Sind die Ulf-Test-Zeiten realistisch? | Posts 16+

ObenbeiMutti wrote:

Do you log the times with VAG-Com support, or do you rely on the speedometer readings for your measurements? It's not just the speedometer that's inaccurate; the tachometer also lies to a significant extent.

I manually adjust the stop point to DZM.
You seem to be right about the incorrect reading: up to approximately 3000 rpm, it synchronizes perfectly with VAGCOM, but above that, it starts to show a slight advance: at 4000 rpm, VAGCOM indicates approximately 40 rpm more than what is shown on the tachometer.

Quote:

I'm just wondering that I can keep up with the PDs quite well with my modified AFN G3, probably because of its low weight of 1120 kg. However, the feeling of "pop" in the PDs is still worlds away from what I experience?

Because the AFN turbocharger's boost isn't as noticeable: it delivers boost even at very low RPMs, similar to a PD (presumably referring to a different turbocharger) with roughly "half the boost pressure."

Quote:

More torque at the same power would mean reaching higher RPMs faster. However, people always talk about power-to-weight ratio, never about power force. Why actually.

Because fixed data (weight and maximum power) are easier to understand than trying to determine the power output from graphs based on specific RPMs, or calculating the power from the Nm maximum and its corresponding RPM.

Quote:

If I were to install a 150 horsepower motorcycle engine with 90 Nm of torque into my car, I probably wouldn't be able to get it moving much, even with the extra power, right?

If the transmission were short enough, then yes. It really depends on how the power is distributed across the usable speed range.

For example, 150 horsepower at 12000 rpm and 90 Nm at 6000 rpm could be realistically simulated as a rough TDI equivalent using a 3:1 gearbox, resulting in approximately 270 Nm at 2000 rpm and 150 horsepower at 4000 rpm .

I rely 100% on VAG-Com.

@Ulf
Why manually stop the process when you can do it with VAG-Com?

@ ObenbeiMutti
How much fuel do you need to maintain a speed of 2000-4000 RPM in 3rd gear?

Hello,

@Marfan,

Just now, it's also about 5 seconds.
The car is very light, weighing 1120kg according to the registration document.
The engine is supposed to produce only 125-130 horsepower (stock is 110 hp), but I've slightly reduced the power output due to excessively high particle emissions.
I don't know if the car doesn't come with an upward-facing sprinkler system as standard .
I can reach approximately 207 km/h on a flat road without the device, and 212 km/h with the device, but I'm not sure if that provides any information about the power output.
I simply don't like relying on my tachometer and the listed weight.
Maybe I should go weigh myself sometime.

Marfan wrote:

@Ulf Why are you stopping it manually when you can do it with VAG-Com?

Because the RPM and, especially, the time resolution of VAGCOM are too inaccurate for my needs.

The RPM resolution is 21 RPM, and the sampling period at 1 MWB is 0.39 seconds on my car.

This leads to the following worst-case assumption, already at the time level:

Measurement 1
The engine will reach 2000 rpm approximately 0.19 seconds after starting, and 4000 rpm in about 6.81 seconds. Please note that both time points are located within the middle of a sampling interval.
Assuming that the log shows values closest to the measured RPM, with 1995 and 3990 RPM occurring at 0.39 and 6.63 seconds, respectively. This results in a measurement time of 6.24 seconds.

Measurement 2
Due to slight headwinds or similar factors, the vehicle takes 0.1 seconds longer to reach certain speeds between 2000 and 4000 rpm. However, the log now shows 1995 rpm at 0.0 seconds and 3990 rpm at 7.02 seconds.

Result: DZR time = 7.02 seconds, which is 0.78 seconds longer than in the previous run, although the actual difference is only 0.1 seconds.

The "noise" or error margin in VAGCOM diagnostic data recordings could, in my car alone, amount to roughly 0.7 seconds on the time axis, which is about 10% of my typical values. Additionally, there could be an extra 40 rpm on the rpm axis, representing about 2% of the measurement.

In total, the VAGCOM-DZR can inflate the minimal differences between two measurements by up to 12%, which results in calculated power differences of around 15 horsepower for the ASZ. However, I find the stopwatch to be much more accurate (less than 2 horsepower in multiple measurements).

That's my opinion on reading the DZR (Direct Reading Zone) via VAGCOM.

If there are any flaws in my reasoning: I am open to any factual discussion (at my level of understanding ).

Maybe they should also go weigh themselves sometime.
I've been wanting to do that for a long time, but where can I do it?

@ Ulf
I do things differently.
I compared the VAG-Com readings with an analog tachometer: 50 km/h (VAG-Com) corresponds to 2000 RPM, and 101 km/h corresponds to 4000 RPM.

Marfan wrote:

I do things completely differently. I compared the VAG-com speed and an analog tachometer => 50 km/h (VAG-Com) corresponds to 2000 RPM, and 101 km/h corresponds to 4000 RPM.

So, you're manually limiting the speed between 50 and 101 km/h instead of 2000 to 4000 rpm, and you're only using VAGCOM for "calibration" of the measurement?

Then, inaccuracies creep in again due to the 1 km/h resolution of the VAGCOM "speedometer":
2% at 50 km/h and 1% at 100 km/h.

Furthermore, you're lucky that you can at least once get the reading on the speedometer exactly "on the mark."
What would you do, for example, if you were traveling between 53 and 107 km/h?

Hi Ulf,

For my part, I interpolate the measurements in VAGCOM because VAGCOM always provides a pair of values consisting of engine speed and the corresponding time. Assuming a linear progression of the speed between two measurements, you can use the rule of three to calculate the values for 2000 and 4000, and this is likely to be more accurate than any manual estimation.

Your example above is not understandable because, in order to interpolate, you always need the values below and above the target RPMs.

Looking at the plot generated with this method, you can see that the error introduced by assuming linearity is quite small and far from 0.7 seconds.

Example (completely made up): 0.34 seconds at 1900 RPM, 0.64 seconds at 2100 RPM, and then, for exactly 2000 RPM, the value of 0.49 seconds is used in the calculation. Similarly, but in a slightly more complicated way, if the values are not symmetrically distributed, you can solve the linear system of equations.

Or have we misunderstood each other?

Best regards,

Jan.

Jan6K wrote:

Or have we misunderstood each other?

Hi Jan,

I think that:
I wasn't concerned with the accuracy of the interpolation (which I omitted in my example by assuming appropriate rpm values), but rather with the "relatively poor resolution" of VAGCOM, especially in the time domain.

A 21 rpm grid for a measurement range of 2000 rpm, in my opinion, still corresponds to a resolution of almost 7 bits – which would still be acceptable...

However, a 0.39-second grid with a measurement duration of around 6 seconds, in my opinion, corresponds to a resolution of approximately 4 bits, and I find that a bit too coarse, especially when dealing with comparative measurements (e.g., before and after a tuning), where the difference might be only 1-2 horsepower.

Okay, so I've once again ventured far outside my comfort zone.
If I've messed something up -> feel free to correct me .

Hi Ulf,

I think I understand the misunderstanding.

From the fact that the measurements are taken in a 0.39-second interval, you can infer that the timing resolution of VAGCOM is only that precise.

To my knowledge, that's not the case; they have a resolution of at least 1/100 of a second. I infer this from the fact that the timestamps in the decimal part (at least upon a cursory glance) can take on any arbitrary "xy" value. If there were a limitation like the one with 21 revolutions per minute, then certain values wouldn't be visible, but only values within a grid (which wouldn't necessarily be related to seconds; it could be something else, such as multiples of 0.0031415 seconds, for example, which wouldn't be so easy to see).

I recall that someone once wrote that this internal timer actually operates with a resolution of 1/100 or less. We've actually experienced this before, when people started using VAGCOM for data interpolation.

The 0.39-second grid is partly due to the system not being able to transfer the data quickly enough via the K-bus. For example, if you log 3 groups in parallel, you'll have three times the distance, but still the same resolution of 1/100 for the measured values. It goes even further: while each value from a triple-log is only read every 1.2 seconds, the timestamps provided for each group indicate that the system also provides values "in between" those readings.

Here's an example of a cell from a 3-log table in my ASV (Amplified Spontaneous Emission Variable) data.

49.66, 23.4, 22.5, 35.1, 50.12, 1008, 440.0, 355.0, 48.6, 50.58, 966, 1040, 1000, 24.7

The first value was 49.66, the second was 50.12, and the third was 50.58. If you continue looking through the data log, you'll also see that the intervals are not the same; they fluctuate slightly, which is logical because the data acquisition with low priority runs and can therefore be delayed by other CPU activities. Additionally, the ASCII transfer (I'm not sure if MSG already sends ASCII data, or if VAGCOM interprets it, which could also contribute to the delay, because for example, a value of 990 requires 3 bytes for the engine speed, while 1000 requires 4, which takes longer to transmit. You can check this in the ASZ). I suspect that not every data group takes 0.39 seconds, but that some are faster or slower. That would be another indication of a resolution of 1/100 or less.

Therefore: We have a pair of value and timestamp every 0.xy seconds, where the timestamp has a resolution of at least 1/100 of a second. This allows us to interpolate well and reliably, except that the 21 U/min (revolutions per minute) is causing interference. The error resulting from this would need to be calculated by performing an interpolation, where you assume a deviation of -10 on one side and 0 on the other , and then comparing that with the result of the reversed calculation.

However, I still believe that the error will be no more than 0.05 seconds, if at all.

Best regards,

Jan.

Jan6K wrote:

Therefore: We have a pair of value and timestamp every 0.xy seconds, where the timestamp has a resolution of at least 1/100 of a second. We can interpolate well and safely with this, except that the 21 U/min is causing a problem. .

Hm... well, okay: I don't object because I don't have any evidence against your statement .

To verify this, one would need to perform a series of DZR tests sequentially and calculate the time it takes to reach 2000-4000 rpm from the resulting data.
Always on the same route, in calm conditions, etc., in order to keep the actual time differences as small as possible.

And then, consider the largest difference in the calculated times for the 2000-4000 rpm range.
According to your assumption, it should be less than 0.1 seconds, but according to my estimate, it should be at least 0.5 seconds or more, with a distribution that resembles noise.

But I won't be doing those kinds of series tests: it's too complicated for me, as I'll stick to using a stopwatch anyway .

Hi!
'Today, I tested the acceleration of my car a few times, from 50 to 100 km/h, with the engine speed at 2000-4000 rpm, and I consistently got values of 5.5 seconds. According to the vehicle registration, the car weighs 1250 kg, and it has a power output of 108 hp (it was originally 90 hp and has been chipped). With adjusted torque values, 6.5 seconds seems like a realistic expectation. I also drove in both directions on the test track! However, these 5.5 seconds are comparable to the 130 hp PD model, which shouldn't be possible. Just for information, it's a Ford Focus 1.8 TDDi.'
Best regards, Olaf.

schnappi wrote:

However, the 5.5 seconds are comparable to the 130 hp PD??? which is actually possible! By the way, it's a Ford Focus 1.8 TDDi, just for information.

Depending on how the torque curve of the tuned Ford looks, you can dismiss the dyno run as unreliable as long as you don't have any reliable data.

The same applies to the overall conversion from revolutions per minute (rpm) to kilometers per hour (km/h). The speedometer only provides a very rough estimate in that case...

However, the 5.5 seconds are comparable to the 130 hp PD - which is actually possible! By the way, it's a Ford Focus 1.8 TDDi, just for information.
Depending on how the torque curve of the tuned Ford looks, you can dismiss the dyno run as unreliable as long as you don't have any reliable data.

The same applies to the overall conversion from revolutions per minute (rpm) to kilometers per hour (km/h). The speedometer only provides a very rough estimate in that case.

I only used the data provided by the manufacturer of the device. I entered that data, and it resulted in a value of 6.5 seconds.
Too bad, it could have worked out .

Hi,
@ Ulf
I'm using VCDS to check the speed, but I never thought there could be such a discrepancy.
Measurement 2.
Due to slight headwinds or similar factors, the vehicle takes 0.1 seconds longer to reach speeds between 2000 and 4000 rpm. However, the log now shows 1995 rpm at 0.0 seconds and 3990 rpm at 7.02 seconds.
I don't understand .
I was expecting your example, and I got 0.25s -> that's the biggest difference.

Hi,
Lupo 3L, 61 horsepower.
Measured at 9.5 seconds with a measured weight of 1020 kg.
greetings

Marfan wrote:

I was expecting your example, and I got 0.25s -> that's the biggest difference

.
Maybe, but it's not that important anyway.

For me, at least, this isn't a topic that's worth discussing to the very last detail.