Boost Control: Charge Pressure via Air Mass Target

Hello,

The boost pressure control operates independently of the mass airflow sensor (MAF) value.
The engine maps are designed in such a way that there is usually "slightly too much" boost pressure: soot reduction takes precedence over torque limitation, which can be clearly seen in the logs from full-throttle runs.

As a result, the turbocharger is working unnecessarily hard, meaning that the low-temperature heat (LLT), exhaust gas temperature (EGT), exhaust back pressure, and turbocharger speed are all unnecessarily high.

Wouldn't it be possible, as a DIY project, to generate an analog correction signal for the voltage of the boost pressure sensor by continuously logging data and comparing the soot level with the torque limit?

The higher the soot limit is above the torque limit, the further the pressure sensor signal is increased to reduce the pressure – ideally to the point where the soot limit equals the torque limit.

To prevent oscillations in the boost pressure control caused by pulsating acceleration and to avoid overloading the turbocharger in the event of a faulty mass airflow sensor (MAF), I believe the circuit should be designed in such a way that...

-> the pressure sensor signal can only be distorted in the upward direction.

-> The response to the difference between soot limit and torque limit is slow.
-> This results in strong damping, so that the circuit essentially only operates in a steady state, for example, at a largely constant speed.


Do you think something like that is feasible?

However, the circuit design, if any, would mostly be created by the digital experts .

Hi Ulf,

I am consistently impressed by your input.

The more I read your posts, the more I wonder (and I'm asking you this directly), why you didn't study automotive engineering and work at VW in engine development.

I believe that people like that should be in the right positions within companies, and things will eventually improve.

Greetings.

Eike.

PS: I believe your idea is feasible. It aligns with the 'Lambda=1 regulation' approach, which requires a highly homogeneous combustion process. Since this is not yet achieved in diesel engines, it's likely that we'll operate with a significant excess of air as a precaution, because NOx (which is invisible) is generally more tolerable than soot (which is visible).
To extract the last bit of efficiency from the engine, I tend to favor solutions like water-diesel mixtures or similar approaches.

eike wrote:

The more I read your posts, the more often I wonder (and I'm asking you this now), why you didn't study automotive engineering and work at VW in engine development.

Well, life sometimes plays out differently... but I'm glad that for my hobby, I didn't have to endure the bloated, theoretical nonsense of a formal education, which you only end up using in tiny fragments later on .

Quote:

It seems to be heading in the direction of a "Lambda=1 regulation".

No... the target air mass value or the soot limit still includes the excess air that is structurally determined or intentionally specified according to the characteristic curve.

I only want to achieve that the blower doesn't practically constantly and draw in more air than is needed for the intended excess of air.

@ulf

Where should the target values come from? From the sensor? I think it will be too slow if it comes through the K-line. According to the specifications, 0.2 seconds can elapse for a request cycle... I assume that at least two reading cycles from the sensor are needed.

It might be even better to implement custom sensor readouts within the microcontroller and read them using various parameters. Then, replace the actual pressure sensor with a digital potentiometer controlled by the microcontroller... Alternatively, the pressure sensor could be represented using a D/A converter.

Here are my random thoughts on the matter.

If you know how, you can also read the measurement values in real-time directly from the control unit, but this requires detailed knowledge, and the solution would be different for each engine (control unit) or easily modifiable.

joergs wrote:

Where should the target values come from? From the sensor? I think it will be too slow if it comes through the K-line. According to the specification, 0.2 seconds can elapse for a request cycle... I assume that at least two reading cycles from the sensor are needed...

A delay of even just 1 second wouldn't fundamentally contradict my idea.
I already see the risk of overcorrection or similar problems with a "quick" solution, and that's not the outcome we want.

Quote:

It might be even more ingenious to implement custom sensing fields within the microcontroller itself and read them out using various parameters. Then, replace the actual pressure sensor with a digital potentiometer controlled by the microcontroller...

Alternatively, perhaps the pressure sensor could be represented using a D/A converter...

"In this process, I see the risk that 'the more calculations are used instead of measurements, the more unstable the system becomes.'"
For example, the emergency protection function in case of excessive boost pressure: how do you want to reliably trigger it if not through the analog sensor?

Where did you intend to obtain those specific parameters from? Were you planning to extract them digitally from the engine control unit, or by tapping into sensors?

Hi Ulf,

I've had that idea before, but I don't want to do any more programming.

With AFN, it's even simpler.

Regardless of the speed, each (analog) control value requires exactly one boost pressure value. I would receive both values without any delay.

A similar process, which measures the injection duration in a gasoline direct injection (GDI) system and calculates the injection quantity based on that measurement, can also be applied to GDI engines.

But to fine-tune something like that, you need a test bench. And from that point on, it would be easier to reduce the boost pressure curves by 20%.
That just leaves the information about the structure of the characteristic curves to be added, and then there's a lack of time, and even more stress...

You could also use a linear lambda sensor to determine the residual oxygen content in the exhaust. (So, run the diesel engine with a lambda value of 1. Then it will definitely produce soot!)

While on vacation, I came up with another purely analog solution, which I had actually seen once as a lambda control in a K-Jetronic control unit.

Monitor the accelerator pedal position and injection duration or duty cycle in the PD system for changes.

2. If nothing changes, or if the change is very slow, the boost pressure sensor value will slowly increase -> the control system will then reduce the pressure.

3. . . . until the soot limit is reached and the injection duration is reduced.
This adjustment of the air-fuel ratio or the RWG voltage is translated into a reversal of the simulated boost pressure, which slightly increases the pressure and allows the cycle to restart at step 2.

Ideally, the result would be a slightly fluctuating boost pressure and minimal torque fluctuations, hopefully not noticeable enough to be disruptive.

To avoid affecting the maximum power output (Pmax), the "control" could be deactivated, for example, via the full-load contact.

The question is whether the effort would be worthwhile in practice.

A significantly simpler solution would be a continuously adjustable "fake" boost pressure voltage setting (achieved via a potentiometer) that increases the voltage, thereby reducing the actual boost pressure, so that the soot reduction system is only just barely activated, even during the hottest summer months.

Ideally, the result would be a slightly fluctuating boost pressure and minimal torque fluctuations, hopefully not noticeable enough to be disruptive.


You can tell.
'I experience a similar problem when I'm driving with the cruise control engaged, and the engine speed is around 1500-1600 RPM. That's when the turbocharger is in the range where it's building boost.'
'The transmission causes the engine to easily slip. The speed decreases slightly, the cruise control slightly increases the throttle, the turbocharger provides boost (torque), the cruise control then reduces the throttle, the turbocharger pressure decreases, and the speed decreases... This creates a self-reinforcing cycle, leading to a 'throttle lag' effect. With the accelerator pedal, there's no problem because the fuel injection remains constant. With the cruise control, it seems like the load compensation system, which normally dampens such effects, is not working.'

The charger, as you know, draws a significant amount of power or presents a considerable resistance. Go to measurement block 11 and adjust the basic settings. Keep the motor speed around 1300 and press 'GO.' It puts quite a strain on the system when the VTG (variable turbine geometry) is activated...
'In my opinion, if the boost pressure control system is fluctuating, it would definitely be noticeable while cruising.'

CU Gremlin.

Gremlin wrote:

This would probably become noticeably unpleasant while cruising if the boost pressure control is oscillating...

Hmm, so, would it be better to have an unregulated fake with a potentiometer inside?
Reduce the pressure until the performance decreases or the "camel driver" (someone who pushes for maximum output) intervenes, then increase it again until the symptoms just disappear.

Repeat the entire process depending on the season/significant changes in outdoor temperature, and that's it... or what do you think?