DI flow rate adjustment

[jamesm]

I'm trying to adjust the DI flow rate to compensate for some fueling error on the DI side without having to hack up my MAF curve and consequently throw my now-perfectly-calibrated port injectors out of whack. It's a process refinement that I've been considering for a long time but was only recently inspired to do by @mad_sb's post on the subject where he used DI pressure targets to accomplish something similar. I want to go the route of adjusting the scalers (DI flow rate is effectively the DI equivalent of a port injector scaler, only rather than cc/min it's the amount of time the system takes to flow 1cc, so inverted and 3-dimensional basically) rather than adjusting the pressure as i view it as a more direct approach to the problem. Not that the other way is wrong, hell i may end up going back and doing it the other way, it's just my initial thought.

The issue i'm having is that I have no idea what the y axis of the map is. It's label is the always informative 'value'. any ideas?

here's a pic of the map itself:

[arghx7]

Two things. For the map you've shown, it looks like it could be Fuel pressure target vs some sensor voltage (rail pressure sensor voltage) vs some multiplier or correction factor. The rail pressure sensor is 0-5v right?

Second, could you explain further what you're looking to do? You want to change the injection mass by adjusting the pulsewidth, as opposed to adjusting the rail pressure? One thing to keep in mind is that the fuel spray in a direct injection system works against the cylinder pressure. The cylinder pressure varies with crank angle, load, valve timing, etc. That's why OEM controls for DI have some way, even if it's a very simplified correction, to account for instantaneous flow rate vs rail pressure vs cylinder pressure.

[jamesm]

Originally Posted by arghx7

Two things. For the map you've shown, it looks like it could be Fuel pressure target vs some sensor voltage (rail pressure sensor voltage) vs some multiplier or correction factor. The rail pressure sensor is 0-5v right?

Second, could you explain further what you're looking to do? You want to change the injection mass by adjusting the pulsewidth, as opposed to adjusting the rail pressure? One thing to keep in mind is that the fuel spray in a direct injection system works against the cylinder pressure. The cylinder pressure varies with crank angle, load, valve timing, etc. That's why OEM controls for DI have some way, even if it's a very simplified correction, to account for instantaneous flow rate vs rail pressure vs cylinder pressure.

The map values are described by ecutek as the time required to flow 1cc of fuel. I figured this would be a more direct way of calibrating the di with the goal of minimizing fueling error. The x axis is di fuel rail pressure, but I have no idea what the 0-5v y axis represents. It doesn't seem as though it would be the pressure sensor voltage or that would be somewhat redundant since the x axis is the scaled output of that same sensor, but I could definitely be wrong.

[mad_sb]

Originally Posted by jamesm

I'm trying to adjust the DI flow rate to compensate for some fueling error on the DI side without having to hack up my MAF curve and consequently throw my now-perfectly-calibrated port injectors out of whack. It's a process refinement that I've been considering for a long time but was only recently inspired to do by @mad_sb's post on the subject where he used DI pressure targets to accomplish something similar. I want to go the route of adjusting the scalers (DI flow rate is effectively the DI equivalent of a port injector scaler, only rather than cc/min it's the amount of time the system takes to flow 1cc, so inverted and 3-dimensional basically) rather than adjusting the pressure as i view it as a more direct approach to the problem. Not that the other way is wrong, hell i may end up going back and doing it the other way, it's just my initial thought.

The issue i'm having is that I have no idea what the y axis of the map is. It's label is the always informative 'value'. any ideas?

here's a pic of the map itself:

Looks like the Y axis is maf voltage. In the raw rom it is maf g/s rather than maf volts. Also, that is not the pressure multipler table that i have been using to scale DI. That table is the GDI flow rate table. I have not touched that one at all. The tables i use allow you to adjust flow rate vs rail pressure AND maf g/s so you can fine tune DI scaling to the maf curve you are using.

Target rail pressure is a separate table, I only used that to tweak a couple spots, mainly i was using the DI pressure multiplier table. See attached. Shown is the stock GDI Pressure Multiplier table.

[jamesm]

so i wonder then what are the advantages/drawbacks to using the gdi flow rate vs. gdi pressure multiplier? i would think that both are inputs to the same fueling algorithm, but i'm not sure which way to go. i've read a recommendation somewhere (can't remember where, maybe ecutek docs) that suggested using the gdi flow rate to compensate for fuel density in an e85 setup that doesn't use racerom, which is a big part of why i was leaning that direction. i'm curious to know if there's any practical difference and if so what that is.

i was also thinking that the y axis may be maf voltage, but it didn't seem to make sense that it's scaled all the way down to zero, since the maf scale isn't and you'd never see that in operation. thanks for confirming that it is (which i believe we can be confident of if the raw rom values are g/s). this will allow me to test and see how it works out using this map to make the corrections.

my car is down for the weekend getting some intercooler pipes powdercoated, but i'll test when i get it back together and report back.

[mad_sb]

Originally Posted by jamesm

so i wonder then what are the advantages/drawbacks to using the gdi flow rate vs. gdi pressure multiplier? i would think that both are inputs to the same fueling algorithm, but i'm not sure which way to go. i've read a recommendation somewhere (can't remember where, maybe ecutek docs) that suggested using the gdi flow rate to compensate for fuel density in an e85 setup that doesn't use racerom, which is a big part of why i was leaning that direction. i'm curious to know if there's any practical difference and if so what that is.

i was also thinking that the y axis may be maf voltage, but it didn't seem to make sense that it's scaled all the way down to zero, since the maf scale isn't and you'd never see that in operation. thanks for confirming that it is (which i believe we can be confident of if the raw rom values are g/s). this will allow me to test and see how it works out using this map to make the corrections.

my car is down for the weekend getting some intercooler pipes powdercoated, but i'll test when i get it back together and report back.

Let me double check the Y axis when i get home, as i was typing i noticed it was the flow rate table rather than the pressure multiplier table you were showing.. I will check after work for sure and let you know. I can also do some testing on the flow rate table since i am still dialing in my changes, i started over yesterday (forcing port only on the stock port injectors is a bit scary so i only ran them up to 3k).

[jamesm]

i just realized that i can do this with a custom map very easily. probably should've looked there first lol. thanks for all the help anyway! maybe someone will see this and need it some day

[SkullWorks]

semi related random question,

Is MAP sensor used in the algorithm that defines DI Rail Pressure? Or does it reference MAF G/S only?

I was contemplating how far off base the flow numbers likely are when you consider the difference in cylinder pressure after FI is considered. Also is there a map defined that relates Injection angle to Cylinder pressure? Or are people still not playing with the SOI angle?

[jamesm]

Originally Posted by SkullWorks

semi related random question,

Is MAP sensor used in the algorithm that defines DI Rail Pressure? Or does it reference MAF G/S only?

I was contemplating how far off base the flow numbers likely are when you consider the difference in cylinder pressure after FI is considered. Also is there a map defined that relates Injection angle to Cylinder pressure? Or are people still not playing with the SOI angle?

the di fuel quantity compensation defined in ecutek uses fuel rail pressure on the x axis and load on the y, so i presume it only uses maf g/sec, unless of course you're running sd and replacing the maf output with the sd calc, which would use the map sensor.

ecutek doesn't define anything related to SOI angle for the di, likely due to liability issues. someone could easily mess up and end up injecting into the spark event . being that these things aren't well understood yet, i'd bet that they didn't feel the need to give us that noose to hang ourselves with. romraider has it, and the maps are load vs. rpm vs. soi angle.

[arghx7]

Originally Posted by jamesm

the di fuel quantity compensation defined in ecutek uses fuel rail pressure on the x axis and load on the y, so i presume it only uses maf g/sec, unless of course you're running sd and replacing the maf output with the sd calc, which would use the map sensor.

ecutek doesn't define anything related to SOI angle for the di, likely due to liability issues. someone could easily mess up and end up injecting into the spark event . being that these things aren't well understood yet, i'd bet that they didn't feel the need to give us that noose to hang ourselves with. romraider has it, and the maps are load vs. rpm vs. soi angle.

I think the more likely scenario is that during initial reverse engineering of the ECU, ECUTek didn't know much about injection timing in a quantitative or numerical sense. Considering very few people understand PFI injection timing, and most of the speculation out there ranges from inaccurate to garbage, they probably didn't find the need or the value in locating the timing parameters for direct injection.

One thing somebody needs to figure out is if it is possible to record end of injection timing for PFI and DI on the FA20. Somebody also needs to find what the limits are for SOI, EOI, and/or injection duration in the software.

You also want to find out if the code is capable of two or three injections per stroke. Under about 3000-4000rpm you can reduce knock sensitivity and run a couple more degrees of spark with two injections. The stock tune doesn't seem to do it. Japanese DI engines in production today generally don't utilize multiple injections for this purpose as much as more established manufacturers with DI--I suspect it is due to limited knowledge and experience with it. See attached paper for some discussion of tradeoffs in injection timing and injection split ratio.

I'm curious as to why you specifically feel that spraying fuel near the time of the spark event is a big deal, outside of smoke & particulate emissions (which nobody here cares about). If you look at the more advanced spark ignition combustion systems out there in production (chiefly high end German engines), they are already doing this for catalyst heating or for low load stratified combustion. Homogeneous combustion usually has end-of-injection timing limits in software for emissions purposes, and also software limits for crank angle degrees between injection events.

Another thing to keep in mind about rail pressure is that under most conditions higher pressure is better for atomization/mixing and combustion. One of the reasons why the pressure ramps up to 20 MPa gradually, rather than "jumping" to it outside of a certain speed/load window, is to reduce friction and pumping work. There may also be some unknown durability issue with the fuel system, but I can assure you that not every rail pressure map looks like that in its general shape.

[jamesm]

Originally Posted by arghx7

I think the more likely scenario is that during initial reverse engineering of the ECU, ECUTek didn't know much about injection timing in a quantitative or numerical sense. Considering very few people understand PFI injection timing, and most of the speculation out there ranges from inaccurate to garbage, they probably didn't find the need or the value in locating the timing parameters for direct injection.

One thing somebody needs to figure out is if it is possible to record end of injection timing for PFI and DI on the FA20. Somebody also needs to find what the limits are for SOI, EOI, and/or injection duration in the software.

You also want to find out if the code is capable of two or three injections per stroke. Under about 3000-4000rpm you can reduce knock sensitivity and run a couple more degrees of spark with two injections. The stock tune doesn't seem to do it. Japanese DI engines in production today generally don't utilize multiple injections for this purpose as much as more established manufacturers with DI--I suspect it is due to limited knowledge and experience with it. See attached paper for some discussion of tradeoffs in injection timing and injection split ratio.

I'm curious as to why you specifically feel that spraying fuel near the time of the spark event is a big deal, outside of smoke & particulate emissions (which nobody here cares about). If you look at the more advanced spark ignition combustion systems out there in production (chiefly high end German engines), they are already doing this for catalyst heating or for low load stratified combustion. Most of the rest of the engines out there have end-of-injection timing limits for emissions purposes.

Another thing to keep in mind about rail pressure is that under most conditions higher pressure is better for atomization/mixing and combustion. One of the reasons why the pressure ramps up to 20 MPa gradually, rather than "jumping" to it outside of a certain speed/load window, is to reduce friction and pumping work. There may also be some unknown durability issue with the fuel system, but I can assure you that not every rail pressure map looks like that in its general shape.

As always... appreciate the detailed info. I've never really thought about it too much, just kind of a 'hey, you shouldn't do that' so i didn't type of thing. Ecutek docs list 1ml as the hard limit, citing the risk of injecting during the spark event. Evokes images of when you were a kid, shooting fire with a WD-40 can and holding the flame there too long , but that's just my first thought as to why it may be bad to do.

you can log port and di injection time with ecutek, then reverse back to the SOI maps to get an idea of when the injection is taking place, couldn't you?

EDIT: looks like ecutek just added logging of PI/DI angle

[arghx7]

You can roughly figure out EOI timing for the DI if you have SOI and duration, using a formula. It would be better to read it off the ECU though, because there may be some other thing/modifier going on in the background, and it would be nice to actually see the EOI hitting a limit at say 30 degrees BTDC. As for the port injection, well you can calculate the number of crank angle degrees duration based on the duty cycle (50% duty = 360 crank angle degrees, 100% duty = 720). But you need a reference point. The key is figuring out whether you are open or closed valve injection. If PFI duty cycle is large enough, you're going to have open valve injection anyway because there's not enough time to get all the fuel on to the back of the valves at a given PFI injector flow rate/size.

There are small benefits for engine output to be had from moving EOI around on port injection and SOI on direct injection, because the factory timing is set according to emissions constraints--mostly HC emissions, smoke emissions, and long oil life (oil dilution). I'm talking a little bit of volumetric efficiency or knock improvement, usually nothing drastic.

Getting back to the original point of this thread: at part load your rail pressure and direct injection timing windows aren't maxed out. You've got room to play around with rail pressure, injection timing, and injection duration. At full load it's more like PFI fuel tuning with the fuel pressure aspect, because you will be at max rail pressure anyway. That's when possible software SOI and EOI limits come into play.

[ztan]

Had a chance to look at the disassembly. A lot indirectly referenced.

X-axis is Fuel Pressure, FFF8936C is the RAM address for data used in the GDI Flow table

Y-axis looks like a Direct Injector Pulse Width value, FFF887E8 is the RAM address for data used in the GDI Flow table

Return gets stored in FFF887EC and logs very closely with FFF887F0 (I believe this is Direct IPW).

[jamesm]

I ended up using the do flow scaling table to do it. It works very well.