RI A6

Group 120-2 provides what the ECU computes as the actual engine torque. Fun stuff. If you log this along with engine rpm, you can compute torque and HP curves. So that's what I did yesterday and today.

Yesterday I was playing, so I used the advanced measurement capability of VAG-COM, and logged a whole bunch of parameters, including torque and HP. This next plot shows a composite of many different acceleration and deceleration runs around town. I post processed the data and left in the highest torque values from each rpm point.

Because it is a composite, there is a bit of noise in the plot, but it looks close to what I might expect. There is, however, an interesting lack of torque falloff at high RPM and some end peaks that are interesting.

<img src="http://www.teraspeed.com/RS4/composite%20pulls%201-3-07.jpg">
<u><b>Yesterday's run</b></u>

So, today I did some more controlled test runs. I started from a slow roll, changed to 3rd gear, and went to full throttle until I hit the rev limiter. This was the cleanest looking plot tonight. Interestingly enough, it looks strangely similar to some of the early RS4 dyno runs that we've seen. There's that funny lip right around 2000 rpm and a very sharp drop off in torque after 6000 rpm, that is not evident in the plot from yesterday.

As I drove today, before I looked at the data, I thought that the car didn't feel as strong as it often does and did yesterday. Yesterday it felt like it just wanted to take off, and it is seen in that ending HP peak. And today, the there was a stong kick at 5500 rpm, followed by a mushy feeling. Not impressive like I usually think.

So, I'm thinking WTF, something real is going on here. I have a few ideas, and I have the tool to test them out. But that impression we all have from time to time that she seems to go at a rocket, and then at other times seems to be a bit sluggish is real. Could be fuel trim adaptation, fuel rail colapse, intake air flow turbulance ....

(UPDATE: The lip is caused by S-mode disengaged. This run was done with the s-button in the off model.) I'm thinking the lip at 2000 rpm might very well be the intake stalling at wide open throttle. Starting out in 3rd gear at 5 mph there probably is not enough change in rpm to keep the air flow out of turbulance. Doing some pulls in 2nd gear should even out the lip at low rpms.

At the high side I'm actually interested in the kicks in torque on the end. It could just be an anomoly when the throttle shuts down and the way torque is measured by the ECU with the mass air flow transducer.


<img src="http://www.teraspeed.com/RS4/3rd%20gear%20pull%201-4-07.jpg">
<u><b>Today's run: S-mode off</b></u>


Tomorrow and this weekend I'll do a few more runs, and I'll also try some VAG-COM Butt Dyno pulls, since the Torque numbers produced by the ECU are not calibrated. However, if you want to do some before and after tuning comparison, it is pretty nifty. Even though the torque number the ECU reports is not calibrated, it should be quite repeatable on one car, and may well track multiple cars, if the MAF transducer is accurately manufactured.

Those who have substituted a K&amp;N air filter should be able to easily do a few before and after pulls to determine if it makes a noticable difference. If used at a dyno day, we should be able to use VAG-COM to calibrate the torque readout of the ECU with the torque readout of the dyno. How cool is that? For those that are interested, on the Ross-Tech info page is a link to a Block 120 torque calculation spreadsheet.

http://www.ross-tech.com/vag-com/cars/index.html

And on this next link is info on the "Butt Dyno" method with VAG-COM.<ul><li><a href="https://forums.audiworld.com/vag/msgs/457.phtml">Butt Dyno with VAG-COM</a></li></ul>

sparkhill

Interesting... Thanks!

nyet

its just a number motronic uses to determine fueling etc.<ul><li><a href="http://nyet.org/fats">if you want to convert group 3 to hp/tq.</a></li></ul>

nyet

<ul><li><a href="http://s4wiki.com/wiki/ME7.1">http://s4wiki.com/wiki/ME7.1</a</li></ul>

VerrÜckt

Log some 3rd gear pulls and use nyet's calculation tool!

RI A6

In the env file you allow the definition of an rpm_start, rpm_end and redline numbers. Now, I'm no Perl expert, but you seem to have hard coded a 6900 rpm limit in the perl script, no matter what the rpm_end and redline limits are, and I don't have a Linux box to run the code on.

<b>
foreach my $row (@{$d-&gt;{'data'}}) {
next if(($row-&gt;[$rpm])&lt;3900);
next if(($row-&gt;[$rpm])&gt;6900);
</b>

I'd suggest that you change that limit based on the redline. If I've misinterpreted the code, I apologize.

Also, your algorithm assumes a fixed driveline loss, doesn't it? I wonder why you don't just plot WHP, which is what you are actually calculating. I'm wondering why you didn't use the Ross-Tech "Butt Dyno" method, to determine driveline loss from measurements.

RI A6

There are algorithms in the MED 9.1 ECU to determine engine torque from Air/fuel ratio and mass air flow. This is passed along to the torque control and traction control systems. Unfortunately, the algorithms are not documented publically, as far as I've found.

Your block diagram of the ME 7.1 ECU torque control is interesting, however, it is simplified. You'll see some better block diagrams of the newer systems in the Bosch Book, Gasoline-Engine Management, 2nd Edition, which also talks about Gasoline Direct Injection. Torque calculations need to be a bit more accurate with GDI in order to accomplish switchover from Stratified Charge mode to Homogeneous mode, without bucking. I believe that the torque numbers reported are not necessarily calibrated to absolute torque, but they are good from a relative torque/HP curve standpoint.

VerrÜckt

I can't for the life of me find it anymore though.

RI A6

My point is that it's not "just a number". It is derived from first principles. Otherwise Motronic torque control theory would be meaningless. But there are some assumptions behind it, and there are inaccuracies in the sensors that can cause the numbers to be off. The only sensor in the system that is truly dead on is the after cat bang/bang lambda sensor. It is the primary control sensor. Unfortunately, it tells us absolutely nothing about torque.

Nyet's algorithm is also derived from 1st principles ... at least until he goes to compute HP by assuming a fixed scalar drivetrain loss factor. That's just plain wrong.

RI A6

After doing a bit of reading, there is one significant difference between the ME and MED units, and that is closed loop control of the fuel system. I can see where the torque number provided by the ME 7 ECU could be wrong, since fuel flow is assumed by the injection timing.

In the MED 7 and MED 9 ECUs, (the D meaning direct injection) fuel rail pressure and injector timing is under closed loop control. It has to be this way for the ECU to run in Europe in the Stratified mode, where the throttle angle is kept wide open, and torque is regulated by metered fuel flow. The calibration procedures for engine parameter mapping with the MED ECUs is quite a complex process. The published claim is that torque is reported with an accuracy of +/- 2%, validated on dyno, and that accuracy is necessary in order to effect a seamless transition from Stratified to Homogeneous mode and back again.

The question then, is which torque is being reported by the ECU. Since Bosch defines several points where torque is computed:

Internal torque from combustion
Torque after losses from gas exchange and friction
Clutch torque (where Losses due to auxilary components: A/C, Pumps, Electrical, are accounted for.)
Gear and Transmission Torque
Wheel Torque (the torque available to the drivetrain)

Since Torque is reported from Traction Control (ASR) Group 120, I am inclined to think that torque is being reported as delivered at the driveline, before wheel and road losses. But WTF do I know?