Tuning a boosted 8 is not that difficult in itself; all the info required is freely avaiable on the internet. For me, it was enough to understand what's going on to decide what I need to do. Surely, experience or lack of it will play a role here. There is no dumb question, there are only dumb people who think they know and don't research or ask others. There is also the babystep method where you take it easy, be super conservative and only vary 1 parameter at a time while keeping others in a super safe area at the expense of making your turbo car slower than it was NA.
Tuning is described by some as black magic, especially by those who do it for a living, to deter others from even attempting to learn about it. News flash, if humans made it, it can be understood by a human.

 

Yeah ...actually it's pretty easy. Just like anything is easy, once you know how.
My point above wasn't not to try it because it's hard but ... don't take it lightly!

 

My understanding is that the table size is fixed, the column values are not. You can change the values of the headers as Brettus has said and the ECU interpolates value in between.

 

Thankyou for clarity. While i have not completed the hardware fabrication and setup for the turbo, i want to get a idea of the process involved in tuning for boost. ill document the build and post at later date.

Project car: 2005 jdm rx8 type s, rhd.
turbo parts acquired (used): greddy manifold, modified garret 3071r (think mmaniac)
tuning: mazdaedit

 

Got that same setup, Brett tuned it, here's a video at Suzuka for reference. Boost is 0.7 bar.

 

Hi! Could somebody explain how RPM limit hysteresis works?
We have a track car with friends (early JDM S1) and recently I was reading a lot to get more info on Mazdaedit that we got for the car. So far I only did some reliability changes like fans and regular stuff like that, but for RPM limit I increased the hysteresis to 400 to understand if it is added to the limit value or subtracted.
Recently I noticed that at the track I sometimes hit RPM limiter at about 200 rpm lower than on some other earlier laps in the same session, which is the exact change that I made to hysteresis value (200=>400). Coolant seems to be at the same level and I'm not really sure what else can affect this. I plan to experiment with it a bit later, but if somebody can spot something obvious that'll help a lot

 

.
hysteresis is when when it goes back to normal operation. For example fans come on at say 90C. They stay on until 85c. 85c is the hysteresis. The rev limiter hysteresis cuts it until rpm drops x. X being the value you put in the hysteresis box. Essentially you will have a larger drop in rpm until you can drive again. Put it 5000 and you won’t be able to drive until 4K. 🤣

 

Interesting... RPM limit is unchanged and is 9000, if I'm not mistaken... But I'm pretty sure that I've seen 9400+ logs..
and car definitely cuts later on earlier laps in the session.
I just thought that I was driving faster since I started hitting limiter more often, but lap times say the opposite 🤣
you can kind of see on the log below with TPS and RPM traces for two laps in the same session, I was staying full throttle for much longer, but RPM stopped climbing at around 9200 while 2 laps earlier they reached 9480...

 

Quick question for someone who has a complete understanding of MazdaEdit and the ECU. Please, no guessing here.

Is the table that is labelled VE in MazdaEdit actually just the base fuel map? The numbers in there seem to almost exactly match the AFR targets (in lambda) that I'm logging.

Even in open-loop, it looks like it's logging these values as the targets rather than the targets set in the "open loop" tables.

Thanks,
Bryan

 

No it's not the base fuel map , but it is part of the fuel calculation that the ecu makes so does have a direct effect on AFRs in open loop.

 

Okay, thanks.

That confirms what I just observed from a couple of hours of testing... basically, nothing works the way I would expect

I just want to be able to adjust a particular AFR target and see it actually do that. But alas, I've learned a lot of what does *not* do that.

I'll have to dig some more.

 

I'll save you some pain ................. Adjust ALL the fuel maps to be the same , then adjust all maps together. Ignore the fuel by gear titles to the maps.

Back in the day ...maybe 15 years ago it took me about a month to find out that one little gem of info by trial and errror...such was the knowledge about tuning these cars back then.

 

You sir, are my hero. I'll try it out.

Thank you!

 

I'll just add a bit of info to that -- despite the map titles indicating a relation to gear, they are used differently -- at least in my car's ECU (EUDM, 6-speed, MT, HiP)

The ECU uses 4 sets of maps, let's call them A, A-limp, B1, and B2. When computing the fueling target, the ECU always computes two fuel targets, A and B and then decides which one to use.

For target A, it chooses between (maps from) sets A and A-limp. I call it A-limp, because it really is a set of 1-point maps with safe defaults (14.7 AFR for both throttle and load-based fuel, no BARO compensation for throttle-based values).

For target B, the ECU uses (maps from) one of the sets B1 or B2, based on conditions I haven't been able to completely decipher yet. In one case, the B1 table is used for some value X <= 4, otherwise B2 table is used, and in another case, the B1 table is used for some value Y <= 3, otherwise B2 table is used. I have yet to find how values X and Y are computed and what is their meaning, but elsewhere (in another big chunk of code that evaluates fueling conditions) they are also compared to values 5 and 6, which might suggest they are related to gears. Anyway, the key thing is that the tables themselves do not relate to specific gears.

As to what the ECU does next... In general, it blends target A and target B based on two other conditions. One condition that is very visible is that if the ECT is _strictly_ below -40 degC (the ECU actually checks for range -43 .. -40 and if ECT is below -43, it considers it to be below -40), the ECU blends target A and target B. The blending is done simply as (1 - target_B_weight) * target_A + target_B_weight * target_B, and the ECU adjusts the target_B_weight over time to make the transition smooth (sorry if I got too technical here). If the ECT is above -40 degC, then the ECU may or may not do the blending, depending on some other value Z, the exact meaning of which I wasn't able to decipher yet. If Z is nonzero, no blending is being done and the ECU uses target A, if Z is zero, the ECU blends the values. The value Z may very well be just a timeout for using target A exclusively before going back to the blending.

The ignition tables are used in a similar way. In the limp case, the ECU just puts a hard limit on the advance (16 deg leading and 11 deg trailing), but other than that, it calculates advance targets A and B in a similar fashion (using A, and B1/B2 tables) and always blends target A and target B using the target_B_weight that is dynamically adjusted by the ECU.

Edit: my description wasn't entirely correct (if it's at all possible, it would be easier to talk code) so I updated it as best I can at the moment. If you ignore my inability to describe what exactly is going on (because of the unknown values X, Y and Z), I think the key point is the ECU is not using individual maps, but it either uses map A alone, or it blends it with one of the B maps to implement various enrichments and if you want to avoid that, making those tables identical (as Brettus suggested) eliminates it.
I just wanted to provide a more technical explanation on why it should help :-)

 

Holy cow. I'm stashing this info in my notes. Thank you both for the leg-up!

 

Welp, I've successfully modified my warm up and open loop afrs.

Now to figure out closed loop

 

Woo. Success!

Changing the closed loop correction tables definitely adjusts the target afr.

Here are the correction values and the corresponding afr targets I ended up with from my testing:

0: 14.7 afr
-.5: 12.5 afr
-.25: 12.5 afr
-.1: 13.2 afr
-.05: 14 afr

Now that I have a pretty good idea of how this thing works, I can finally start working on smoothing out that nutty open loop fuel map.

 

Doesn't correspond to my findings
Max. possible adjustment : plus or minus 0.03
approx. Result :
plus 0.03 = 15.7 ....................minus 0.03 = 13.7

suggest you log several mins and take your readings when stft has settled, then average the result from the log.

 

Interesting. I'll definitely play with that some more.

 

I always thought that whole point of MAF calibration and the rest of documented Mazdaedit workflow is specifically made to get that result. If your inputs are correct, you should hit your AFR targets from the tables, right?

 

After flashing this tune, warming up the car, and then logging a ~3 mile drive, this is what I've got. The second graph are the averaged closed loop targets I'm getting and the last is of the averaged STFTs.

If I'm a complete idiot and missing something, please point it out

 

I'm not an expert in this area at all, so I'm not going to argue.

 

Looks like I do have some minor maf scaling to do. This is commanded vs actual afr in open loop plotted against maf volts and flow.

 

Before you read this: I have read everything I can find about tuning with ME, watched Kanes videos, and have taken some ecu tuning courses; some specific to rotary engines. I'm just trying to get all of what I've learned to match up in some logical manner. In the end, I have no idea what I'm talking about, so If I'm making some dumb observations or assumptions, please point them out.

My current goal is to have the ecu commanded/target afr be the one I'm looking for whether it's in open or closed loop, so I can use it to more accurately tune the engine; especially when nitrous is engaged. I know It's not strictly necessary but I haven't been able to find much info about tuning closed loop in ME. So I'm posting what I'm observing.

This first screenshot shows full scatter plots of commanded afr on the left vs measured on the right. The right side looks like what I'd expect my car to be doing, and lines up with my target base tune. But the commanded afrs on the left do not match up at all, which makes them pretty useless. Again, my goal is to make them useful.

The picture becomes a little clearer in the second screenshot when I filter open loop out of the left side. It looks like it's almost always commanding 13.5 afr, which lines up with my closed loop corrections table. It looks like the exceptions are when it's making stft adjustments.

Then the third screenshot shows what it looks like with closed loop filtered out. Again, commanded afr is 13.5 except under overrun. It looks like it spits out the closed loop target, even when it's in open loop. That's interesting to me because I haven't read anything that explains why that happens and is contradictory to the ecu tuning courses I've taken.

So I'll keep digging and testing unless someone knows and is willing to explain the answer

 

I used to monitor commanded values many years ago but eventually decide it's not a useful parameter, and stopped doing so. I never looked at it in as much depth as what you are doing though so am interested to see what you find out.
My strategy is always to get the open loop fuel tables as close as possible to actual in all circumstances but don't stress too much if it's out by as much as a point in some places.