Audi B6 series S4 Throttle Research

Audi B6 S4 (4.2 V8) Throttle Research

One of the most common remarks you hear from anybody buying or test drving Audi's latest S4, the 2004 model (B6 chassis variant), is:

"Wow, this thing is hard to shift!" or

"Man, I can't get the hang of this clutch" Well, the problem is not you - and especially, the problem is not that "you can't handle a car with this much power." The problem is that Audi has programmed all kinds of tricks into the "drive-by-wire" throttle, and some of them unfortunately make you a worse driver. Skeptics love to think that you just can't handle this much car and that if they were in it, they wouldn't have any problem. Not so.

(before I go on, here is a link to a response that I wrote to somebody recently, which captures some reflections from a while after I wrote this original article and a pseudo-solution for the non-feint-of-heart).

The purpose of this page is to track the research I am doing into the fact that I find the behavior of the throttle on my 2004 Audi S4 unpredictable. In particular, this car has a "drive by wire" throttle. This means the gas pedal ("accelerator" as it is actually called) does not directly control how much gas goes to the engine, which is the case in most cars. Instead, in this car, there is a sensor on the gas pedal which tells a computer how far you have pressed it, and the computer controls a motor in the engine that controls the throttle.

Here are quick shortcuts to each major piece of reasearch:

low rpm full throttle (you are here)
a note about the graphs
mid rpm half throttle - doubling
unpredictable delayed reaction
This is the worst of all the problems only because it has the most pronounced and frequent affect

It seems that the program running this computer has some "features" to it that result in the engine not always doing what you tell it to do. The three main problems are:

Unpredictable delayed reaction
Over-reaction to slight input (blips)
Lifting accelerator does not close throttle

At this point, Audi has acknowledged that all of these beahviors are as-designed, and there is nothing wrong with my car.

In some cases, there is a significant delay before you accelerate. In other cases, delays can be much shorter, but if they happen while shifting, they cause a jerky ride and burning of the clutch, which is not the driver's fault.

And in the worst case, the car can keep accelerating after you lift the accelerator. Note that lifting all the way always works, it is only partial lifts that are sometimes ignored or have a delayed reaction.

Research

I am going my research with a HEX-USB adapter and VAG-COM software, both from Ross Tech. This is a kit that does pretty much what an Audi service department's diagnostic computer does. This kit is a little more expensive than you might guess, but it is a hundredth the cost of the real Audi tool. Plus there is a limited market of people who are both computer and car geeks and who own Audis. Given that, the fact that a piece of hardware that does this costs less than M$ Office is pretty amazing. Please forgive the shameless plug for Ross Tech, the guys there have been instrumental in enabling everything I've figure out so far, and they deserve it.

Results

Ok, here's what I have figured out so far:

When you hit the gas at low RPMs (<2000), the computer will not give you full throttle. It looks like at 1000 RPMs, it gives you only 70% throttle, and it gradually increases it to 100% throttle as you approach 2000 rpms.
the disturbing side effect of this (other than that you don't accelerate as much as you might), is that as you back off the pedal, you may still be past the computer's idea of "full throttle" at that engine speed. This means that the car will not slow down as you should rightly expect, but in fact, it will keep speeding up, until your foot, and the computer's idea of what rpm that throttle position should be allowed, cross.
There is another case at higher rpms of acceleration which continues after you have lifted the accelerator. I don't fully understand this one yet, but I'll show the graph.
I also have evidence of truly delayed throttle response, as in you hit the gas, and nothing happens for a half second or so, and then you get the right response, just not when you expected it. I don't understand this one yet; I suspect faulty sensors, and possibly that they are more likely to fail when they are cold. Graphs forthcoming.

A note about the graphs

The left y axis is %. The right y axis is engine RPMs. The horizontal axis is seconds since the start of that data collection session. I've gotten more sophisticated as I've gone along, learning mor about the VAG, etc. So, earlier graphs often don't have RPM data collected. And the latest ones include both RPM data and speed data. The speed is also graphed on the left axis, but is to be read in MPH, not % (but the scale is still 0-100, even though no tests go much over 30 mph).

In short, green & yellow are accelerator position, and red & blue are throttle position. These values have all been adjusted & normalized b/c the real values in the car don't actually rage from 0-100.

In all of these graphs, green and yellow show acccelerator position. There are two sensors that measure it. They agree on my car, so you often can only see one b/c the the other covers it. Note that I have adjusted the graph to show a 100% scale. In truth, the sensors read from about 14% to about 76%. Also the reason one is called "accelerator50" is that this sensor is actually designed to read in values that are half what the other sensor reads. This has also been adjusted.

Red and blue show actual throttle position. Similarly to above, you can usually only see red. Thes have also been adjusted. The real ranges are about 14-86%. The second sensor is called "throttle0" because it actually reads the opposite of the first (it tells you "how closed the throttle is." So, it's "maximum" theoretical value is actually zero. I've adjusted this as well.

Note also that in some graphs you may see rpms drop below what appears to be idle. This is real. I did much of my testing at or below 1000 rpms, in 6th gear. This was deliberately to put the engine at as low as possible a power output, so I could get several seconds of data without ending up at an absurd speed.

RPMs are the thick white line.

Low RPM Acceleration

Here you can clearly see two things;

that with the pedal floored, you don't get full throttle for quite a while, and
even after you start to lift the throttle significantly, the car keeps accelerating. You can see this because the revs keep rising, and the throttle stays wide open.
The fact that the throttle reached 100% exactly when I started to lift is purely coincidence.
In this graph, I deliberately lifted the throttle slowly, rather than just popping my foot off completely.

High RPM acceleration

Here you can see that even above 4000 rpms, going from floored pedal to less than floored results in continued acceleration. Alarming because you are "lifting the accelerator" and have every right to expect that you will thus "accelerate less."

Notice that RPMs continue to rise steadily as the throttle is lifted, for a little while (about a full second).

Mid-rpm Throttle Doubling

Another discovery I've had is that in some cases, full throttle can be activated by as little as half accelerator usage. In other words, pushing the pedal half way down gets you full throttle. Pushing it more does nothing, and lifting can still result in continued acceleration. This isn't surprising, think of it like this: 50% accelerator gave you max throttle. you keep pushing to say 80% accelerator - you still have max throttle. Now you lift to say 60% accelerator - but that's still past what was giving you max throttle. So, you think you told it to slow down, but it's still speeding up.

I've only researched this breifly. It appears to be RPM depenedant; at higher rpms, it appears to take closer to 100% accelerator to get 100% throttle, but at this point I need to collect more data. Also, I apologize, but I forgot to actually collect RPM data on this run (I was actually going somewhere, the data collection was an afterhought).

Following are two different graphs showing that far less than 100% accelerator can still give you 100% throttle

note: In these graphs I have improved my scaling of the values, so "no pedal/ no throttle" now both show up as zero.

Delayed Reaction

This problem happens primarily when the car is moving. It is easiest to reproduce when coasting, with the engine at idle, but that is not the only time it happens. This problem has three versions:

Sometimes you press the accelerator, and the throttle either doesn't respond at all for a moment, and then responds way too much, resulting in the engine racing
Sometimes it responds for a moment and then cuts out
Sometimes it responds with only a small fraction

This is the worst of all of these issues because it is unpredictdable. The others are frustrating and make you a worse driver, but you can at least guess when they're going to happen, and make room fodr them. For example, you can lift the pedal more than you want to avoid over accelerating - it'll make for a jerky ride, but it will stop you from sliding off an on-ramp. This one is so variable that there is nothing you can do about it. It will burn your clutch, it will knock you & your passengers arouind, and it will leave you stuck in intersections.

You might be wondering why you would ever be at idle while moving. Sometimes it is simply that you are downshifting from low rpms, and it drops to idle while you're getting into the next gear, and you want to rev the engine a little so disengaging the clutch doesn't make the car buck. Being in heavy traffic is one reason. Going around a tight corner is another, sometimes you have to go too slow to actually be in a gear. Or if you're in first gear it's too easy to accidentally jerk the car around, and it's safer to coast.

Some Graphs

Here we have a delayed response where the throttle opens much slower than requested at first, and then suddenly catches up. Here are three examples of me "blipping & pressing" to match the engine's revs to the car's speed. This is what you would do to quickly get the engine to speed (moderate throttle) and then hold it at that speed (less throttle).

The first two exhibited immediate (0.1 sec) response. The third one gave only about 5% response, which was effectively nothing, and the revs didn't respond. Then as I pressed (more than normal, which is the subconscious reaction to the fact that the blip had no effect), there was a significant delay (about 0.4 sec) between my press, and a proportional response from the throttle. Not the difference in behavoir of the revs between the 3rd & the other two. It took 1.75 sec to get the engine to 1500 rpms, vs. 0.4 sec the other two times.

Here we have unexpectedly low response. This is me starting in 1st gear from a standstill. I blip the throttle to get some revs, and successfully get th engine to about 1500 rpms. I begin to engage the clutch, and come back with 25-30% pedal application. The engine responds with very little power, and as you can see, only about 12% throttle. It feels like I have a wimpy engine, and revs start to drop because the clutch is overpowering the engine. I increase the pedal from 26% to 38%, and the throttle suddenly overreacts, going from 15% to 55%.

In this case, the exact application of my clutch allowed the clutch to spin as the engine revved, so the resulting acceleration was smooth, but very hard on the clutch, and resulted in the 'burnt clutch smell'.

Here we have unexpectedly low response. This occured after engaging 2nd gear. At 4392 secs I accelerate in 1st gear, to about 15 mph. Then I declutched, and depressed the pedal to engage second. I accidentally gave too much, so I backed off, and came back with an appropriate amount. At 4395 sec, I backed off to about 8% pedal. The engine responded with 0 & throttle, so it backed off too much. Then I came back on with 30% pedal, and the engine only gave me 15% throttle. This felt like the car was "falling on it's face" to quote another owner; like I suddenly had a 1.8 liter 4 cylinder engine.