Injector size, injector placement and dual injectors

[Belgian1979]

I'm not convinced it will play such a big role. Evaporation of fuel takes time. I do not see that happen sufficiently enough when the air is injected when the cylinder inhales. That means that a lot of the fuel has to evaporate from the heat created during compression.
If too much fuel is present in the runner because of a batch fire system, the ecu will correct based on ego.

I can see the advantage of a sequential system on the level of the ignition with cop, where you have a fully charged coil available at high rpm on the one hand and can do individual trimming of the spark based on knock detected in a specific number of degrees (combustion stroke) but even with an ms system the extra cost associated with such a move is prohibitive in my opinion at least when having to make the move from my ms2 setup on itb's (where the airflow is balanced on each cyl seperate to a pretty close tolerance)

[Belgian1979]

I have used MS2E with some modified code to run dual MAF sensors, so I KNOW that it can be done with MS3...the question is who and how they want to implement it.

I'm really looking at an MS3, instead of using MS2E with the Megashift to control an E4OD in a drag application...so many features I just don't need though.

Did you get them to modify the code for you ? And even then, how does TS react to it ? Currently I do not see ts running each bank on a seperate maf and seperate table. It would be interesting though

[mk e]

And the processor is so fast when it does knock control it can isolate what cylinder it is and only retards/richens that cylinder. Now that's amazing. And it's a true VE based system, not pulse width based.

Any OEM ECU can do individual cylinder knock control (including mine). It doesn't require a crazy processor to do.

Yeah, the OEMs ALL have pretty powerful processors these days. They're all RTOS, floating point, 32bit, multiple co-processor. It's amazing what they can do.....and how hard it is to touch ANYTHING on a modern car because everything goes though the ECU

[Belgian1979]

One of the many reasons a modern car will never reach the 'oldtimer' status

[900HP]

MSIII-pro is 1 micro-second resolution. The megasquirt is all of that and more. It is amazing.

Here are just some of the features.............. and at the price you will NEVER come close to the performance that this offers. I think you'd be hard pressed to beat it at ANY price level. The data logging alone (up to 88 channels) is unreal..................

Software features:

•1 microsecond injector pulse width resolution

•16 x 16 fuel and spark tables - can be reconfigured to function as 30 x 16 or 16 x 30 by switching tables based on RPM or load

No it's not. The new AEM will run circles around it;
http://www.aemelectronics.com/?q=produc ... finity-ecu
1/10th of a micro second.

And the processor is so fast when it does knock control it can isolate what cylinder it is and only retards/richens that cylinder. Now that's amazing. And it's a true VE based system, not pulse width based.

And don't be fooled by heresay. Locating the injector further down the runner is always better. Just ask Nascar, or research an article in RET magazine last year on EFI in Nascar and what they've found to work, and not to work.

AEM still can NOT do closed loop AFR per cylinder
AEM can NOT do variable cam timing on 4 channels (think Coyote motor here)
AEM data acquisition and logging still behind capablilities of MSIII-pro
AEM does not have the assignable PWM output controls the MSIII-pro has which are very useful.
To say that "locating the injector further down the runner is always better is a bit of a stretch. There is no such thing as always. You should also phrase it further UP the runner, as in farther away from the intake valve usually makes more power.

[mk e]

Data logging too many things at too high of a resolution is not useful for most "end-users" of EFI. An oscilloscope, etc... should be used if you really need to find something wrong. The software itself should be able to let you know if a sensor is bad. 10Hz logging on 100 ram variables is a lot of data. This has nothing to do with what the CPU is using to do it's calcs, though.

Exactly so. The high logging rates are used for suspension and ABS tuning, miss fire detection, time lost during shifting the fancy new dual clutch transmissions and stuff like that. The base motec only give you like a dozen preset logging channels ...that I think are at 100hz which is generally all anybody needs (the crank error detection is built into the firware so its separate I think). To get access to the high power logging you have to pay extra but then you can log not only sensor values but also log pretty much anything the ECU is calculating/doing which is pretty cool.....but as you say most people don't need that and I know I've never been willing to pay for it but a lot of racing teams love it form what I read.

Also, too high resolution for tables just makes more typing necessary, and more chance for big errors. Most engines don't have seriously large changes in VE or spark requirements. A 10x10 can cover a lot of engines, if the breakpoints are adjustable.

Making good hp doesn't ever very big tables 10/10, 12x12....but I always end up with a lot more rpm points by the time I'm happy with idle, off idle, starting, ect. so personally I find 24-32 rpm point is a pretty nice to have available. For load point 10-12 seems enough but 16 can help in cleaning up the low power stuff. For me anyway.

Also, most software will use interrupts to read the crank and control the coils and injectors, which can have some seriously high precision for control, but throw in a few multipliers, tack on an offset, a little sensor wiggle, and whammy - you get steps in the adjustments. But luckily, an engine doesn't need 0.01% control. 1% control is actually quite satisfactory. Closed loop is going to throw it all around anyway.

For sure on fuel 1% is plenty good unless you're an F1 team tracking fuel weight to the gram . I read that timing needs to be 0.1 degree max error to hit emissions but I'm not sure any more than .25 or so is needed for performance....Iv'e never been able to measure a hp change below that anyway.

[mk e]

I have used MS2E with some modified code to run dual MAF sensors, so I KNOW that it can be done with MS3...the question is who and how they want to implement it.

I'm really looking at an MS3, instead of using MS2E with the Megashift to control an E4OD in a drag application...so many features I just don't need though.

Did you get them to modify the code for you ? And even then, how does TS react to it ? Currently I do not see ts running each bank on a seperate maf and seperate table. It would be interesting though

Wouldn't it be easier to just run 2 ECUs if you want separate control bank to bank? We end up with 2 on V12s due to channel limits and just share the sensor and it works fine but take more time to tune. Motec allows 2 of the M800s to be linked master/slave via CAN so it still acts as 1 ECU for main tuning, traction control and such, but 2 M800s is an EXPENSIVE solution ($15kish)....I heard they created it for 1 of the OEM Le Mans teams.

[twl]

Quick question for you EFI experts, if you don't mind.
With a single injector located in the manifold, and pointing into the port, is it important for it to aim at the back of the valve, or is it okay for it to point generally down the port?

[mk e]

Quick question for you EFI experts, if you don't mind.
With a single injector located in the manifold, and pointing into the port, is it important for it to aim at the back of the valve, or is it okay for it to point generally down the port?

$5 says your quick question doesn't have a quick answer

When I do EFI conversions the goal is always to point the injector at the valve....but I'm almost never able to achieve that and it ends up pointing in the general direction of the valve and it works just fine. How much if any performance or fuel consumption rate am I losing? no idea really to be perfectly honest.

[nitro2]

I'm not convinced it will play such a big role. Evaporation of fuel takes time. I do not see that happen sufficiently enough when the air is injected when the cylinder inhales. That means that a lot of the fuel has to evaporate from the heat created during compression.
If too much fuel is present in the runner because of a batch fire system, the ecu will correct based on ego.

The "burr finish" thread discusses how slowly "wall fuel" travels, as well as the fact that simply having the "correct" AFR does not tell the whole story, not at all. Fuel vaporized by an external source of heat is a no-no for power, unless for some reason the engine can't adequately prep the AF mixture without it i.e. summer gas run in the winter without preheating the intake air.

[Ed Wright]

My GM engineer buddy says back of the valve. I'm sure there will be 47 differing opinions here.

[900HP]

We found that port speed and CSA affect how sensitive an engine is to injector placement as well. A port with even velocity and good floor speed will be less sensitive to injector placement and angle.

[Belgian1979]

Due to the intake constraints on my setup the injector points straight down into the intake, so 90° into the airstream. It runs pretty well, but I would have liked to see it different.

[PackardV8]

Due to the intake constraints on my setup the injector points straight down into the intake, so 90° into the airstream. It runs pretty well, but I would have liked to see it different.

Constraints are inherent in engine building. As said previously, TBI at the top of the intake is good, EFI into the runner is better, EFI onto the back of the valve better still, but DI in the cylinder at the spark plug is best.

As forever, "speed costs money. how fast can you afford to go?"

[twl]

Thanks for your answers, fellas!
I have the injector going generally down the direction of the port, maybe a little up toward the higher part of the port(long side), so I guess it's going to be okay.