at all. This provided great insight into my previous googling on whether or not we should be paying that premium on 93 octane for my wife's GL550. Correct me if I'm wrong because I'm just extrapolating data I've gone over from the p car so if I'm completely off please feel free to correct me...
The two most important facts I've come across is that cars with compression ratios higher than 10:1 (most turbo cars) do require premium fuel and that most newer cars have knock sensors which negate this necessity. While these facts seem to contradict eachother they're actually both true. The knock sensors do a great job (maybe not as good in a family SUV or sedan but good enough job I suppose) controlling knock and preventing detonation, especially if you're in an RPM range that isn't too fast for these sensors. However, as mentioned above they can't control EVERYTHING, ALL the time in the higher RPM range. Now finally here's the all important application to all this...

Of course Porsche didn't design a car that would detonate just because you floored it and didn't use the recommended 93 octane or even 93 from a bad source. They designed a conservative (still very fast) tune with relatively low boost at ~0.8 bar max. HOWEVER, that all changes when you tune and increase it to (from what I've read) 1.2-1.3 bar. The RPMs get to that high range much faster, fuel is required to burn quicker, knock sensors may not be able to keep up and you may get way closer to the -15 retard than you'd really be comfortable with.

My new view on the premium on premium for our family car is 87 octane. Sure it's twin turbo but I can't remember the last time I floored it or the RPMs went above 3K. I'm not sure what my wife is doing but she just learned to use the paddle shifters a few months ago so I'm not too worried

Bskyoon

One thing I didn't mention and I will have to address is that the 4th column indicated intake air temperature and clearly there is a very significant difference between the two runs as I ran the 'bad' fuel last week during the day and did the Shell run last night in much cooler weather. I'm pretty sure that made a difference since the whole concept here is efficiently burning fuel so I will look into that later.

saabin

Thanks for the data, glad you are really digging into the numbers.. I presume those logs are from the stock tune?

I remember getting some high knock numbers when I was logging stock, too, although I don’t think I ever saw as high as -9.. We have pretty good/consistent fuel down here in TX.

Yes you will definitely see a difference with your IAT delta (24.8C vs 36C), so your knock differences could be attributed to BOTH IATs and the fuel quality.. I learned when I was logging to try to make sure the outside temps were as close as possible between runs so as to eliminate IAT from the equation when comparing things.

It was interesting to me to see the overall timing and timing knock, stock vs Kevin’s tune.. I have access to MS109 here 20 miles away and even with Kevin’s 93octane file it makes a difference with the timing retard (going from very little to none, for example)

Bskyoon

Yes those were stock tune with 'bad' gas and 'ok' gas. I just uploaded the tune yesterday, I can see the boost gauge flying up much faster and higher than before but also feel the pull. It's much smoother than I thought, I was expecting a kick in the butt but I can't complain about the driveability. Haven't had a chance to run logs yet...

Bskyoon

I finally got around to running this log, here's a snapshot of a hard pull in 4th near the top end. If nothing else the boost numbers will show that 33% increase in forced induction which can alone testify for increased power. The engine load also significantly increased and the IAT is fairly consistent in both the stock tune and this run this time. I definitely can appreciate the timing retard actually improving from nearly -8 to -9 to just a maximum of -6.

Butt dyno results:
I have 330 rears and 305 fronts and easily break traction near the top of 1st gear now.
No clutch slippage yet thankfully* (only 16K miles on the car).
Gas mileage going down, useless trips going up.
Hopefully this will keep me happy for longer than a few months, I never wanted a 900hp car but I guess all those journeys began somewhere...