Wednesday, May 25, 2005

visualization of gearing/shiftpoints

NEW DOWNLOAD, version 2.3, now with graphing! :)

So while my spreadsheet is cool, not enough people have precise numbers and dynos in different gears and such to really take full advantage of its capabilities.

Thus, I'm gonna combat the stupid myth of 'if you peak at 6k then just shift there' that you hear so often on LS1tech on a realistic example.

Pictures are worth a thousand words, so I grabbed some data from Trackbird's stockish 2002 camaro, as he's the only guy I know that has dynos in different gears. Oh, it's good to be a crewchief ;) I used 3rd gear data for 1st and 2nd, as 3rd and 4th were the only data I had.

I graphed 4 curves, picturing the relationship between horsepower at different speeds, in different gears. Some of them overlap of course, as gears do.
The blue dots represent the horsepower curves for the entire range in each gear. Pink dots are the HP which would be used to accelerate, with given parameters (shift points, gearing, tire size).

Now for the cool part:
T56 is a good trans, well spaced for LS1 powerband. 4L60E however is a different story. I used my old setup (3.23 final gear, 4L60E trans, 315-35/17 tires) with Trackbird's dyno graphs (since it's stockish enough) for the purpose of this demonstration.

Look at the 1-2 shift: 6000rpms drops to 3100rpm, which brings it down from 290hp to only about 180hp. That's sad, becuase it kills the whole myth of big american V8's with a lot of down low grunt. Sorry guys, but it's just not there anymore. 180hp while your car weighs 3500lbs will yield a lousy acceleration.

2-3 shift is not that much better. This time we're going from 6000rpm to 3700rpm which yields us a drop of about 70hp. So this time not only you end up with lousy 220hp, but you're already in 3rd gear, which is 1:1 (3.23 effective) ratio.

Now let's look at the modded scenario (3.73s, 6400/6300/6200rpm shifts):

1-2: 6400rpm drops to 3400rpm 'landing' on lousy 195hp again. It's better than before, but still lousy. The good part is that for that few miles an hour, it wasn't going from 180 to 195hp, it was dropping from 280hp to 270hp. That's the trick about shifting past the peak power. You might be losing power, but you're not losing it nowwhere near (almost 100hp difference!) as badly as if you were to shift to a higher gear earlier. So the gain is twofold, you don't lose lower gear, and you gain on the higher gear--nice!
2-3: 6300rpm drops to 3900rpm 'landing' on much more pleasant 225hp. The logic is the same as with 1-2 shift: you don't lose in 2nd anywhere near as much as you'd lose by shifting to 3rd earlier. Then as a added bonus, you start off in 3rd on more power.

So how can we quantify this?

I set up a scenario where we'd use 1-2-3-4 gears on both auto and manual transmissions. 30-140mph run, should demonstrate how higher shiftpoints combined with shorter final gear, yield a better acceleration.

I added up all the hp cells 'participating' in this simulation and added them up. Also, I counted how many of them there were, giving us all we need to see the average horsepower rating throughout the whole accelerating process.

case 1 (6000rpm shift, 3.23s): 263.89HP
case 2 (6400/6300/6200, 3.73s): 270.65HP
That's a 2.5% increase, but more importantly it's throughout the whole range! Not only that, but it's also rearwheel horsepower rating, this is exactly what causes acceleration (well, if we ignore the air resistance, but it's not like you can cheat out of this one). So practically speaking, it's power for free, but in that effective sort of way, not some theoretical instantenous peak rating that marketing guys love so much.

Also, this is on a stockish aplication. In cases where power does not drop off 'till much higher rpms, proper shifting will reward you even more.

If you're stuck with the A4 transmission, you can learn where it hurts the most, so you can spray there. Right after the shift in 2nd would be NICE, but remember that a to make about 400hp (let's say ~200shot) at 3100rpm means you're making 677lb/ft of torque, so while it will help tremendously, it might not be the best thing to do as far as reliability is concerned.

I hope that demonstrates a bunch of points:
1. Various applications: where to spray for drag racers/where to shift for road racers/what speed would be the best to start off with for a highway race
2. Common sense sucks as far as numerical integration is concerned. Some things you just gotta calculate for yourself, you can't just rely on myths and popular opinions to.
3. Peaky powerbands (yes, that's you, LS1 people!) do need properly matched transmissions to go alongside with them. There are reasons why little Hondas with peaky VTEC powerbands rev beyond 8000rpm and come with 4.3+ final gears. While they don't have the sheer stupid power, they always stay on it, and in combination with lower weight (which ultimately is the other part of the acceleration equation) they yield similar results as far as acceleration is concerned.

So in conclusion: don't just concern yourself with making power. Learn to configure your setup so you actually get to use it.

Thursday, May 19, 2005

gears, powerbands and shift points

Long time ago, I read THIS.
It really made me think, especially that Andi is a _really_ smart dude.
For the longest time I've been pondering how do I write a program that would take gearing, tires size, shiftpoints and come up with the most optimal configuration. In video games (like the Gran Turismo series) you always end up with a 'magic' trans with which you can set up just about any combo you'd like. I spent hours (yes, I know, it's sad, shut up) tweaking it so the top speed on the track would match the redline in last gear available. Then I discovered that depending on the track first gear would have to be long enough that it wouldn't be useless (aka permanent spinning) or short enough that I could just ignore it and don't use it for anything but standing starts.
In real world, we have slightly different problems. We can't change internal gearing of the trans (well, we can, but that's pretty much left for real racing teams and people who really don't have a budget), we can only pick from a limited selection of final gears, and we also we have to watch out for things like not using overdrive for racing situations as it's too stressful. (Doesn't Porsche 911 turbo have like 3 overdrive gears though? Well, I guess that trans can take it, you get what you pay for). So in my case, I got the GM 4speed automatic with lovely 3.06,1.62,1.0,0.7 gearing. These are HUGE drops in between gears. With stock gearing it would shift 6000->3200, then 6000->3700 which totally made it fall off the happy range. First thing was to bump up the rev limiter to 6600rpm and set up shifts for 6200rpm, which depending on temperature and how much I've been beating on it would do it between 6200-6400rpm.
that was slightly nicer, after a shift I'd end up 200-300rpm higher than before. That's not enough however, so I had to go out and get a 3.73 gear :) With that, and the raised shiftpoints, it would shift 6300->3500 (1-2) and 6300->4000 (2-3). That made it much snappier, 2nd gear would shift out at 72, not at 90, 3rd wouldn't go past 140, it would hit 6500rpm at 125mph. It's a much more track oriented setup now, makes for a good daily driver too.

But of course, me being me, I didn't wanna go by seat of pants, I wanted some numerical backup of what it does and by how much.
So all these things combined, lead me to create a spreadsheet that takes into consideration different power curves, gearing, tire size, and shiftpoints and then counts exactly how much work was done in each of the intervals. The problem was to figure out that I had multiple gears with multiple power figures for each rpm and each gear, and I also wanted separate shiftpoints, as you saw at Andi's writeup, it can vary quite a bit.

So here it is: DOWNLOAD

Put in your data and take a look at the average hp rating. Start playing with it. Put in shorter gears, up your shiftpoints, watch the numbers change. I'm gonna post some useful scenarios for which this spreadsheet is useful later.

If you wanna know what's behind the scenes, I've basically taken the HP numbers relative to the gear they're in (yes, you can have separate maps for different gears, all you turbo people should be happier!) and added them up as long as fall fall within rpm range (bigger than the 'landing point' from the previous gear, and smaller than the shiftpoint). These are the points highlight in orange. Under columns tagged W you got a sum and a count of the 'valid' cells for each gear. Then on the right you got it all summed up.

That's where 'interpretation' problems start. Because depending on gear one 100rpm cell can cover more mph, once you end up counting them up you might end up with a different number of cells counted, even though the speed range through which you went was unchanged. Change gears from 2.73 to 4.11 and I can just about guarantee you're gonna see it happen. So what's the problem? Well, you if you add up more cells, you're probably gonna end up with a higher sum, making the comparison moot. So there's two ways of going about this:
1. somehow 'normalize' the hp values according to rpm/mph factor of each gear (there's already some values for doing this, on top of Range colums)
2. go by the average hp per cell values. That's in red on the right. Seems to work fairly well, but I'd rather do it with the first method.
Let me know what you think.

Thursday, May 05, 2005

Injector Duty Cycle and Volumetric Efficiency

In few previous posts I was playing a lot with injector sizing, and figuring out duty cycles, to gauge what kind of injectors do you need for your application.

Few days ago I came across this . Basically it outlines how we can convert injector duty cycles and air fuel ratios into a fairly close approximation of volumetric efficiency.

First you start out with converting injector pulse width into duty cycle. That I covered few posts ago. Once you know how much fuel gets injected, you calculate a total fuel flow. That takes the duty cycle, number of injectors and fuel pressure into consideration.
Airflow is not that far off, since you know your AFR (if you have a wideband; if you don't, you just go by the target/commandered AFR from PE tables). The problem with this one is units. You got lbs/hr and you somehow have to end up with cubic feet per minute. That's just math though, and not a complicated one, you just have to watch it.
The actual calculation of VE is a division of airflow vs displacement, as that's theoretically the maximum of what you're supposed to be able to flow though it. Of course forced induction and big cam/NA applications can go over 100% VE.

So DOWNLOAD spreadsheet. All you really need is RPM and Injector Pulse Width. If you have a wideband, then you can also go off that for your AFR muliplier. This spreadsheet is set up with 12.2AFR in mind, so if you have a different goal, please change it.

In the spreadsheet you downloaded, you'll see two tabs with two different sets of data. One is for stock '99 F-body 26.4lbs/hr injectors during a high speed run with 'good air' where the Injector duty went over 100%. The other graph is for SVO 30lb/hr injectors with a much more healthy Injector Duty Cycle. The weird part is that the VE changes from one sheet to the other quite a bit. Both FuelFlow and AirFlow are actually higher on the smaller injectors. So what makes the big difference? Is it:
1. weather conditions--the 'good air' day was really spectacular, car ran like mad. Just before I swapped the injectors I did some runs just so I know where I was on Duty Cycle that day, and it showed 96% max, while on the good day I've seen it as high as 109%.
2. Old injectors were being pushed beyond their limits as far as flow was concerned, and the flow might have been actually misreported, so that whole set of data as far as VE is concerned is just plain wrong.
3. different day, different injectors, injectors going static, too many variables, stop worrying, you can't compare the two?

Monday, May 02, 2005

The new injectors and O2 sensors are IN!

Last week I did something very strange: I wrenched on my own car! As strange as that sounds, I am severly 'mechanically challenged' and usually all my alterations/modfications/new parts are done by my mechanic. But even with most of the parts being bought not at retail prices (forums, ebay, friends, etc) it gets expensive quickly. So some simpler things, I've wanted to do myself.

So last thursday, right after work (5pm) I put my car up on ramps and jack it up on a side, and slide myself underneath, looking for the front oxygen sensors. Not only I was able to find them, but I ended up exchanging them almost entirely by myself! Granted, due to them being there for almost 80k miles everything was quite fused together with heat (they're up there, right after header flanges and before catalytic converters) so it took multiple takes of PB Blaster and some other rusty bolt removing spray thingy (can't think of the name right now) to get them to budge. That combined with the fact that I only had an adjustable wrench, and not a real 7/8 open end one, made this a long affair. What I was told was a 15 minute swap, took me 4hrs. But it fired up on first try, and after they warmed up, started oscillating quickly, which I could actually verify by looking at the HPTuners scanner. I was very happy, and proceeded to go get drunk in celebration with my neighbours and a friend that have been my peanut gallery laughing at my grease monkey skills (or rather the lack thereof).

Saturday was injectors day. As per previous posts, on a day with 'good air' I hit 109% duty cycle on stock 26.4 lb/hr injectors, which made me rather scared. I've picked up some Ford SVO 30lb/hr injectors, as I figured that at one point I will need them. And what do you know, that moment finally came.

So a guy I've known for a long time told me to come over and we'll do them. We started at 2pm, and took out the LGMotorsports 3point strut tower brace (which is a bitch, unless you have the suspension completely unloaded) as it was completely in the way. Bunch of other stuff was in the way too, throttle line, all the cabling going to IAT/MAF sensors, and a bunch of things I don't even know what they're called. Once everything was out of the way, we let out the fuel rail pressure, took out the clips holding the injectors, and disconnected the electric harnesses. Few more bolts, and the entire fuel rail was loose, so we lifted it and flipped it around. Once on it's back, we took out the first injector and got sprayed with gas. Lovely, that's exactly why I don't like wrenching. There was a decent amount of gas still left in the rail and injectors, so we spent quite some time letting it leak out on some rags. Then we popped out all the injectors and o-rings left in the injector seats. Cleaning the new injectors was next, we checked to see if there aren't any visible clogs/damage on them. I wish I had the time to send them out for real flow test and a cleaning, but this whole 'out of injectors' situation sneaked up on me from nowhere, and I have a trackday to do in about a month, so I needed to get this stuff done ASAP.
Once the injectors were in, everything went in fairly smoothly, and we mounted it all up without major hassles.
The funny moment came of course once it was all done and ready for the first 'fire in the whole' test, as early on we decided to cool the engine down faster by turning on all the engine fans, but running it entirely off the battery. At one moment the fans just went out, and I though it was some automatic precaution of draining the battery. No such luck as 'timeout' for fans apparently, the battery was dead. So we fired it up from a gigantic charger, and it seemed to work ok, but the idle was surging and hunting, as we didn't flash the PCM with the new IFR tables yet, as we just wanted to see if it works (and waited a bit for the battery to charge up, as we didn't want it to die half way through flashing the PCM). Then the new calibration went in, and the car idled perfect right away. We took it for a spin, and top of second gear at 6300rpm yielded somewhere around 75% duty cycle, while on stock injectors earlier that day it would be closer to 96% (the air wasn't as good as the 109% day). So I can now finally start safely proceed with more tuning and not be afraid of going lean because of injectors not being up to a task.

The amazing part happened on the way back home. I was in MAF mode, logging for all my usual stuff, and the car ran perfect, with ZERO other alterations. That's what happens when it's all calibrated right folks. None of that old school 'lets calibrate AFR with IFR multiplications and MAF shifting' crap pseudo-science.

The only thing right now that's a bit off that I just noticed going through logs are the values of oxygen sensors at WOT. They consistanly average between 820-830mV for Bank1 and 835-855mV for Bank2. I should be at 12.2AFR, so shouldn't that be closer to 900-910mV?
Something's off about PE, but now that I have the hardware in place, I can finally go play with PE/WOT/Open loop mode.

That's probably gonna be another story though...

Logging for MAF calibration made easy

Apparently if you just want to adjust your MAF calibration you do not need to get into Speed Density mode! A guy named Drew approached me on the HPTuners board, asking why we need to go to SD for MAF calibration. I never really questioned this procedure, as usually I do both at the same time (as new MAF will need a new VE) so I must have it in Speed Density anyway.

So I told Drew to do a little experiement, do few logs in SD and then another few in MAF. We compared the results, and they were nearly identical, any differences were mostly based on not having enough samples. The more samples for given calibration point we had, the closer the two calibrations got.

The good part of this is that you can grab any number of data you got, any mode too, and just keep using the historical compounding function of my spreadsheet to get large numbers of samples, creating a truly universal calibration.

This little fact gets rid of the most annoying fact: you never seem to have enough data for the higher frequencies, as it's hard to get on the gas that long without getting into some hairy situations. But over a course of few days or weeks of more casual data gathering, with an occasional WOT run here and there, you will eventually get enough data. How much is enough? I'm satisfied with my data once I go over 100 samples for each interval. 100 samples at higher frequencies especially, create some very little discrepancies as far as the calibration is concerned. At lower frequencies, I guess since there is more 'leeway' for the air to move in, they still differ, but you end up with so much data for these (because you get it just sitting at stoplights and getting groceries), that the calibrations come out within just few percent of each other.

Hope that speeds up the process. I just spliced a bunch of data files I've had after my last volumetric efficiency altering modification (LS6 intake) and now I have thousands of samples for each calibration point, hopefully yielding a true calibration of my Granatelli MAF.