Sunday, June 25, 2006

Boost Perspective

After looking at some tunes and logs of turbo systems, I realized that people don't adjust their NA tune to the new reality of forced induction. I mean, they know to adjust their target AFR and go easy on timing, but what I'm talking about is not knowing _when_ to switch over to the boosted settings.

In a normal NA setup you usually trigger PE with >60%TPS@ and 15kPa MAP, which is very tame, but sensible.



Boost doesn't happen purely as a function of throttle input and rpm, it is really a side effect of increased load and airflow, which are results of throttle and rpm. With these few indirections involved, you should not rely on just TPS/RPM triggers.

First you must figure out WHEN does the boost hit.
In a first attempt, I just looked at the common LTFT scan with all data loaded, and it looks like MAP, which is the closest thing we got to a boost measurement in this log, maxes out very early.



I wanted to see in context what happens before, during and after MAP maxing out, thus I just played the log in the charts window, concentrating on TPS, RPM, MAP, DynAir, and DynCylAir.



In the chart above we see that even at 84% TPS and 2700RPM we've already maxed out the MAP! this is great news as it means a quick and easy spoolup of the turbo. It also means that whatever data we'll get from this log qualifies as ballparkish at best, as we will not know the true MAP reading, which is taken into consideration for each Airflow estimate. (this log is done in Speed Density)

Another boost situation, and a similar result:



Very similar numbers (74%TPS and 2650RPM), yet we're already getting 0.80g/cyl of compression, which is a great PEAK result for most cammed NA cars, while we're getting it partial throttle and <2700rpm!!!



Airflow readings hit a flatline at about 58lb/min and stay there until redline. What the real readings are, we won't know until everything is calibrated correctly with a 2barSD setup.
The amazing part about this graph is the torque/compression: 1.24g/cyl!!! I have not seen a NA stock cube motor do that ever. Another great part about it, is that it peaks late (5300rpm) and doesn't drop off much at the redline, suggesting a fantastic powerband.
Again, due to the fact we're looking at maxed out readings, this is all pure ballpark, no specific numbers.

So now that we know that we the car is healthy, the boost hits early, hits hard (for a basic non-intercooled 5psi kit) and carries way more power that we can currently harness all the way to 6200rpm.

Remember at the beginning of this article we were talking about triggers? They were optimized for little torque (0.80g/cyl is a nice number for a small cam, 0.90 for bigger cam, and 1.0+ for a nice head/cam setup), and hitting that torque late on the powerband (LS6 intake usually forces it to be between 4400-4800RPM).
Now, we have a whole different paradigm on our hands. This setup maxes out at 1.26g/cyl at 5300rpm, with going into boost at a very early 2650RPM already pushing 0.80g/cyl.
Let's try to pinpoint where the triggers should be:




This histogram shows how much compression/torque/air in cylinder we get at different RPM intervals. Looks like somewhere around 2100rpm we can already hit 0.80g/cyl (note this histogram shows max values) which normally is amount of torque we'd be prepared (as in triggered PE mode) at 4400-4800RPM.

Based on this information, we set Delay RPM to 2100RPM.

Now let's look at how much throttle input does it take to cause a full 1 bar absolute pressure (101.3kPa) which is 1bar MAP's max value:



Somewhere at mid 30s %TPS we can already achieve boost, thus:
Set PE Enable MAP at 35%

Now the only table left is the PE Enable TPS table. We already know that we need only 35% to make things dangerous, now we need to tell the PCM how soon on the powerband that can occur, thus we graph MAP vs RPM:



We see at 2000RPM we already generate 90kPa MAP pressure, thus:
Set the PE Enable TPS table to 35% starting at 2000rpm, and for safety's sake, let's pick 20% all the way at the redline, and blend between them for a sensible curve.

Hopefully this demonstrates how you have to watch out for the new conditions created by your new modifications. Also, I posted a lot of configurations of custom histograms, which while not precise with the data gathered, it provides plenty of trending information to make it a worthwhile exploration.
Another thing I know I haven't explained yet, is my liberal use of horsepower/airflow/DynAir and compression/torque/DynCylAir. It's another writeup in the making, but for now you just have to take it for granted, so go ahead, push the 'I Believe' button and enjoy your new safer ride.

If you have any boost logs taken with a well tuned, well set up (2/3bar MAP sensor, 2/3bar SD mode/CustomOS) please send it to me, as I'd love to have a followup article to this one.
If you use the information in this article and you notice that things work better/worse, please let me know, I'd like to know how it works out in a day-to-day operation, as I don't have access to the Camaro with the STS setup anymore.

Thursday, June 22, 2006

Roadracing logs analysis

So Dennis went racing to Summit Point and brought me some logs. For the underinformed crowd, this is a full fledged, although home made Camaro race car. Less than 2900lbs with fuel and driver, probably upwards of 420rwhp, huge wheels with roadracing slicks, aero, the works, you get the idea. He said that last time he was out, only one car was able to pass him, out of 50 or so at the track. NICE!

We've been tuning this thing for a while using EFILive and the blackbox logging capability and it's been coming along nicely. Dennis is happy with the power because he has no traction in 2nd gear if he floors it, even with the big slicks on. But on the roadcourse, 2nd is rarely used, so all this power should come in very useful.

So this is my newbie attempt at hacking the data into somewhat readable and hopefully educational graphs.

First I wanted to see if the car is healthy, and if it does what it supposed to do, so I checked injectors: pulse width never went past 12 and duty cycle never went past 60%, so I'm going to skip these boring graphs.
Next is Air Fuel Ratio when going partial throttle and full blast.



It looks reasonable, car goes into PE mode as MAP/RPM values increase, as these are the PE triggers.

What made me a bit curious though is why was the spread of AFR at WOT so large:



I thought to myself, maybe I'm looking at the wrong thing, throttle is just the input, intake manifold pressure (MAP) is the effective dictator of what amount of air is going to get pushed in, as well as it's used in Speed Density airflow calculations. So I graphed it another way to see if there is a difference:



I dont see any frankly. we still get 0.7 to 1.0 point spread in AFR final values. I started to wonder what could be the culprit behind it. From experience, I know that a lot of information is to be extracted from plotting raw MAF signals and look at the noise levels, as an indicator of trouble. Here's what I came up with:



Damn it Dennis, clean your air filter! :)
The areas circled are noise, the signal has a very strong signature and range, but this is more than usual amount of noise. So if you ever see anything like this, go shake off dust, chunks of rubber from tires of cars in front of you, dead birds, whatever you got in there. Here we're gonna let him get away, as this log was the last one of all the sessions so stuff probably accumulated over the course of 2 days.

I did notice one more thing: maximum pressure in the intake dropped as the RPMs increased. Does this mean that we cannot feed air into the engine fast enough? Is this a sign of a bottleneck? Is it the intake manifold? Or is a just the overlap of the cam letting some air in and out fast enough that it shows up as never quite reaching the peak pressure values? Take a look:




Ok, so with all the checkups of the stuff that might make the car blow up (there's another boring graph or timing but that's really dead on the money), we started talking about more unscientific, but definitely tangible aspects of tuning. This track has a long straight, on which you could really scream through 4th (3.42 gear M6) up to 6500rpm, or try to shift a bit early to a not so dramatic 4400 rpm in 5th. But apparently, at 4400rpm in 5th, combination of greatly increased air resistance and falling off the powerband(cammed LS1 are the VTEC of pushrod V8's, they're very rev-dependent) leaves so little horsepower left that the acceleration is dismal. So of course, another debate of longer gears and 3-4 or shorter gears and using 3-4-5 at the track. This dispute is to be solved numerically, as quickly as I get done with chores and errands and finish rewriting my shiftpoint/gearing optimizing spreadsheet in C++ with increased precision, added air resistance and more gears.
For now we just get to look at what it looks like in terms of air consumed at various speeds:



You can clearly see that after shifting to 5th, you only get 30lb/min of airflow, out of ~45 possible at higher revs. while normally this would be only a slight annoyance, at greater speeds it becomes absolutely detrimental.
Let's do a simple example: if at greater speed it takes you 150hp to just fight off the air resistance, and you have 450hp max, then you have 300hp left for actual acceleration. If you shift early and you're left making 300hp and still using 150hp to just cut through air, then you have 150hp left for acceleration. Combine that with taller gearing of 5th gear versus 4th, and you are effectively stuck with 1/3 of accelration you'd get in 4th!

And here's the biggest discovery of this datalog: I was watching Formula 1 the British Grand Prix from Silverstone, where with smaller engines these guys spend over 70% of the track going WOT!!! I never thought of measuring that, but now that I have the data, why not, right?
But of course, I tried to run this idea through some people, and they brought up a good point: how do you define the proportions? is it measured as:

  • "wot acceleration"/"total number of samples"

  • "wot acceleration"/"acceleration but not wot"

  • "wot acceleration"/"number of samples in which car was not-braking"



I went with simple version and went with the first choice, as it provided me with what I think describes the issue best: how often around the track do we get to actually utilze the car's power to full potential?

I had few logs, and the results varied from about 19% to 22%.

Now this is LOW! And this is a very well suspensioned, well driven, light car on a lot of rubber! I expected a lot more, afterall this is a track with a pretty hefty straight (150mph+ max).

Some time ago on www.frrax.com (<--- awesome site, btw, if you haven't been there yet, go visit!) I did a quick calculations of how much does having extra 10hp on a 280average hp car help in actually decreasing times, and It came back with something suprisingly, yet laughably small, like 0.1 sec if we get to use that 10hp for 10 secs! So not only we dont get to use the extra power very often, but even when we do, it doesn't have that much influence.

So does adding power to amateur roadrace cars mostly increase chances of blowing up, and does next to nothing when it comes to actually reducing lap times?