john banks

The sprung internal wastegate controlled by a restricted manifold source with a solenoid bleed is subject to boost offsets in different gears, environmental conditions or changes to breathing. I am trying to think this through and explain it a bit better to work out strategies to avoid it.

My present thinking is that the system will reach equilibrium when the flow of air bypassing the exhaust housing is sufficient to stop the turbine speed increasing. One may assume that there is a linear response at the wastegate actuator perhaps, but the wastegate itself is a side hinged circular plate and the flow through this will not change linearly as the actuator moves, hence the offsets?

Just my theory, please add to it/criticise, as I have not seen the reasoning behind offsets on the net or in the literature.

The JECS closed loop boost control is also subject to offsets, because it uses a maximum duty cycle and aggressively ramps up to this when below target, and the main control of the peak seems to be the actual maximum value you supply. Perhaps the aggressiveness could be damped down (losing ultimate response?) to find a nice compromise between the value in the boost map and the duty cycle? At present it is far biased to the latter.

Pavlo

I had a solution in my head, relating to using the dawes in conjuction with a 3 port solenoid.

Feed from compresser goes to solenoid in.

Solenoid Output 1 goes to dawes valve input.
Dawes output goes to actuator.
Solenoid Output 2 goes direct to actuator.

At low loadinput on the boost map, you run normal dury cycles, beyond a certain point, you run 100%, effectively giving boost control to the dawes.

You may need to add a bleed/check valve somewhere to allow actuator to return, haven't got that far.

At high load, ie full throttle, you effectivel have a dawes device, so you can tailor your duty cycles to be more progressive at lower loads.

This sees okay to me, but I might be nuts.

Paul

Pavlo

Does anyone actually understand what i described here.

Paul

john banks

It sounds good, but it also feels like we would benefit from a more linear response wastegate to solve the problem, or somehow get a closed loop boost control that does not have offsets, but is also stable. Some of the EBCs would have a role hear, but none seem to have all the features you want - ie RPM, TPS aware, plus a strong integral setup that doesn't make them oscillate, AND THEN retain ECU control of boost.

Would an external wastegate be easier to control?

Pavlo

Dunno about external.

However, I did think more about my idea, and realised that as soon as you have anything but 100% solenoid duty cycle, the ecu has complete control of boost.

So you have leave the solenoid wide open from 0-6000rpm at full load, and then progresively feed it in from 6000rpm onwards, this would allow you to use the ECU to taper off the boost.

Relying on the dawes at full load during spool up should prevent boost overshoot.
Use the ECU at lower loads to give smoother more progressive boost changes, to give better on-track feel.

Paul

Andy.F

The key word you mention is equilibrium. Through the lower gears the system never achieves equilibrium, the turbo is always playing catch up. Hit 5th or a big hill and the revs stabilise and you start overboosting.
It needs to be closed loop control. Porting the wastegate orifice gives more progressive control due to less turbulence but you still get 'load dependent' offsets due to the different engine/turbo acceleration rates.

Andy

Pavlo

Exactly, make the ecu use agressive duty cycles at low rpm to prevent wastegate creep and help spool up. At higher RPM the dawes would limit the overboost you would undoubtadly get from running high duty cycles during spool up.

Also, I have seen a number of mods involving using a larger wastegate paddle, and enlarging the hole.

Paul

john banks

Steve Done had a setup using rate of change of RPM as an input to a boost compensation map which gave him clutch slip. Sounds ideal I will investigate more.

Pavlo

Sounds like he used a bit of spare ram space to add some assembly code to do this.

Time to learn Mitsubishi assembly language.

Paul

BruceWarne

Pavlo, I had the exact setup you describe...it eliminated the overshoots, but caused boost flutter...