GDI firing angle values, looking for explanation

[Tor]

Did some excel graphs. Don't know how easy they are to interpret. My excel skills suck! Y axis 0 is TDC, negative values are exhaust stroke, positive intake. Lower/upper red line intake open and intake peak. Green exhaust close. Orange is SOI controlled by GDI, blue is EOI as a function of IPW from SOI.

Stock A01G, IPW taken from an old stock log:


OFT values, IPW from a current log but with 20/80 PI/DI split through mid rpm:


My cams and playing with GDI values. If only there were some clear guidelines to what is desirable to target:

[Kodename47]

You want to do the graph in 720 degrees, or at least the region that covers. That graph isn't that helpful. You also want spark timing as that's possibly far more relevant than the cams as you're not reliant on VE to draw the fuel in, you just want to avoid any escaping past the exhaust valves. That is unlikely to happen though.

Most of the papers I've read were on Toyota/Lexus GDI systems that have been linked on older threads.

[Tor]

Originally Posted by Kodename47

From memory you want the DI to inject as late as possible to help with thermal control in the cylinder, but early enough that there is a good mixture.

From a logical POV, wouldn't injecting early cool more? Stock they inject early at high rpm, even though there is still at lot of time/degs to ignition. That could of course be for mixture as you say.

Originally Posted by Kodename47

You want to do the graph in 720 degrees, or at least the region that covers. That graph isn't that helpful. You also want spark timing as that's possibly far more relevant than the cams as you're not reliant on VE to draw the fuel in, you just want to avoid any escaping past the exhaust valves. That is unlikely to happen though.

As said, my excel skills are not that good. It would be much better to have it overlayed on your cam lobe graph, with boxes like the screenshot I posted before, but that's way beyond my excel skills.

But if my formulas are correct the injection times are fairly short, even at high rpm (highest 144 deg). With GDI angle of 360+ in that area, I only need a range of +150 to see what's going on. So I'm not sure what you mean isn't covered? Maybe it's the way it represented? I would have preferred rpm on the Y-axis, but couldn't get excel to cooperate. Ignition is very far away in all cases which is why I didn't choose to include it because of the resolution. How do you mean I could improve the graph?

The cams are overlayed to see where is injected with regards to overlap and not injecting with a closed intake cam, which I suppose isn't particularly useful?

I think I am going to try this out and see what the effect is. I don't expect big surprises as it's not far off from what I run now, except the 290 at low rpm.



And the graph it would result in with ignition added:
The red lines from below are intake open, peak, and close - just for reference. Worth noticing, I run 50% port to 3200 rpm and 20% above to redline.

[Kodename47]

Originally Posted by Tor

From a logical POV, wouldn't injecting early cool more? Stock they inject early at high rpm, even though there is still at lot of time/degs to ignition. That could of course be for mixture as you say.

Think about it this way, if you inject early how much cylinder fill has occurred. Definitively it's less than later in the cycle. The DI's helps with charge temps due to the fuel injection pressure and expansion that happens, so to make the maximum effect for in cylinder cooling you fire as late as possible. It is believed that the D4-S uses single injection timing where as some cars utilise dual stage injection, where you would fire early in the cycle to get the mixture and then another as late as possible for thermal control. Let's not forget that this is why DI is used and why it is seen as beneficial over.

I did this spreadsheet some while back, the DI calculator might be what your after to picture things.

That 2nd graph above isn't quite what you want. You want 0 to be 0 spark advance, 360 is the start of the intake stroke, so say start at 540 on the upward stroke of the exhaust. The best way to depict it would be, in some order

- Ex Cam close
- In Cam open
- DI SOI
- In Cam Close
- DI EOI
- Spark

The top 3 will change order depending on the overlap and DI timing.

[Kodename47]

Perhaps it's easier to picture it like this:



That's not using any real comparative figures butt give you an idea how it all works as a cycle.

[Tor]

I'll give your tool a close look the next days. Thanks!

Originally Posted by Kodename47

Perhaps it's easier to picture it like this:



That's not using any real comparative figures butt give you an idea how it all works as a cycle.

That's just what I do, only vertical because I'm an excel jerk. And On the other axis is rpm, so I can see the change over rpm. The cams on my graph is not the lobes, but just points of open, peak, close at different rpm.

I'm better with photoshop than excel, but maybe this will make it more clear....

[Kodename47]

Originally Posted by Tor

That's just what I do, only vertical because I'm an excel jerk. And On the other axis is rpm, so I can see the change over rpm.

It was hard to read due to the colours and upside down to what I expected so I was just confused The RPM axis for WOT is a good idea, mine was just to demonstrate the cycle for a given RPM and load. However I would always set 0 as TDC 0 deg spark as it's a much preferred reference. So long as you don't have negative ignition timing then all the graph should be above 0 so flows down with time.

More things that I thought about to consider:
For the same AFR throughout the rev range the IPW will be the same if the engine load is the same. The same IPW covers a larger range in degrees of the cycle as the RPM increases, this is also countered with ignition timing that generally increases with RPM. So as RPMs increase, all conditions being the same, the SOI angle will have to be earlier. Then consider the fact that most make the car richer at higher RPM, this moves the SOI earlier again.

[Tor]

Originally Posted by Kodename47

It was hard to read due to the colours and upside down to what I expected so I was just confused

Again, apologies for my excel skills.

My point with the graph was to be able to visualize how different GDI moves EOI and if a pattern would emerge. Of course, it doesn't make it easier that my tune has changed injector ratios. Neither is it easy to see a pattern with the stock tune because of the changing injector ratios. Interestingly enough, the stock MY17 tune runs the same GDI as previous years, even though it has changed injector ratios too. And maybe there is not supposed to be a pattern at all.

The point of the modified table (which I flashed today) was to get a similar EOI across the rpm range. Perhaps it would make more sense to look at it in milliseconds than deg? Since, obviously, it's going to take a lot longer at low rpm to cover the same deg. On the other hand, maybe the movement of the piston also contributes to getting a good mixture. And hence, maybe deg is not a wrong reference after all?

The RPM axis for WOT is a good idea, mine was just to demonstrate the cycle for a given RPM and load. However I would always set 0 as TDC 0 deg spark as it's a much preferred reference. So long as you don't have negative ignition timing then all the graph should be above 0 so flows down with time.

Yes, that what I did with the boxes I painted on your graphs as well. But my way it's easier to see the whole table at once. I took your reference of 0 because I had gotten used to looking at the cam lobes. But I see 720 as 0 reference would make more sense with ignition and injection.

More things that I thought about to consider:
For the same AFR throughout the rev range the IPW will be the same if the engine load is the same. The same IPW covers a larger range in degrees of the cycle as the RPM increases, this is also countered with ignition timing that generally increases with RPM. So as RPMs increase, all conditions being the same, the SOI angle will have to be earlier. Then consider the fact that most make the car richer at higher RPM, this moves the SOI earlier again.

The question is, how early can we start to inject and if it would make sense to start the injection even with a still closed intake valve. I think I'll try that at high rpm so EOI will not get the later-snip at high rpm.

I flashed the table and put in through some stress tests to check for influence on knock. For low rpm: High gear/high load. And for high rpm: Autobahn WOT in 4th, 5th and 6th. This is probably not the right way to adjust these values, as far as I read they should be adjusted to get the richest AFR. But I'm curious to see if it's possible to reduce knock this way.

It was moderate successful. At low rpm, it held IAM pretty good, with some minor drops to 0.98. But quite a lot of FLKC, nothing too serious though. With Autobahn WOT pulls I had horrible FLKC in midrange rpm, but not that much at high rpm. Perhaps I advanced it too much in mid rpm? Or there is something else at play (high atmospheric pressure at the moment as an example and also my TCPC/ECT hack failed to activate timing reduction the first pulls).

I'll play a bit more with it. It would be nice to find some cause and effect.

As a side note, I find it fascinating the more one dives into these things how much everything interferes with each other. I think one will never be done tuning these engines unless it's a decision is made to stop.

[Kodename47]

Improved the sheet a little with more information. It also plots the cycle as below:


EcuTek recommend a min of 2ms between the EOI and spark timing, which is where I made my comment. I don't think you'll hit that limit NA though even if you ran 100% DI. You might get close though. But as I said, I would really be looking to inject as late as possible. After the intake valve has close, or as near to it for best cylinder charge management. But then you need to allow time for the fuel to mix well. On NA I wouldn't be advancing the timing but seeing if retarding works at all.

[Tor]

Originally Posted by Kodename47

Improved the sheet a little with more information. It also plots the cycle as below:

That looks a lot nicer than mine!

EcuTek recommend a min of 2ms between the EOI and spark timing, which is where I made my comment. I don't think you'll hit that limit NA though even if you ran 100% DI. You might get close though. But as I said, I would really be looking to inject as late as possible. After the intake valve has close, or as near to it for best cylinder charge management. But then you need to allow time for the fuel to mix well. On NA I wouldn't be advancing the timing but seeing if retarding works at all.

2 ms at 7200 rpm is 86 deg. My IPW is 3.34 ms or 144 deg, so from the Ecutek recommandation I could retard to 230 GDI firing angle. But why is stock then 370 with IPW 3.56 or 154 deg?

Could it be, that Ecutek's recommandation is aimed mainly at forced induction and they assume a lot longer IPW and hence, a lot earlier SOE? For the entire duration of the IPW mixing would also take place.

Added "2 ms and 6 ms seconds prior to ignition lines" for reference and matched your colors (but still upside down):

[Kodename47]

Originally Posted by Tor

2 ms at 7200 rpm is 86 deg. My IPW is 3.34 ms or 144 deg, so from the Ecutek recommandation I could retard to 230 GDI firing angle. But why is stock then 370 with IPW 3.56 or 154 deg?

Could it be, that Ecutek's recommandation is aimed mainly at forced induction and they assume a lot longer IPW and hence, a lot earlier SOE? For the entire duration of the IPW mixing would also take place.

Their info is geared towards higher volumes of fuel, but just though it was worth passing on. My graph is all part of my tool, fully cross integrated. I've also updated the chart with 2ms and 5ms lines before spark timing, which is derived from the timing tables. All available to download

I still think the PI mix plays a huge part of what you can achieve here. If you run more PI at lower RPM at WOT you may gain from some retarded timing, you may not. I guess we're in the realms of trying to research and test. Most DI documents I've found are on diesel engine or comparing them to gasoline, the Lexus/Toyota papers are interesting and the most in depth. If you find any worthy of a read them post them up here.

[Tor]

Originally Posted by Kodename47

Their info is geared towards higher volumes of fuel, but just though it was worth passing on. My graph is all part of my tool, fully cross integrated. I've also updated the chart with 2ms and 5ms lines before spark timing, which is derived from the timing tables. All available to download

I still think the PI mix plays a huge part of what you can achieve here. If you run more PI at lower RPM at WOT you may gain from some retarded timing, you may not. I guess we're in the realms of trying to research and test. Most DI documents I've found are on diesel engine or comparing them to gasoline, the Lexus/Toyota papers are interesting and the most in depth. If you find any worthy of a read them post them up here.

I suppose another thing to consider, besides from cooling effects, is how it mixes. This study of injector/spark plug location quotes from a previous experiment:

They reported that the early fuel injection resulted in the formation of homogeneous mixture which led to stable combustion.

Another point is emissions. Although the conclusion is regarding the placement of the injector/spark plug:

Finally, it is concluded that, the case 1 results in the better mixture stratification, which leads to better combustion and performance. However, the NOx emissions are higher, but the soot and CO emissions are least

Combined with the info in this abstract:
http://papers.sae.org/1999-01-0502/

The results presented here indicate that the impingement of liquid fuel on the piston top and cylinder liner is an important source of HC emissions from direct-injection spark-ignition (DISI) engines, and that this mechanism is at least partially responsible for their high HC emissions.

What I mean to say, is that the stock firing angles are probably not adjusted mainly for performance, but rather with a huge compromise of what provide the lowest emissions.

I don't mean any conclusions can be made from any of this. As you said, this is borderline R&D, but even worse since there is no clear goal (Reduce knock? Increase performance? Etc.). I think I will revert to a known table (stock or OFT) and await the webinar from HP Academy for now.

[Matt@Cosworth]

If you move the inlet cam then the GDi angle should also move with it as the two are connected and as Kodename states this can also change the ignition angle

these changes are quite subtle and very difficult to pick up without having the car on the rollers especially on a NA car

[Tor]

Originally Posted by Matt@Cosworth

If you move the inlet cam then the GDi angle should also move with it as the two are connected and as Kodename states this can also change the ignition angle

these changes are quite subtle and very difficult to pick up without having the car on the rollers especially on a NA car

Thanks a lot for that insight.

So does that mean if I advance the inlet cam I should advance the GDI firing angle as well? And is there a relation that it's the same amount in deg?
What do you look for when at the dyno? More torque? That it goes richer? Less knock?



What I decided to do for now is keeping stock values to 2400 rpm, then slightly more advanced from 2800 to 4800 (1/2 missing IPW in deg at each rpm, to account for the lower injected volume due to running PI thoughout the range). At high rpm I sticked to what I done so far already.

Looking like this using proper excel tools:



Previous tune, the dotted orange line:



I have to admit that I can't say it's the GDI making the difference alone, since I also made POL 0.16 richer around 4500. But this version:



The FLKC in 5th gear might look bad, but normally my TCPC/ECT hack removes 2.4 deg timing there, but it's too cold for it to activate. So in that light, it's really not too bad and nothing that I wouldn't expect to see. Also after doing AVCS, I leaned it out 0.5 AFR, so compared to what I saw in the fall without the hack it's even so about the same. I should have added another piece of ducktape infront of my oil cooler too, which would block some of the radiator too, but I forgot before driving off.