Alan 91 C2

The SD concept only requires four pieces of data to "determine the transform function" the VE for the engine.

When you know RPM, MAP, AIT, and O2, you have defined the thermal performance of the engine. This approach requires more processing power than was available in our original DME.

To look at the problem in an alternate way, an engine is an air pump we throw fuel into. So the MAF approach measures the airflow directly, and adds pounds of fuel, with allowance for RPM to richen the AFR.

SD, does the same analysis. By knowing the MAP, O2, and AIT, we know the pounds of air, feed fuel and adjust AFR for RPM. The SD approach breaks the engine operation into multi-dimensional cells within a computer map, essentially a reverse look at what the MAF does.

The difference is SD learns flow from sensors, and MAF measures flow.

The equivalent in the way we teach children math. For example, the math problem 4x5=20, can be viewed as 5+5+5+5=20. Where the first solution is the MAF equivalent and the second solution is SD.

Note I have omitted knock, fuel octane, and others as they apply to both MAF and SD.

V2Rocket

Thank you for breaking this down.

Confirm this please:
1) You have to tell the SD the engine/cylinder displacement in the initial programming.
2) RPM x displacement tells the ECU the total airflow
3) MAP + IAT tells the ECU the air mass going in
4) O2 sensor feedback tells the ECU if too lean or too rich, based on MAP+IAT+RPM
5) VE is figured using O2 feedback? (MAP+IAT says 14700 units of air should be going in at a given RPM, ECU commands 1000 units of fuel, but O2 later says that was too rich, which means that VE at that combination of factors is less than 100% since there is not enough air for stoich?)

Alan 91 C2

Hi Spencer,

As always, the Devil is in the details. There is a starting VE table for the engine.

A good site to view is
http://support.moates.net/theory-speed-density/

Recognize the the discussion of air density, as a problem, is manageable for our cars as the real power is made under boost. And any car with a MAP and AIT sensor does have density information to the computer. Density is defined by the gas (our air mix 21% oxygen and 78% nitrogen plus other gases) temperature and pressure.

Each ECU vendor may look at the engine setup differently. And how table values are established. So this is somewhat like any database system, think accounting, cash or accrual basis, both work.

We have omitted discussion of the Throttle Position sensor (TPS) and assumed the discussion was for wide open throttle. Off throttle engine fuel demand are handled separately.

Confirm this please:
1) You have to tell the SD the engine/cylinder displacement in the initial programming. -- The displacement is factored into the setup tables.
2) RPM x displacement tells the ECU the total airflow -- Provides a reference mass flow corrected by fuel in and O2 out in a table
3) MAP + IAT tells the ECU the air mass going in -- Provides a reference when viewed against RPM, fuel, and O2
4) O2 sensor feedback tells the ECU if too lean or too rich, based on MAP+IAT+RPM -- Yes
5) VE is figured using O2 feedback? (MAP+IAT says 14700 units of air should be going in at a given RPM, ECU commands 1000 units of fuel, but O2 later says that was too rich, which means that VE at that combination of factors is less than 100% since there is not enough air for stoich?) -- Yes, recognize we do not actually calculate mass flow for air (although we have the info). What we want to determine is the correct fuel flow for a given RPM, MAP, and AIT, as verified by the O2 values.

The known safety factors for AFR are used to make the combustion products somewhat rich, at max power, 12.5-11 to protect the engine components. What we know from testing is the max power at 14.7 is risky for EGT, so we sacrifice a few percent power to save our engine components, at lower EGT.

odonnell

Here's the actual equations the MS series of controllers use. It's basically implemented SD theory, based on the ideal gas law. It works quite well and any "error" is compensated for when you tune your enrichments and VE table.

http://www.megamanual.com/v22manual/mfuel.htm

In short, your engine displacement/number of cylinders/injector size/fuel type/other physical parameters don't directly play into the calculation of the VE table. Instead, you calculate a number called the "required fuel." In MS this is abbreviated "req_fuel" and can also be referred to as "base injection time." It's usually between 2-20 ms. It represents that at 100% volumetric efficiency, your engine needs each injector to fire for X milliseconds per cycle to achieve a stoichiometric AFR (dependent on what fuel you run).

So the VE table, while Alan is right, can also be looked at as a multiplier table. MS looks at where you are on the VE table, of course a function of RPM and MAP, and looks at the programmed value. That's treated like a percentage and multiplied by req_fuel. So if your req_fuel is 10ms and your VE is 75, that means that engine can at best, only use 75% of that fuel since it can only flow 75% of it's displacement right now.

In real life it won't be 7.5ms, because MS modifies the req_fuel figure with other parameters as well. You have to factor in the air temperature/density correction, the injector dead time, and other enrichments if applicable (cold start, acceleration etc). The MAP is also factored in, as a percentage of the MAP reading divided by barometric pressure. The final equation is:

PW = REQ_FUEL * VE * MAP * E + accel + Injector_open_time

The "E" above is the multiplied result of all enrichments, like warm-up, after-start, barometer and air temperature correction, closed-loop, etc.

Example: MAP=40 kPa, VE=74%, GammaE=97%, and constants: Req_Fuel = 10ms and Inj Open Time = 1.3 ms.

so: PW = 10 * 40/100 * 74/100 * 97/100 + 1.3 = 4.17 ms in this case.

With regard to O2 correction on the VE table, I assume you're asking in reference to self-tuning. I'm not positive how it works, but if you understand the above math, you'll see that VE table can be adjusted to directly control fuel pulsewidth. What VE analyze software does is look at a programmed AFR table, and simply adjust the VE table until it's in line with that goal. So let's say you want 11.5 AFR in boost. Cool just set your table, warm up the car until enrichments are minimal, and hit auto-tune. The wideband obviously has to be in contact with MS for all that.

Alan 91 C2

Thanks Michael,

The key take away is that we have the computation power, in the ECUs of today, and the sensor technologies, to do this analysis on the fly.

And record all the data for review after a power run.

Jay Wellwood

Great topic and discussion.

Alan 91 C2

i knew this topic would stir up the as-is responses and the as it should be team.

We need these engine to tell us what is the issue, rather than the begin to replace parts approach.

These are great fun to drive, and we want it simple.

Alan 91 C2

As I look at the ECU systems out there, there are two discriminators; Do they offer Plug and Play (PnP) for our cars, or Do we have to develop the configurations.
To help those considering an alternate solution for ECU, I would like to compile a list of systems, note if PnP, and who can offer help with installed systems.

The list of systems to date is: (please cut the existing list into your message with ECU provider additions, list is alphabetical)

Emtron engine management
Megasquirt and MSDroid
Rogue
TECs
VEMS.hu
Wolf EMS

JBrown

Electromotive tech ECU

odonnell

MS has a PnP kit for the 944S. http://www.diyautotune.com/diypnp/ap...8-4cyl-mt.html

If the S2 and/or 968 has the same DME pin arrangement as the S, it would work too with tune changed up.

V2Rocket

In a week or two I will be placing an order for VEMS PnP.

Alan 91 C2

The list of systems to date is: (please cut the existing list into your message with ECU provider additions, list is alphabetical)

Electromotive Tech,
Emtron engine management
Megasquirt and MSDroid, PnP for 944s
Rogue
TECs
VEMS.hu, PNP for 951
Wolf EMS

Cyril

Link...

Alan 91 C2

Cyril,

I was unable to determine if LP Motorsports is developing a PnP version on the Link ECU for the 951. The wiring layout table looked like a prototype for our new harness. Can you provide additional info on the LP Motorsports work? The 951 in the pictures is very nice work. Are there more pictures of the car? Engine details and dyno runs?


The list of systems to date is: (please cut the existing list into your message with ECU provider additions, list is alphabetical)

Electromotive Tech,
Emtron engine management
Link ECU
Megasquirt and MSDroid, PnP for 944s
Rogue
TECs
VEMS.hu, PNP for 951
Wolf EMS

Thom

I love the Link G4/Vi-PEC ECUs. Easy to use, very flexible. Have had great success using them on Porsche, BMW and Audi.

Here is a log from this morning, FWIW.