Hello everyone,

I mainly do code for the older H8 ecus(DSMs, Evo5/6, Mirage,etc.)
I have made great progress with my set of Mods, with help of people from this great forum.

One of the things the H8s still lack is ECU Boost Control.

I have started to look over the Evo8 and 9 Boost Control Tables and Routines.
Everything seems overly complex for what is being done.


I was wondering if someone could explain the benefits of the Evo Boost Control vs a simpler system? Or has a good info on how the Evo Boost Control works?


The simpler system I was thinking of implementing, would have a single table of Desired Boost vs RPM. WasteGate Duty would be 100% until a few psi before Desired Boost then taper down to 0%. Does anyone see any faults in this?


Thanks for any help, I haven't been able to much real info on Ecu/Electronic Boost Control Theory.

 

I don't know much about it... but if you want some reading material, I think you can look up closed loop control systems and that will help you understand what the stock ECU is doing.

 

Yeah if you do it like that, then the waste gate wont respond fast enough and you will over boost. The ecu then will probably pull too much WGDC and drop the boost below target. It may finally stabilize out after a few oscillations. The base WGDC and the WGDC correction table per unit of load/boost is a more controlled system.

 

Thanks for info. Updated my definitions and got better table descriptions, much clearer now. All the extra tables had me confused at first.


Seems the only important tables are:

Error Correction - Prevents Oscillations/Feedback, Better Control.

Desired Load - What you want Load/Boost.

Base WGDC - Prevents OverBoost that Error Correction couldn't control at certain rpms.



Is the Error Correction cumulative over time?
Like if 10% Boost Error = 15% WGDC Change.
The first cycle would be 15%.
And if that didn't fix anything, the next cycle would be 30%.
45%, 60%...

Is that how it works?

And WGDC Correction Interval is the time between cycles?

 

Exactly!!! But the Wastegate correction is like in half percents from +5 to -5 for a load range of about 10 to -10. Just download a 94170015 rom of the likes and use ecuflash to check out the default values.

 

one thing ive noticed about the TBEC is that it always pulls the max WGDC drop when it has any overboost. I havent been able to get it to run any other perctentage than the one called for in that specific table.
...just wish i could remember the table name.

 

I'd shoot for a simple PI controller. It is important to have the integral portion if you want consistent boost pressure. I'd avoid a derivative based system, as they tend to be trouble for tuning and give little benefit over a well tuned PI system. You do need to watch out for integrator "wind up" though if you pursue this method.

This is pretty much what the old school Greddy Profec B was.

I would imagine the "cheddar cheese" knob was just an amplifier gain and the boost set knob was just a potentiometer to set the reference voltage.

FWIW though, the stock system has parameters for a lot of fail safes and corrections. The stock boost control seems to be based on the idea of keeping power delivery consistent, not boost pressure as well.

 

The Evo system with a cumulative Error Correction is a PI system?

A derivative based system would take into account the change of error rate?

Control Engineering stuff is new to me.


The stock Evo system isn't so complicated once you get rid of the ten or so non-important maps. I think I will be able to code something similar for the H8s. The H8s do have a pseudo ecu boost control already in place, but it only changes the boost level a couple psi and doesn't have a real error correction system, but it should be upgradeable.

 

I'm not real familiar with the workings of the EVO system, but yeah, it seems to be a form of a PI controller.

It's actually a little more sophisticated, really. The baseline wastegate duty cycle table would be analogous to the proportional gain, if you put the whole table at one value. However, it allows it to be RPM dependent, thus adding an extra dimension to the system.

The target boost tables set up your actual feedback loop targets, which the error correction tables then form a non-linear integral correction table. A normal integral system would just have one value that defined how fast it would correct for any errors and you would have to tune that single value to be the best compromise between over shoot and oscillation. You can change how it deals with the error in the EVO ECU dependent on the amount of error and you can tune for overshoot and oscillation independently with it.

In essence, it seems to be a PI controller, in function, it's better.

 

dunno if it makes any difference but theres only a single boost solenoid on evo5/6 ,also uses a restrictor in the boost line

 

The Evo 7-9 boost control system is pretty simple in concept. It sort of blends together proportional and integral control, so it is not even a PID controller. The system uses baseline WGDC curve (set by the tuner) that gets the boost close to the desired boost level without any feedback. This is nothing like how a PID system works. How close the boost is to the target boost depends somewhat on the level of effort put in by the person tuning the system. Weather conditions will have an effect, and I'll mention more about this in a moment. The only feedback mechanism is the TBEC table. The code periodically calculates a boost error which is the difference between the target boost and a spot measurement of the actual boost. It then uses the TBEC table to determine a WGDC adjustment. The adjustment moves the entire baseline WGDC curve up or down. This has features of both proportional and integral control. A true proportional curve wouldn't move the entire curve up or down, and a true integral control would integrate the boost error over a period of time rather than making a spot measurement. This system is much simpler to implement than a true PI control. There are also some tables that limit the WGDC value based on TPS, coolant temp, and engine load.

One feature that is lacking from Mitsu's system is a compensation for air temperature. Cooler air makes more boost for a given WGDC, and while the TBEC table can correct for this, it is not very graceful at doing it. What tephra has done in his v7 ROM is to add a WGDC correction based on IAT. Cooler air pulls down the WGDC by a set amount.

 

Thanks much for the info.


Does anyone know the stock systems Correction Interval units?

Would 10 units be, one correction every 10 loops?

Or does it equal a unit of time, if so what does one unit equal approximately?

 

Mrfred, doesn't the factory boost control system use compensated load as the target?

At least the way I understand the factory system, it is load based and the load used is an IAT and baro compensated load. Thus the system compensates for IAT indirectly through the load parameter?

 

We're getting a bit more into the details of boost control. Yes, the factory boost control system targets baro+air temp compensated load. It does provide a compensation for IAT, however the effect is to push the turbo to run higher boost in hot weather and less boost in cool weather. It does this to try to hit the target baro+air temp compensated load which, for a given boost level, gets deflated in hot weather and gets inflated in cool weather. Its not a desirable response, at least in hot weather.

It is preferable to use boost or uncompensated load as the control parameter because there will be no drive to run higher boost in hot weather. That is one of the advantages to switching over to direct boost control.

Even when using boost or uncompensated load, there is still a need to tweak WGDC as a function of air temperature in order to hit the target boost. tephra's WDGC correction vs IAT is a non-feedback system that effectively anticipates air temperature effects and preemptively adjusts the WGDC curve without waiting for a boost (or load error) to occur.

For those that want to run less boost in hot weather, the way to accomplish this is to limit "max total upward WGDC correction" table to a small value (or zero).

 

Thanks for the explanation. Makes sense. The Factory system likely is just trying to keep power constant and in stock power levels, there is some additional over head it can tap into to keep power up at higher temps without hurting anything.

Start trying to run 25+ PSI though and all that reserve capability goes out the window.