NefMoto

ME7 M-Box (5120 hack and 4bar MAP) with Tial 770R's. The wastegates crack at about 22psi, so the ability to control the boost in low-throttle, high RPM situations is limited and is resulting in 17963 Charge pressure: Maximum limit exceeded. Closing the throttle plate and retarding ignition are limited mitigations, i.e I have seen 20PSI+ with ~25% throttle.

My approach so far has been to increase KFMIRL (and generate the KFMIOP inverse), in areas where I essentially cannot control boost, to something that (kind of) follows the boost profile. The result is that mimax_w has been raised to something that looks reasonable. Unfortunately, mifa_w has not come along with it which eventually results in my requested load/boost being to low. One of the torque monitor interventions in MDFAW must be limiting mifa_w.


Following from wped_w, mrfa_w is requesting enough torque. I don't have the memory address of mivbeg_w, but I don't think mivbeg_w is the problem unless there is something very strange happening with mrfa_w(mimax_w - mimin_w) as MDIMX = 99. mivbeb_w is not the problem assuming mivbeg_w is reasonable since mimax_w is not the limiting factor.

DMFABEG has a number of potential interventions.

• SAWE: Change limitation during overrun fuel cutoff and reinsertion
• DASHPOT: Change limitation with negative load changes (Dashpot)
• MISMEUS: Change limitation with fast torque intervention for operating mode changeover
• LSD: Change limitation during positive load changes (load impact evaporation)
• MDBG: Momentary gradient limitation

Only MISMEUS, LSD and MDBG seem like candidates.

• MISMEUS: Not clear to me what is happening here (and very little information). I do not have the memory address for B_mismeus. Can anyone elaborate?
• LSD: I can log B_lsd and it is 0 so not a factor.
• MDBG: KFMIFABG looks interesting, but it looks like SY_CVT = 0, so it isn't used (?). I do not have the memory address for B_mifabg. Can anyone elaborate?

Any help is greatly appreciated.

None of what you posted is relevant to your problem.
Literally none of it.

If you are running retarded actuators, then do a run with 0 DC at WOT and configure the cracking pressure in the ECU correctly, so it knows when to limit with the throttle and when not to.

KFVPDKSD/KFVPDKSE for setting base spring pressure. Setting these properly will allow the ecu to control boost with the throttle for values under the wastegate spring pressure

I have done this. As I mentioned, limiting the throttle does not appear to be enough. Even with the throttle limited at 20%, I have seen over 20psi at or above 5,000RPM. Attempts to close the throttle down even more have had limited success, other undesirable side effects and even then I am not requesting enough load and getting significant boost deviations.

If 20% throttle is not enough to prevent overboosting, I'd reconsider the wg setup :/

How about a csv instead of png bullshit, so I can actually look at relevant stuff.

Either:
a) Your throttle maps are not configured correctly and rl is much lower than rlsol
b) Your dump valves have WAY too stiff springs, because the differential in pressure should push them open.

Or the combination of the above.

The turbos are set up the way Johnny recommends. This is the reality of a 20PSI WG crack pressure and my intention from the beginning of the build was to be able to make around 40PSI. I knew what I was getting into and I was aware that tuning part throttle at high RPM might be interesting, but I was expecting to be able to do more with limiting the throttle. Maybe I can do a better job of closing down the throttle, but tuning for an arbitrary torque request limit is silly.  I am really more interested in what the mystery torque intervention is.


a) rl is not much lower than rlsol.
b) See above. The configuration is no different than most Tial setups, i.e. the WG's won't start to crack until 20psi.

The forum is refusing to upload the CSV. https://drive.google.com/file/d/1ZHahZk2lD7ozHtq-y7CuqP_Lxx5cpHnu/view?usp=sharing (https://drive.google.com/file/d/1ZHahZk2lD7ozHtq-y7CuqP_Lxx5cpHnu/view?usp=sharing)

I appreciate you looking at them, let me know if you suspect I have done something ridiculous.

Now re-read my b), and actually pay attention.
I was not talking about wastegates.

If you fix that problem (did you delete N249?), you will also fix the problem with your insane boost level without touching any of your actuators.

You are talking about the 710s?

As an aside, is there a way to feed the single wg n75 signal to an external wg and external dual wg solenoid?

I am talking about whatever he is using for recirc.
With that boost difference they should be open. Whether via mechanical means or through N249.

His pressure drop across throttle plate is 1 bar.

N249 has not been deleted. I have new Forge DV's that replaced old Forge DV's during the build. I just looked and they appear to have the 15-23PSI spring instead of the 5-15PSI springs. You are likely onto something here (can't thank you enough for helping). I will replace them and report back.

15-23psi???

Do those things ever recirc? That would require an obscene amount of vacuum to overcome spring pressure

The lighter spring has not helped. From the design of the Forge DV's, it is clear that a positive pressure differential in the charge path relative to the intake path will only push the valves closed. They will only open under vacuum (or if for some reason the intake path had a positive pressure differential relative to the charge path). I am not sure if 710's are also designed in this way, but from what I can tell from reading the documentation the N249 was not intended to be used as a boost limiter. Only as a turbo bypass under vacuum and re circulation coming off boost.


They do. They are quite soft (it doesn't take much vacuum to get them to compress) and those numbers are presumably an (arbitrary) conversion to boost pressure. Although, based on my comments above, I don't see why you would need a stiffer spring in the first place.

Logs: https://drive.google.com/open?id=1pcFiR09vHxvMN9J_I_kOhsEPIoFWS_7i (https://drive.google.com/open?id=1pcFiR09vHxvMN9J_I_kOhsEPIoFWS_7i)

Haven't had a chance to look at them really closely, but on the last pull (the interesting one) I had TM 2 and it shut down the ECU. It was't immediately clear to me why.

I'm a little unclear on this as well. I thought that was the whole point of any properly functioning dv - do not leak any pressure unless actuated. My assumption was that the DV should not open unless there is vacuum (compared to atmospheric) on the actuator.

Installing them backwards may make them act more like a pressure release valve (too much pressure relative to atmospheric) but that isn't really what you want either.

I could be wrong, of course.

If you install them backwards they will be held closed with spring pressure as the force from the inlet is the same as the boost force (since the throttle plate isn't a restriction).

If it is, the boost will blow it open.  As far as my understanding goes.

Correct, I would purposefully install them in the "wrong" orientation in this case, so that if there is a big difference in pressure, they will get blown open.
However, that actually probably NEEDS a stiffer spring, because the DV input is pre-intercooler so you have to account for the pressure drop across intercooler, both due to intercooler flow as well as the adiabatic process within the intercooler.

In the last month I have gone through about 50 more tunes trying to sort the issue of controlling the boost with the throttle while the boost pressure is under the waste gate cracking pressure. I have had limited success.

WOT is great, car is fast and runs as expected. Less than 50% throttle up to 7,000rpm is good. It isn't "butter smooth", but it is fairly smooth and totally reasonable to daily drive. It is 60%-80% throttle at above 4,000rpm where everything goes wrong and I get the horrible "kangaroo-ing" throttle/boost.

Kangaroo-ing -> https://drive.google.com/open?id=1E6USTu7CwxQg5MGBVDAwTqEszm3N8jLZ (https://drive.google.com/open?id=1E6USTu7CwxQg5MGBVDAwTqEszm3N8jLZ)

KFVPDKSD ->https://drive.google.com/open?id=11qR5cZURsyw3TWhiqz-ZQUMmJZqiKZEX (https://drive.google.com/open?id=11qR5cZURsyw3TWhiqz-ZQUMmJZqiKZEX)

WDKUGDN -> https://drive.google.com/open?id=19HPbVIqw8H-5UvlYQh5WQTAlIFI9JiVe (https://drive.google.com/open?id=19HPbVIqw8H-5UvlYQh5WQTAlIFI9JiVe)

KFVPDKSD/E

I agree with prj that the only rational way to define this map is by taking logs at 0% WGDC and 100% throttle to see where you need to close the throttle to control the boost (i.e, requested pressure ratio can be achieved and exceeded at a given RPM) and where the throttle should be 100% open (i.e, requested pressure ratio cannot be achieved at a given RPM). This map is fairly trivial to rough in, much more difficult to fine tune when you consider what WDKUGDN has to look like.

One of the issues I ran across was that I have these logs in 30 degree F ambient temperatures and 90 degree F ambient temperatures an there is a big difference in what pressure ratios can be achieved for a given RPM. The difference between 30 degrees and 90 degrees might be a pressure ratio that varies as much as 0.5 for the same given RPM and that can shift you a few columns to the left or right in KFVPDKSD/E. I opted to take the most aggressive logs (cold weather) and use those. Worst case is I am over throttled in hot weather.

WDKUGDN

To actually control the boost with the throttle (especially at high RPM) WDKUGDN has to close down the throttle aggressively. A throttle position of somewhere between 16-22% seems to be the sweet spot. Closing the throttle more than 16% seems to put the engine close enough to vacuum that the DV's will release in a rapid cycling manner, or maybe it is a surging issue...  However, even at 20% you will see the boost creep up as the RPMs increase.

Log -> https://drive.google.com/open?id=1QREWRBu_SXp0FObmGIPpV1okiY_K9lXM (https://drive.google.com/open?id=1QREWRBu_SXp0FObmGIPpV1okiY_K9lXM).

Because the difference between WDKUGDN and the requested throttle angle at a high throttle/high rpm is so large, even a small change in the Z-values in KFVPDKSD/E have a huge effect on the throttle angle. Another issue I see is that the MAP sensor is upstream of the throttle body and the compressor wheels of the Tial's 770's can have a lot of angular momentum. This means a big change in throttle position at high rpm can cause a big pressure wave back upstream towards the turbos. The result is pvdkds spikes (MAP is pre-throttle body) but the pressure in the intake manifold can drop to near vacuum which causes the DV's to dump followed by a series neck snapping kangaroos. Also, this can't be good for the turbos.

The Point:

I have basically exhausted every reasonable set of values for KFVPDKSD/E and WDKUGDN. Because of the differences in PR for ambient temperatures, the low resolution of the maps, the basically infinite number of ways a pair of X and Y values can be interpolated for in the maps, the transience of the turbos, the pre-throttle body MAP and the throttle angle required to control boost at high RPM... it is extremely difficult (maybe impossible) to get this right. I know others have claimed they have nailed this with similar setups, I am skeptical.

The Future:

I think prj is onto something with inverted DV's and the DV's needing to dump excess boost. I have seen AMD cars with inverted DV's and wondered for a long time what the reasoning behind them was. It would be reasonable to assume the issues I have been seeing are why the DV's have been inverted. So, next step is to try that. Feel free to explain why I am totally wrong and everyone else has "butter smooth" big turbos.

Update: First try with inverted Forge DV's with a stiff spring is a failure. I can't get over 2-3psi of boost. Obviously a big pressure differential, I am thinking I can't get the engine completely out of vacuum, but I'll trouble shoot and give it another go if I can find something wrong.

Great post, lots of good information there.


I have yet to experience any non k03 tune regardless of wg cracking pressure that has anywhere near the stock part throttle smoothness on k03s.

The people who say their "part throttle is butter smooth" generally spent so much money on their set up that they can't admit their part throttle might not be so great.

Thanks, Nye. Unfortunately, at this point I can go on and on about what hasn't worked and all of the many reasons why it probably didn't work, but I don't have much to say on what has worked or ideas I have left that might make it work.

K03's have a low cracking pressure, they spool very fast and they have a relatively small amount of angular momentum. That is to say with K03's, you don't spend much time in areas of KFVPDKSD where vpsspls_w << 1 and even if it is, the K03's are more responsive to throttle inputs and overall it is more of a 'smoothing' effect rather than an outright 'control' effect. That is to say, I am fairly confident the throttle control was never intended to be used as full form boost control.

At this point I would take moderately smooth even if the throttle/power delivery is very non-linear at high RPM's (just no kangaroo-ing), but that is going to take some serious tweaking to the load maps to get requested load up to the point that the turbos are going to make and I just have to deal with not having full control over (which takes me back to the original point of this post after all of that...).

Seems like the throttle is kangaroo-ing despite vpsspls_w being more or less steady while plsol ~= pvdkds, too. Something else must be going on here...

Log -> https://drive.google.com/open?id=1-o0hy9IzFHYl7fph7Uddvh8W4ko93_GT (https://drive.google.com/open?id=1-o0hy9IzFHYl7fph7Uddvh8W4ko93_GT)

There is no desired vs. actual boost in this last log, but to me it looks like over-boost throttle interventions, I see them on my stock ME7.9.10 setup from time to time. So perhaps a separate issue from WDKUGDN / KFVPDKS tuning?

Another thing - on the mentioned ECU of mine there is no WDKUGDN (well, there is, but it is turned off with a code word), and the values are calculated from KFWDKMSNVP. If anything, this tells me that these two have to stay in close relation for the whole thing to work. So perhaps it is unwise to touch WDKUGDN as long as the throttle body is stock?

Edit: not untouched but checked to be in sync with the other one.
 

You absolutely have to touch WDKUGDN.
And to set WDKUGDN plot mshfm vs msdk, and adjust WDKUGDN until they are more or less equal at WOT.

I tried this approach. I think it gave me the "smoothest" results so far, but the actual boost was much higher than desired and I went back to closing the throttle more aggressively to keep actual boost more in-line with desired. The other option would be to raise KFMIRL in areas above 4,000rpm/50% throttle to something that better resembles the boost I see when MSHFM ~= MSDK , but I haven't had great results with that so far because of the mysterious torque interventions.

v101 - MSHFM vs MSDK -> https://drive.google.com/open?id=1HbIspqY1jUT7qpWPuegWl0awcIYoxUMo (https://drive.google.com/open?id=1HbIspqY1jUT7qpWPuegWl0awcIYoxUMo)
v101 - PVDKS vs PLSOL -> https://drive.google.com/open?id=1nM08ceuV2m_PBBN8Rh5ZiWq63VGGrXIo (https://drive.google.com/open?id=1nM08ceuV2m_PBBN8Rh5ZiWq63VGGrXIo)
v101 - Logs -> https://drive.google.com/open?id=1_xrOgcrv_61DFtuPHSkNuiAAdEsS3XPK (https://drive.google.com/open?id=1_xrOgcrv_61DFtuPHSkNuiAAdEsS3XPK)

The approach where I try to keep the throttle angle small to keep PVDKS != PLSOL results in MSDK << MSHFM.

v121 - MSHFM vs MSDK -> https://drive.google.com/open?id=17wzhhWmbcH6GUpT1IhP5iu-i_dKmRH8o (https://drive.google.com/open?id=17wzhhWmbcH6GUpT1IhP5iu-i_dKmRH8o)

It is odd that msdk is increasing while the throttle angle is decreasing. Modifying KFWDKMSN/WDK to get MSHFM ~= MSDK might help fix the throttle cuts, but that seems questionable.


v121 - PVDKS vs PLSOL -> https://drive.google.com/open?id=1rNogbvzKOQw0dSLqdGnE0hcHT2N8jMPS (https://drive.google.com/open?id=1rNogbvzKOQw0dSLqdGnE0hcHT2N8jMPS)
v121 - Logs -> https://drive.google.com/open?id=1ZtC6FUZCcMWlrIANaQlxHNTeWONrayGm (https://drive.google.com/open?id=1ZtC6FUZCcMWlrIANaQlxHNTeWONrayGm)

So yes, it looks like over-boost throttle interventions could be the cause (although I thought I had disabled them). I think decreasing WDKUGDN above 4000rpm to prevent overboost and trying to get KFWDKMSN/WDK to output something more reasonable for msdk are the next steps to take. Maybe increase KFMIRL in places I can without hitting the mysterious torque interventions.

I told you exactly how to set WDKUGDN.
It is the maximum throttle position which still restricts airflow. Doing it any other way is wrong, simple as that.
KFWDKMSN and the inverse are physical properties of the throttle body calibrated on a flowbench. You never have to touch them.
You need a lot more studying of the algorithms and model before just screwing with random shit that you have no idea how it works.

Your overboost pre-throttle is because of super stiff actuators and it's not going away anywhere.
You're not even logging or comparing the right thing which shows you have no idea how this works even though all your wording implies that you are an expert on this.
Of course pre-throttle there is going to be high pressure when you are choking the intake with the throttle and have stiff actuators.
You're supposed to compare ps_w with pssol_w. Pressure in the manifold. That's what everything works off of. Maybe go back to square 1 and understand a little more about how ME7 works :/

Get rid of your aftermarket DV's and fit standard 710P valves in non-standard orientation so that excess pressure actually gets vented, if you're worried about it. Or don't.
As long as msdk_w and mshfm_w are fine and your actual ps_w and pssol_w are fine it is of 0 importance what pvdk actually is.

Starting to feel like i'm wasting my breath here.

This is my understanding of it as well, but it is not what I am experiencing. I still see overboost codes, throttle cut, etc.. presumably because of pvdk since ps_w is always less than pssol_w. What am I missing here?


To be clear, my DV springs are comparable to what is in the 710P. I have inverted them as you suggested and I couldn't make any boost, even with the most stiff springs in the DV's. I see no reason that 710P would perform any better in this orientation.

In general I agree with what you are saying, I am just not seeing good results when I take that approach. Could you provide some logs from a BT car so I can see your approach to ps_w vs pssol_w and pvdks vs plsol?

Is your MAF underscaled?

It is by ~25%. It would be right at the limit if it were not.

Would it be wrong to make a spreadsheet and calculate WDKUGDN at setpoints the way it is done dynamically from KFWDKMSN (looking for CWUGD in MED9.1 FR reveals how)? Taking as reference for pvdkds the cracking pressure for example? Or plgru line if the ECU has it? (Both of which should be the roughly equivalent, shouldn't they?)

By the way, I wonder what is the practical difference, for ECUs supporting both ways, between calculating it dynamically and having it static... One thing that comes to mind is no need to re-calibrate for different turbos, but what about daily throttle operation?

And here's the problem. DUH. What limit are you on about?

Overboost codes specifically can be removed by adjusting KFDLULS.
Throttle doesn't give a damn about your boost, it's only about manifold filling.

Tune the ECU like the FR says and you won't have issues. Rather than "scaling this" "scaling that" "doing this" "doing that" - of course when you screw up all the models your throttle control goes nuts, what do you expect?

Also I've not had issues making boost with DV's inverted. Maybe something is not fitted right?

I have a 90mm MAF housing with a Hitachi sensor. In theory, at 5v the maximum airflow I can measure (without under scaling) is ~560g/sec. When I initially did the fueling, I looked at the few logs I could find and estimated that they were well over 600g/sec based on injector size, idc and afr. So, I under scaled my MAF by ~25% to give me a little bit of breathing room. Based on some of my logs at 32psi in cold weather, it looks like I have hit ~590g/sec. I am at 5,200ft, so I am limited to ~35psi (PR limit is 4) and I could probably reduce the under scaling, but when I go to sea level I don't want to hit any limits, either. I don't like that I have to under scale the MAF, I just don't see an alternative.


KFDLULS is FF. I'll give it another try, but in the logs I have posted I am fairly sure I am still seeing throttle cut due to pvdk (no codes, however).

Update: This worked. prj is right on (can't thank you enough). Numb/disable KFDLULS, mshfm ~= msdk, ps_w < pssol_w means no throttle cut and is reasonably smooth regardless of RPM or throttle.

Wouldn't moving the MAP downstream of the throttle body resolve having to numb/disable KFDLULS?


Or maybe something is just broken. It is all 20 years old and I haven't replaced any of it. It seems like the signal line on the DV's isn't seeing positive pressure (only vacuum/near ambient) and the pressure differential is just pushing the valve open. Not sure if this is by design or if a check-valve is stuck or whatever. I'll take a look. Regardless, I don't really care. As long as I can get actual filling to meet requested filling without the throttle cutting I am happy.

Where does this "560g/sec" limit come from?
If you're out of MAF then underscaling doesn't help you in any way, it'll still max out all the same. Explain your reasoning.
Ok, let me say this a third time, so it gets through to you - the ECU will never cut throttle due to pvdk > plsol EVER. Apart from overboost related limp mode.
Which you are hitting because your models are screwed up.
It must see positive pressure, your stuff is plumbed wrong then. The only time it it doesn't is when the N249 is actuated.

I was under the impression the MAF/MLHFM should stay under 1738.20kg/hr (~488g/sec) from KFMSNWDK. Is that not the case? Also, is there a hard limit for the MAF g/sec on the ECU? I have never seen a log with numbers anywhere near 600g/sec...


I get it now. Nye, this should be made more clear in the wiki, especially since ECUxPlot and the translated FK refer to pvdk as "actual" and it is littered all over the common tuning threads.

"If you don't get all of this just right, and your actual boost goes too far above requested boost (by ~200mBar), you may experience overboost throttle cut due to negative deviation, which is ME attempting to get boost back under control by temporarily closing the throttle plate."


One of the few things on the car that is still plumbed from the factory. The N249 might be stuck open (which would explain the increased time to spool with a lighter spring), probably worth replacing.

I have no idea where you got that "impression" from, but it couldn't be further removed from the truth. KFMSNWDK gets multiplied by PR, now think about what you said...
The hard limit is ~1800g/s. I don't think that's gonna be a problem for you, now is it?

This statement is wrong, read FR not wiki. Negative deviation is limp mode, yes it goes into limp mode, and the reason it cuts throttle then is because you're running on the limp mode limiter for ... drumroll ... load.
Throttle control doesn't care about boost.

Yes, I think there is a ton of wrong "throttle cut" stuff in the s4wiki which was written before I had access to the FR, and it seemed (at the time) that slow path intervention was lde based, not load/torque based. That said, prj, I have a ton of throttle cut logs that show the plate closing WITHOUT throwing a code (I assume because it isn't actually negative deviation, just slow path torque intervention).

Obviously, nobody but me has ever really bothered to correct anything since then. More fail on my part to get anyone interested in documenting anything.

Yes, slow path intervention due to rl > rlsol