Dutycycle table suggestion
After trying to tune the current (1.0.36) firmware PID BoostControl, I realized that it's hard to tune.
Tried different methods and PID settings, but never got a really satisfying bahaviour => write on your page:
- how you verified the valves and actuator are set up OK
- what PID (an uptodate mcd,mct to start with, and the b table, t table, P, I, D and ilimit for every experiment)
- what was behaviour
- and how it was changed. The [logs] are useless without the PID settings that were used for the given log.
Because every BIG ECU brands have openloop Turbo boost control, I suggest implementing this 'ASAP' to ours.
A 6X6 or 8X8 RPM / TPS or MAP (selectable) table would be fine, with loggable Dutycycle.
Such an "openloop" (it is not openloop, because the MAP changes and it provides feedback) table is very similar in effect as using only the P component (with I=0 and D=0) if the "refpos" is set right (20, 40, 60 or 80%, whatever gives closest balance for the boost-target) for the given setup.
- uint8_t boost_conf; // boostcontrol bit7:4 PWM period [8msec]; bit3:bit2 refpos; bit0 0:boosttarget,1:MAPtarget; 00=disabled
- Ok,i dont understand this,how to set in MegaTune,which variable is for what etc,etc....
We definitely need a "force boost solenoid duty" command to make boost system measurements (and tuning) easier. In the meantime the idle can be configured as PWM solenoid to drive the boost valve, so duty can be forced:
- mdi01 (1 / 256 almost 0 duty)
- mdi13 (19 / 256 duty)
- mdi40 (64 / 256 = 25% duty)
After doing (logged!) runs (which gear ?) with forced duty increased a little every time, an analysis (result can be like: +10% duty raised max reached boost by 20kPa or similar) can help to get good initial values for P (and likely I and D as well).
Overview of boostcontrol
- Configuration: GenBoard/Manual/Config/BoostControl
- Testing: GenBoard/Manual/Testing/BoostControl
Typical Pneumatic connections
- For both installs, activating the solenoid raises resulting boost. We need to control PWM-duty to get to boost target. Raised duty = raised pressure.
- An install where raised solenoid duty is needed to (open wastegate and) lower boost is just dangerous. Firmware supports it, of course, but avoid it.
- For an unpowered solenoid, the wastegate should open with relatively low boost.
- therefore solenoid is usually
- Since most often a 3-way valve is used, words like "opened" and "closed" can be misleading (even if 1-nipple left open to athmosphere). Avoid them. Prefer the words "powered" (or raised duty) solenoid or "unpowered" (or lowered duty)
Boost control system layout:
Primary system for internal wastgate:
- adjustable leak valve between the turbo compressor housing (or after IC to compensate for pressure drop over IC.) and node A
- Bottom chamber of wastegate connected to node A
- Solenoid from node A that in it's rest position is blocked and in it's activated position is open to atmosphere
Primary system for external wastgate:
- Wastegate bottom chamber connected to compressor housing (or after intercooler to compensate for pressure drop over intercooler)
- Solenoid from the top chamber of the wastegate, switched to either atmosphere (rest state) or directly to compressor housing. Or to bottom chamber if you are lazy and don't want long pressure lines and don't plan on ever using the extended range system below.
Improved spoolup and range for external wastegate:
- Primary system for internal wastegate connected to the bottom chamber.
- Primary system for external wastegate connected to the top chamber.
The systems are sequenced, first pressure is applied to the top chamber but if that is not enough (remember EBP pushing on the wastegate valve) the boost control working on the lower chamber will start to bleed off boost. In addition to this the system acting on the bottom chamber is always bleeding off as much boost as possible until the system is getting over the adjustable 'spoolup' threshold.
Note that the atmosphere connection can be replaced with a vacuum connection. A vacuum reservoir will NOT work, it has to be a fairly high capacity vacuum pump. (If a zero break pressure check valve is found and used (never seen one) to atmosphere a vaccum reservoir can be used.)
-Jörgen
Audi external wastegate example - under investigation, no final word on this yet; Actually, text below might be bad ! Please review
Connection
- big pipe on the bottom chamber (closest to the valve) connected directly to the compressor (or after IC)
- small nipple on the top chamber connected to the compressor with a solenoid valve in series (I doubt that this is good).
Is this (trying to be) similar as the "boostvalveexternal" example connection ? - Yes, I was wondering if this was the best solution.
Operation - must be verified against the connection
- When the solenoid is powered (routing pressure to the top chamber) the pressure in both chambers should be equal and then it's just the spring acting on the WG valve, and the stock spring should be able to hold the valve completely shut so boost can build up as fast as possible
- to open the waste the (unpowered) solenoid vents the top chamber to atmosphere, and the wastegate would act as a regular WG.
- solenoid is PWM-ed to keep boost at configured level.
Alternative:
The alternative would be to put the solenoid in between the compressor and the bottom chamber (big pipe), like a regular boost control and let the top chamber vent to atmosphere.
Pros:
Solenoid would only be powered when boost needs to be regulated which is better because with the other configuration the valve would be powered constantly at idle and cruising, not very good.(unless there are different types of solenoid valves? i.e ones that would vent to atmosphere when powered)
Possible Cons:
Very slight loss in performance due to bleeding off boost, the other configuration would theoretically not bleed any boost(not bleeding off boost could cause a problem with the wastegate returning to closed state). Possibly slightly slower actuation of the WG valve due to restriction in the solenoid valve.
After some thinking it seems to me like the regular boost controller style setup would be the best choice. Except I don't get this:
Primary system for external wastgate:
- Wastegate bottom chamber connected to compressor housing (or after intercooler to compensate for pressure drop over intercooler)
- Solenoid from the top chamber of the wastegate, switched to either atmosphere (rest state) or directly to compressor housing. Or to bottom chamber if you are lazy and don't want long pressure lines and don't plan on ever using the extended range system below.
What would make this configuration spool up faster and have an extended range? (this is the config which the above section is about, not the alternative here) And from your text Jörgen it seems to me like the valve would vent to atmosphere in the unpowered(rest) state, contrary to Marcell's correction above.
And just to make clear, the bottom chamber (closest to the valve) is the one where boost would be applied to open the WG valve, top is the chamber with the spring in it and a higher pressure in this chamber would only keep the WG shut harder.
Honda VTEC / Toyota VVTI / BMW Vanos / etc... control
(the more complex type) is very similar to boost-control:
- boost-target (actuator position/phase) dependent of RPM
- actuated by PWM signal (any solenoid valve)
- feedback from a (usually analog?) signal (choose a free analog input and select boost-target, NOT MAP-target)
The simpler/older "single Vanos" basically only requires RPM dependent on-off output, no feedback and no PWM => can be implemented with misc1/misc2/wot or rpm actuator.
It seems possible to control the complex version with proper configuration of the boost-control code (or small firmware mods?) but this has yet to be proven.
Does the BMW double Vanos have feedback from the Vanos actuator (position) ?
See also
- GenBoard/Manual/Config/BoostControl - config variables
- GenBoard/Manual
- GenBoard/Manual/Config
- GenBoard/Manual/PneumaticTube
- InTake/VariableIntake (same type of small solenoid air valve)
- BoostController is the old, developer and brainstorming page