How's my LV?

[iNVAR]

Yeah, I'm not sure what's going on. Hang on, let me double check it, because according to this, I'm making more power right now than I was last year during the cooler temperature. Going to doublecheck that I put everything in right. Also, the 3 logs I'm using are these:

1) Last year October, clean log, approved by Infamous on my final tune. 60F outside.

 

2) Yesterday's WOT log, same map as (1), but significantly hotter outside. 83F let's say.

 

3) Today's WOT log, final map I made up. 83F outside.

romraiderlog_20101003_191155.csv

romraiderlog_20110818_155452_3rd_wot_3.csv

romraiderlog_20110819_152327.csv

[Spec B]

^ PM me your email, I want to send you a bill.

 

Wierd...may be the road you used - gotta always use the same one.

[iNVAR]

Hmm, crap, you know what it could be. The stretch of road I did the pull on might have been ever so very, very, very slightly inclined/declined in the other direction. That could be it. Ugh. Can't really compared the dyno plots then since the stretches of road are slightly different.

 

edit: Haha, yeah, just saw your reply. I think that's what it was. The roads. I was on the same street but on different portions of it or going in different directions.

 

http://www.crimetank.com/misc/plot-full1.jpg

http://www.crimetank.com/misc/plot-full2.jpg

http://www.crimetank.com/misc/plot-full3.jpg

[iNVAR]

So the dyno plot looks really weird, the shape I mean. Don't really care about the numbers. Should that be cause for concern? Or just chalk it up to heat?

[Spec B]

I suspect it is the road...maybe a dip or something.

[iNVAR]

Ah yes, you're right. I plotted a few of the other 20 or so LVs and it appears to be related to the road. Pretty cool how little imperfections and variations on the road will all show up on the graph!

 

So I'm pretty much done, except for two questions:

1) If I want to keep things smooth, I can't really add timing back in anything because that area knocks. I should just leave things as they are now, right?

 

2) The area I mentioned earlier, around 1.05/2200RPM, it's pretty much consistently knocking there. I logged knocks in there about 2-3 times during a daily drive today, and yesterday, and the day before. If I hadn't been resetting it, the ECU would've relearned it. I know you said you write off knock under 2.5K as noise, but if it's fairly consistent and it does show up on the LV, should I just drop a degree in that area and see what happens? If the knock still shows up after dropping 2 degrees, it's probably noise, right?

 

Thanks. I think I owe you more than a drink at this point. Lunch/dinner on me!

[Spec B]

1 - I think it is fine as is - it is actually pretty smooth.

 

2 - Typically in that area of the map the timing should be pretty much identical to stock and you are not producing much boost. If Subaru thought it was safe, I feel pretty comfortable it is. Also, it is widely believed that the stock knock sensor is not very good at distinguishing knock from noise below 3,000 RPM. I have my fine learning correction set to ignore anything below 2500 RPM and 1.25 load. - I would get so much low load noise/knock that my IAM would drop. Even after upgrading to E85, the knock is still there. Based on the above, I concluded it is just noise.

 

With that said, you could try to pull timing and see if that helps in your specific case.

[iNVAR]

Sounds good, I'll probably pull a degree or two and see what happens.

[NSFW]

Yeah, that's what stops me from doing road dyno stuff. I know of some level roads private tracks that aren't smooth, and some smooth roads private tracks that aren't level, but the only smooth-and-level road private track that I know of is almost an hour away.

[sjwelna]

So I'm pretty much done, except for two questions:

1) If I want to keep things smooth, I can't really add timing back in anything because that area knocks. I should just leave things as they are now, right?

 

If you only pulled a degree out, I'd just leave it alone and watch the tune to see if you start knocking again as the weather cools off. If you pulled 3* out (I see in one cell you did) and now you have consistently no knock, you can experiment adding back in a degree at a time and see if it starts to knock. This is the fun in slow tuning!

 

-Steve

[Spec B]

Off topic - was wondering if fast polling is causing some wierdness. I reviewed a few of my new logs and the rpm is not smooth in places - it sometimes drops on a WOT run from for example, 3212 rpm down in one cell followed by 3198 then and then back up? I started to freak thinking my 1500 mile old CM FX300 clutch was slipping. But now, looking at your logs, it looks like it does the same thing...either both our clutches are slipping or something is up with fast polling..

[iNVAR]

Yeah, I noticed that too. The drop is a very miniscule amount, and I'm almost certain my clutch isn't slipping. The clutch is only just coming up on 2 years old and has less than 15K miles on it. I don't drive much.

 

I'm going to hazard a guess that fast polling is doing it... I would theorize that it's now polling so much faster than it's able to catch the little inaccuracies of the speed sensor. I would think that a slipping clutch would cause much more than a drop of about 10-20 RPM.

 

Here's a snip from my own log:

time/load/RPM

2604 2.72 3192

2668 2.71 3249

2732 2.72 3266

2797 2.73 3297

2860 2.73 3353

2926 2.77 3334

2971 2.78 3418

3036 2.77 3484

3100 2.77 3437

3165 2.73 3534

3230 2.73 3566

3275 2.71 3554

3341 2.68 3626

3407 2.65 3644

3468 2.66 3663

3534 2.66 3760

3580 2.66 3745

3644 2.67 3784

3711 2.65 3823

I mean, it would REALLY suck if my clutch was slipping.

 

But it also happens at low RPM, so I'm guessing it's really just a peculiarity of the fast polling

 

0 0.17 2290

47 0.18 2295

110 0.21 2304

172 0.28 2264

236 0.36 2324

300 0.48 2335

350 0.64 2344

415 0.82 2288

478 1.09 2334

540 1.16 2287

605 1.16 2377

652 1.18 2336

717 1.21 2370

780 1.24 2403

845 1.27 2390

909 1.32 2418

973 1.36 2442

1021 1.4 2460

1086 1.44 2471

1148 1.49 2504

1212 1.54 2516

1277 1.58 2540

1327 1.63 2581

1389 1.68 2570

1455 1.75 2624

1517 1.8 2652

There's just no way in hell I'm slipping at such a low load, with less than 8PSI. If it was slippage like that, I would be slipping a LOT more at high load, high boost, high RPM.

[sjwelna]

Off topic - was wondering if fast polling is causing some wierdness. I reviewed a few of my new logs and the rpm is not smooth in places - it sometimes drops on a WOT run from for example, 3212 rpm down in one cell followed by 3198 then and then back up? I started to freak thinking my 1500 mile old CM FX300 clutch was slipping. But now, looking at your logs, it looks like it does the same thing...either both our clutches are slipping or something is up with fast polling..

 

Mine does the same thing too, I doubt it's all 3 of our clutches. Mine only has about 37k on it too. .

 

-Steve

[Spec B]

Great...the anomaly not a big deal considering all the additional data fast polling provides.

[NSFW]

Yeah, the RPM values are noisy. It kinda boggles my mind that the ECU can control timing at all when it clearly has only a vague idea of how fast the crank is turning. Makes no sense.

[sjwelna]

"It's what makes a Subaru a Subaru."

 

-Steve

[fahr_side]

Yeah, the RPM values are noisy. It kinda boggles my mind that the ECU can control timing at all when it clearly has only a vague idea of how fast the crank is turning. Makes no sense.

I doubt that the snapshots of data are at exactly equal intervals. I also doubt that logging gets the highest priority on the processor.

[NSFW]

I doubt that the snapshots of data are at exactly equal intervals. I also doubt that logging gets the highest priority on the processor.

 

You're correct on both counts, but those aren't the problems. I've been studying the ECU software lately... The RPM value that the logger gives you was read from the ECUs RAM, and the ECU uses the very same number to do lookups in the timing tables (and any other tables with an RPM axis).

 

When you log a WOT pull, that number doesn't just increase at an unsteady rate, it's actually common for it to decrease from one sample to the next. The increases more-than-compensate for those occasional decreases, but still... So, the actual ignition timing advance has got to be fluttering as well.

 

And yet the cars run fine. Go figure.

[iNVAR]

Well, each row of the timing and advance tables covers 400 RPM, at least on my table. The only time timing will decrease when it should be increasing while RPM is going up is if it JUST so happens to decrease JUST enough to cross that 400RPM border.

 

So a blip from 2430 to 2401 RPM isn't going to change the timing because they both fall into the same cell but a blip from 2430 to 2399 will. I would wager that the occurrences of that happening are fairly rare, and even when it does happen, the timing should be close enough that it won't matter much.

[NSFW]

The values you see in the tables aren't really values of cells in the way you're thinking. It's really more like the values of the corners between cells.

 

It's easier to visualize in 2D. I picked the table below at random and opened EcuFlash's Graph pane on it. Notice how the line segments between the points are sloped. If it worked the way you describe, the graph would look like a staircase rather than a curve. The ECU interprets tables in the same way that EcuFlash plots them.

 

It's not a big deal, since the RPM variations you see in the logs are pretty small on a percentage basis. But still, it just bugs me on an intellectual level that the ECU thinks that RPM flutters that much. Even more alarming (to me anyway) is the possibility that RPM really does flutter that much. Maybe the noise in the RPM parameter just means that the crank accelerates and decelerates that much between combustion events.

[iNVAR]

So what you're saying is that the ECU interpolates the values in between? Hrm, interesting... Disregard my previous post then!

[NSFW]

Exactly.

[iNVAR]

After working on the tune constantly for the last week and having put more miles on each tune through varied conditions (traffic, stop and go, normal city cruising, highway, aggressive, etc.) I'm at this point with my tune, compared to the original stage 2 from infamous. I removed a bunch of timing from 1.05 because of a consistent learned knock, even though it's below 3000RPM. It appears to have eliminated the knock, but I'll have to drive some more to say for sure. As such, I don' believe it was due to noise.

 

Again, thanks everyone for the help so far!

[Spec B]

^ your are officially a tuna now! Woohoo.

[iNVAR]

Haha, yeah right... Big thanks especially to you.