Sunday, April 09, 2006

IFR tweaking Jihad

A lot of people lately been asking me about IFR tweaking. I couldn't believe that after all these years of using VE with great results people still used the old hacks.

So these are two things I've written up to people about it. The first one is short and purposefully silly visual, the other one is more technical. I hope this will once for all put the end to all stupid IFR tweaking.

Take one:
So let's say you're mixing cement... ;)

Your shovel can move 1 pound of cement per second to the mixer (or whatever that's called) in which you supposed to mix cement and water in 1/13 ratio. You have a hose going to the mixer that you believe delivers 13lb/sec of water. however, the hose is clogged, and you don't know about it. All you know is that your mixture despite your best effort is a bit off. So instinctively, instead of checking what's wrong with the hose, you decide to grab a bigger or smaller shovel depending if your mixture is too watery or too solid to compensate for the change in amount of water.
In the end, did you get the right mixture? yes.
Did you get the right amount of it? No!

This is exactly what IFR tweaking does to your engine. Instead of looking at the source of the problem, you just try to make up for it somewhere else.

Take Two:
All the mods you do, all they do is increase the Volumetric Efficiency
(aka. be able to cram more air in, increase airflow). However,
the idiot tuners, they keep the Volumetric Efficiency table (which
exists exactly for this purpose) unaltered! So with more airflow and
same amount of fuel, you're basically making your car run leaner, first to the point where it makes more power, but eventually it goes well into the dangerous territory.

Here's a scenario to demonstrate this in numbers:
To keep the AFR sane, you gotta tell the computer to dump more fuel.
You can't do it directly, as most of this stuff is calculated dynamically or proportionally, so you just lie about the size of the injectors. This way every fueling operation is proportioned to the new 'size.'

So let's say you had 39lb/min of air coming in, and you're trying to
keep a 13.0 AFR (easy numbers example)
All the injectors have to deliever 39/13=3lb/min of fuel. there's
8 of them, so they flow 3/8 lb/min (0.375). Injectors are usually
rated in lb/hr so this is 0.375lb/min*60 min/hr=22.5lb/hr of fuel.
You don't want to operate your injectors at 100% as they will simply
overheat at one moment and stop sending fuel at all, exploding your motor.
80-85% duty cycle is the safe range in which most injectors can operate
without deminished reliability. So you want 22.5/0.80=28lb/hr
injectors. Oh, what do you know, that's a typical airflow and a
typical injector size in a stockish but tuned LS1! See, all these
numbers come from somewhere, and if you have enough equations for it,
you can see it how got there.

But to get back to IFR bullshiting...
Now let's say you put some swanky exhaust/lid/intake on your car, and
it flows 44lb/min of air, not 39. But if you don't touch VE table,
now you're gonna get 44/39 *13=14.666AFR!!! That's very dangerously
lean! So far we've been dumping up to 39lb/min relatively worth
of fuel. If we say we have injectors smaller by the ratio of 44/39,
it will automatically try to make the injector work harder, by
commanding more pulse width. The trick the 'IFR tweakers' rely on
is that the injecotrs didn't really get smaller, so now we're
commanding bigger pulse width on the same injector as before, effectly
spraying more fuel!
So it does what we wanted, right? Wrong! Pulse widths don't
scale linearly, so you're not really aiming for the right amount of
more fuel, you just know you're spraying more than before. This of it
as a sharp shooting competition but you're blindfolded and have a
helper that can only tell you 'up/down' or 'left/right' while what you
need is '30mm to the right and 3mm down.' You can see why the results
of such tunes are so crappy.

Then you toss in a cam with a naturally steepy VE (inefficient at lower
pressures, and very efficient under high pressure), and the whole IFR
tweaking scheme falls apart.
IFR has only one dependent variable(MAP), VE has two (MAP,
RPM). This means IFR can describe what VE could, except that it has
to consider RPM to be constant. Thus when tuning with IFR, you will
command the same amount of fuel whether you're at 1000rpm or 7000rpm.
big cams have VERY different flow characteristics at such broad rpm range,
and therefore IFR will never be able to reflect the precision
necessary to reflect such changes.

Another numeric exercise to demonstrate this:
A stock cam has about 50% VE at idle, and about 90% at WOT.
A nice big cam has 30% VE at idle, and up to 120% at WOT.
So to make a big cam idle nicely with an unmolested VE you're gonna have to
seriously bullshit about the size of your injector sizes, as you're needing only 30/50(60%) of fuel you used to need.
On the other end of the spectrum, you will need 120/90(133%) more fuel than before. So you lie some more about the injectors, making your IFR table look very sharp and jagged, what in reality is a nice smooth progression.
The real problem is that the IFR table has only one axis (pressure at the fuel rail) and only 17 values. VE table has 380 cells on two axis. If the natural VE dictated by the characteristics of the cam shaft is very sensitive to RPM (and most aftermarket ones are very sensitive) you will not be able to have the computer dump appopriate amount of fuel in all possible situations. So you have to simply your tune (that's why people do it), and tune it either for decent drivability (and have it horribly lean up top) or optimize it for power, and have it stall 5 times every time you try to get in/out of the pitlane.

IFR tuning to VE tuning is what carbs have become to electronic fuel
injection. They both get the job done, just one is way more flexible
than the other.