LMM Simulator: Train Large Language Models

Hello,

From the series: Things the world doesn't need .
Is there a lookup table that limits the maximum air mass at a given engine speed (-> otherwise an error occurs)?
Does anyone have a 'maximum' characteristic curve or any relevant literature references?
I've built a device that can simulate different voltages to the engine control unit (ECU) based on the engine speed, instead of using a mass airflow sensor (MAF). Now I want to calibrate it so that it simulates a MAF sensor reasonably well.
So far, I'm seeing a voltage of 2V at idle, increasing to 4.2V from 4000 RPM.
The performance follows a more or less linear curve. The acceleration up to 2500 RPM is still quite modest.
After that, things are fine. The values between 1000 RPM and 2500 RPM are causing problems. If I increase the speed too quickly, it goes into emergency mode. It's clear that the setting can only be a compromise.
Sure, I'd appreciate it if someone had a tip.

Regards,

Maciek.

Deleted upon request.

Hello Maciek,

Why don't you simply measure the voltage curve of a functioning HFM (which was also referred to as an LMM, or mass airflow meter, here) under full load?

Best regards, Rainer.

Hello Gustav, Hello Rainer.

It's a TDI AFN engine (Audi A4) with a Bosch mass airflow sensor, installed in a VW T3 bus.
'I have a characteristic curve for the mass airflow sensor (MAF) output voltage, but it's not useful because I'm not actually measuring airflow. Instead, I'm generating voltages based on a specific lookup table, depending on the engine speed (Pin 2). I need the engine control unit (ECU) characteristic curve for permissible MAF values, if such a thing exists.'
I have already performed measurements using a functioning mass airflow sensor (MAF). At full throttle, the voltage is already 3.4V at 2000 RPM and goes up to 4.3V at over 4000 RPM. What happens between 1000 RPM and 2000 RPM is difficult to say because it happens very quickly. The measuring device is slow.
That's also the problem. The transition from low RPMs to acceleration. Either it works and the car accelerates properly, or it goes into limp mode.
I'm missing some context about what other data might be relevant.

The idea actually came about when I wondered why the 'closed' EGR valve wasn't causing any issues with the engine control unit (ECU). So, it's just a bit of tinkering, but it would be great if it worked.
I'm still working on my speed curve, which ranges from 1000 RPM to 2000 RPM.

Regards,

Maciek.

Hi,

Ideally, the signal from the mass airflow sensor in a FUNCTIONING engine (i.e., especially without any leaks in the intake pressure area) should be replaceable using intake pressure, ambient pressure, ambient temperature, and engine speed.

"If you have the boost pressure and engine speed, you can calculate the airflow. Of course, if your EGR valve is broken, that simplifies the calculation. And with the ambient pressure and temperature, you can then determine the air mass. For a simulator, you could also reduce the latter to standard conditions, which might be sufficient (in that case, only engine speed and boost pressure would be needed)."

Just a question for everyone: Why doesn't the engine control unit simply perform a specific test to check the plausibility of the signal from the mass airflow sensor? This would likely reveal both faults in the MAF sensor itself and any leaks in the pressurized air intake system.

It's clear that this isn't being done, otherwise there wouldn't be a risk of over-revving turbos due to cracks in the intercooler pipes, and presumably no soot either.

Or could the reason simply be that the tolerance of the LMM signal is so high that it's impossible to perform meaningful calculations (i.e., calculations that would actually identify the errors mentioned)?

Best regards,

Jan.

Hello Jan,

Quote:

Just a question for the group: Why doesn't the engine control unit simply do something to test the plausibility of the signal from the mass airflow sensor?

This would likely reveal both damage to the MAF sensor and leaks in the boost pressure area. }
How do you plan to control the exhaust gas recirculation rate without an HFM (Hot Film Mass Flow Meter)?

Best regards, Rainer.

Hi Rainer,

Not at all. The question was aimed at whether, with the AGR (Abgasrückführung - exhaust gas recirculation) system closed, the control unit should not be able to check the plausibility of the HFM (Heißfilmdruck - hot film pressure) signal.

I certainly didn't want to suggest replacing the HFM, but rather to explore the idea behind the "simulator" and whether it could potentially be used for diagnostic purposes.

However, a prerequisite would be (as mentioned in Bertil's thread about the LuFi), that the HFM measures the signal relatively accurately and reproducibly, which I have doubts about based on Bertil's measurements.

Otherwise, the control unit should, in principle, be able to determine with a reasonable degree of accuracy whether the signal makes sense or not.

The AGR (Abgasrückführung) must be present; that's clear. Or, conversely: If the AGR is intact and functioning as intended (i.e., without any dirt or obstructions), then not only the HFM (Hauptflußmasser) and the pressure hoses, but also the AGR function itself should be testable.

In my opinion, the question is why this isn't already being done, because the computational effort involved wouldn't be that significant, and it would simplify debugging.

The answer that one doesn't need to do something because one doesn't have to monitor the AGR (Abgasrückführung - exhaust gas recirculation) doesn't seem entirely convincing to me.

I currently believe that the HFM signals exhibit very high levels of noise, making it impossible to obtain meaningful counter-signals.

What do you think?

Best regards,

Jan.

Well...

The EDC monitors the LMM for plausibility in exactly the same way.

BUT: the evaluation and display are not mandatory in Europe, but are only required for OBD-II systems.

Therefore, the familiar LMM drift remains ineffective, while in the USA, TDI drivers receive glow plugs and cannot operate with a faulty AGR valve.

What remains is solely to monitor the zero-point drift of the mass flow meter (which is done after the run), but this is not dependent on the contamination of the sensor surface...


The question regarding the LMM simulator is beyond my understanding; I don't see any real point in it. After all, the amount of air drawn in cannot be derived solely from the engine speed, and a new LMM does less work .
However, the documentation explicitly mentions HI and LOW limits for the air mass. Unfortunately, the LOW limit is not being detected (as mentioned above), while the HIGH limit is being detected (because it is independent of the monitoring mentioned above).

@maciek
btw: if the speed through that section is too fast (between 1000-2500), then just shift into 3rd gear and drive up the steepest hill you can find.
If necessary, bypass the brake pedal sensor and apply the brakes. Doing it a couple of times won't be a problem, you're not going to do it for 10 minutes.


CU Gremlin.

Hello again,

It's still not working correctly. Without a test run using a diagnostic tool, it will be difficult to determine the values (using a multimeter). So, I still have to wait for a diagnostic tool. Additionally, there are still problems with the potentiometer when the engine speed changes rapidly (the voltage fluctuates uncontrollably). Either the controller or the potentiometer is too slow, or the speed signal is experiencing interference. More investigation is needed... But, as Jan suggested, if we were to add the boost pressure signal and it actually worked properly, wouldn't it be a replacement for the mass airflow sensor (MAF)?


The question regarding the LMM simulator is beyond my understanding; I don't see any real point in it. After all, the amount of air drawn in cannot be derived solely from the engine speed, and a new LMM does less work .


Yes, actually. But it's just a whim. I got the controller for a...
'Intelligent' tuning boxes and potentiometers can be easily controlled (and still are, for now ).


However, the documentation explicitly mentions HI and LOW limits for the air mass. Unfortunately, the LOW limit is not being detected (as mentioned above), while the HIGH limit is being detected (because it is independent of the monitoring mentioned above).

Do you have more information on that? Is the way in which the characteristic curves are stored defined by a standard? I haven't found anything about it online yet, and it would be helpful to know what's included before I invest money in a description of it.


@maciek
btw: if the speed through that section is too fast (between 1000-2500), then just shift into 3rd gear and drive up the steepest hill you can find.
If necessary, bypass the brake pedal sensor and apply the brakes. Doing it a couple of times won't be a problem, you're not going to do it for 10 minutes.


I'm using the AFN engine in a VW T3 bus. The strongest production engine for that model had 173 Nm of torque. The AFN has a bit more... and I'm not willing to risk breaking the gearbox for the sake of that extra power.

Regards,

Maciek.

Jan6K wrote:

I certainly didn't want to suggest replacing the HFM, but rather to explore the idea behind the "simulator" and whether it could potentially be used for diagnostic purposes. However, the prerequisite would be (see Bertil's thread on the LuFi), that the HFM measures the signal relatively accurately and reproducibly, which I have doubts about based on Bertil's measurements.

Hi Jan,

I cannot dispel your doubts.

If one were to assume malicious intent, one might think that the LMM (likely referring to a specific component) is not checked for plausibility to reduce the number of warranty claims related to this expensive wear item.
As long as the indicator light isn't flashing, it's easier to dismiss complaints about a lack of power .

Hello,

Is there anything new here?
It would be ideal to have a type of amplifier that increases the LMM voltage according to a predefined characteristic curve, to disable the soot limit below a certain point and improve the response in all other areas. In principle, it would use the same method as with a diode, but the 'amplifier characteristic curve' would provide more plausible signals, so that the engine control unit (ECU) wouldn't complain.

Best regards, Matthias.