PorKen

One side intake plenum cylinders doing most of the work?

(EDIT: No, it's a software problem I created, sorry. Some good info here by others, though.)


As part of an upcoming major update for my S4 chips, I have been working on knock-feedback self-learning WOT ignition advance maps. There is only enough free semi-permanent memory (SRAM) for two maps with enough detail to be useful (64 rpm divisions), so I split the maps to serve the right, 1764, and left, 5832, distributors...wouldn't you know, each intake plenum feeds the same cylinders as the distributor on the same side. (Firing order is 13726548.) The results of a week of learning was surprising. More importantly, the car consistently logs the same or better as the fastest times I have recorded in two years of results from my local 'logging road', even after repeated runs. From previous experience, I expect it will dyno as well as the best visits there, if not a touch better.

The right plenum/distributor cylinders seem to be filling a lot better, requiring less timing. The left side plenum cylinders less so, handling a lot more advance. (The left side is the one with the cover too close to the intake horns.) For much of the map, the left side won't knock at all, except at the torque peaks, ~3100/~4200/~4600/~5200, and at max HP, 5600~6200. As the engine heats up, the right side becomes more knock sensitive - be it intake harmonics w/air temp and/or hot areas in cylinder.

The factory EZK WOT map applies to all cylinders and can only support the lowest common advance. In fact, the stock WOT map looks kind of like the lower line, smoothed out, but there are areas that have too much advance, which benefits the left side, until the engine heats up and the right side knocks.

I don't think this is related to cam timing ± per bank as both cylinder banks are fed by each plenum.


I would like to get a set of intake spacers to see if the left side can breath better = less advance. If anything, the right side spacer might be best left out or be a dummy/filled in. (Previously, plenum spacers have not shown a consistent benefit, but that may have been due to stock-tuning limitations.) The right side also does not seem to like the transition state of the flappy at ~5500? I should try changing the switch point rpm a bit.


Relative maximum advance per side at full throttle. Left side advance, blue upper line, is capped. Higher doesn't knock or make more power.


MAF voltage, green and AFR, purple, have also further evened out (no smoothing used, measured in a balance pipe before X).
Injector duty, red, tends to follow the horsepower graph on the dyno. RPM, black, flattens out at 6300 due to a soft rev limit (retard).

PorKen

It has been frustrating to try and get more power out of the S4. Advance the timing and it will gain HP...sometimes.
Oddly, the left plenum needs less timing at the low end and the high end. The high end is annoying for peak HP, but now the right side may pull up the slack a little.


I expect the story is the similar for GT and GTS, if it is due the left plenum volume/shape.


(64-column) dual-sided WOT vs (16-column) factory ROW non-cat EZ WOT map. Right/black line is similar if the factory columns are connected?

joejoe

Jeez Ken you make me feel dumb. (no great thing as my 3 legged husky can do they same) I have no idea what I'm looking at or how it relates to how the motor runs. But it does look cool. This has obviously been a great disappointment and now it is seen there is no reason to go further. I believe it is time to give up this unneccessary avenue and get back to the s3 model. I have one here that you could get acquainted with (I am sure you have forgotten what they even look like)

Truthfully I hope you are making progress and that all is going well in your area. (keeping tabs on this one to try and learn)

Tom in Austin

Fascinating, keep it coming Ken!

MjRocket

Following

fraggle

I almost wish my compression was stock so I could buy these!

worf928

UpFixenDerPorsche

bump

ptuomov

Ken --

If it's a simple flow restriction, wouldn't the impacts grow with air flow demand (i.e., rpm)? To me the difference between banks going away at higher rpms suggests that it's not a simple flow restriction that impacts all runners in one plenum and none of the runners in the other plenum, something else must also be going on.

Do you have a table of knock probabilities / relative frequencies by cylinder by rpm from your logging? If you have that, we can all stare at them and see if it's really the plenum that matters or something else. I have a hunch which I can rule out or not if I can see that table.

The runner lengths:

https://rennlist.com/forums/928-foru...ml#post6151810

1: 19 cm ~ 20 cm, passenger side plenum
2: 27 cm ~ 30 cm, driver side plenum
3: 28 cm ~ 30 cm, ds
4: 23 cm ~ 20 cm, ps
5: 20 cm ~ 20 cm, ds
6: 32 cm ~ 30 cm, ps
7: 28 cm ~ 30 cm, ps
8: 23 cm ~ 20 cm, ds

PorKen

I'm sure that one or more cylinders are affecting the whole group. Perhaps different ones at different RPM.

The only thing I can imagine is that it might have something to do with resonance + airflow. That is, there needs to be enough room to pack the air down the horn? Is it possible the right side is 'stealing' from the left? Half the time the plenums have only an indirect connection through the TB housing, though.

This is assuming of course that it is not a factor of EZK timing or some other flow problem which demonstrates as increased knock resistance on the left plenum.

I only log the knock light output and later transcribe what the EZ has incremented at particular RPMs (abusing the four EZK info outputs of the Theo diag box).

I have been looking at splitting again the WOT map into four groups of two cylinders (as above), which might be more telling. This will allow only a ca. ±16° adaptation in 1.1° increments over whatever base map I supply, but I am convincing myself that is probably OK. It might be good to separate the groups into 20s and 30s, but I only have left/right (distributor) and front/rear (knock sensor) flags to work with. Maybe I can figure out the logic for that...

ptuomov

So here are the runners:

2: 27 cm ~ 30 cm
5: 20 cm ~ 20 cm
ds plenum
3: 28 cm ~ 30 cm
8: 23 cm ~ 20 cm

1: 19 cm ~ 20 cm
6: 32 cm ~ 30 cm
ps plenum
4: 23 cm ~ 20 cm
7: 28 cm ~ 30 cm

That's a nice way to present Rob Edward's data without jpgs!

I think that the Sharktuner allows one to log the knocks by cylinder and then one can grid them based on rpm from the log. Hard to make progress without that data set, at least with the creativity that I have left at 4:05pm on Friday afternoon.

It may still be that it's just the cam timing. When the flappy is open, only the long runners are (based on a priori reasoning) likely to knock at the lower and mid range rpms. The long runners lead to better filling at the mid range rpm for the same reason that long runner manifolds make more torque at mid range rpm. Therefore, we can pretty much guess that it's cylinders 6&7 filling the best and knocking in the mid range rpm in your case. Now, they may knock because the runners 6&7 flow the best or because the cam timing is a bit off between the banks and the whole driver side bank fills better. It's just that the short runners don't knock at that mid range rpm at any cam timing.

FredR

This dialogue seems to reflect my experiences with ST2, exhaust mods, x-pipe and Carl's manifold spacers.

I discussed my findings extensively with Jim/John on ST2 and with Ken on cam timing/tuning in general. my cams were timed with gauges and I found the readings with Ken's cam timing tool spot on.

Before I fitted the spacers it seemed that knocks events were somewhat random and were likely to occur in the flappy transition range and then from about 5200 to 6000, predominantly on cylinders 2 and 6- more so on 6. For high load/high rpm cells I targetted AFR's in the 12.5 region but I could not see a relationship between onset of knock and AFR in the range 12 to 13.2 or so. in the end I selected 12.5 as my target average value at the top end of the scale.

After fitting the spacers [both sides] it seemed to me that onset of knock was predominantly on No 6 cylinder and I needed to pull the timing back to about 24 degrees to reduce the knock retards to about degrees when they [inconsistently] occurred [as advised by Jim Corenham].

I was concerned that the oil lifting and running mineral oil might cause knock but it did not explain the preponderance on No 6. Having modified my breather extensively and having oil consumption drop noticeably, I am thinking of putting ST2 back on in our current cooler weather and test to see whether I can get any further advance dalled in.

I also wondered if my current set up can be improved by cam retardation [more advance?] but I concluded that my current setup had in effect increased airflow on some cylinders given my assometer suggested that Carl's findings when fitting the spacers [ a bit more oomph from mid range upwards was reasonable. Just wish I had access to a dyno.
This all leaves me puzzled as to whether having to pull back timing to tame knock was a good thing because of more airflow or bad because other less efficient cylinders are suffering as a consequence.

I understand the idea of different runner lengths is to spread the power band and prevent the motor from being "too peaky". My conclusion after my efforts was that Porsche did not leave too much on the table but there is something to be had.

There are so many interactive variables it takes quite a good mathematical mind to understand the machinations and even then not sure it is something home DIY tuners should undertake lightly.

One thing is for sure, if Ken is struggling to figure all this out it is no wonder mugs like myself struggle a bit.

I have wondered if with my 30lb injectors I can/should set the transition point for the injectors to 6500 rpm [i.e. beyond the rev limiter] and whether the flappy transition at 5500 is inducing knock to some extent.

Regards

Fred

I

PorKen

Cam timing might then be a factor, then? I forget that this car came with a Racing belt, which FAIAP does not stretch.

My theory is that the regular belts stretch at higher rpm, retarding the cams, especially the right, 1-4 cams - a sort of variable valve timing. Also, as the engine warms, the 1-4 cams advance about 2° as the cam gears move away from the crank gear. Belt stretch may take up some of this effect, so it may actually be more, so a Racing belt might need more than 2° retard on the 1-4 when set cold.


I am now driving with the 'WOT/8' code (4 base + 4 learned/overlay 64 cell WOT spark maps ). With only a handful of runs it is starting to show some small differences between the four corners, but the cylinders are still mixed, long/short - 6/1, 7/4, 2/5, 3/8, so I don't expect huge differences.



As an aside, I see the advantage of having an open differential in the rain. Too impatient to wait for dry roads I tried my second gear hill climb WOT road again without problem. With a LSD car this usually meant a scary ride with the back end trying to come around when hitting the torque peaks, but the open diff's single spinning tire is a fuse which keeps the car straight.

ptuomov

Since you are chasing the difference between banks and plenums, I don't think the belt stretch is a factor. In my opinion, you can forget about the belt stretch. The tension between the cam gears isn't going to be that large. Just my instinct, for what it's worth.

We don't even know for sure what the right questions is. We have to guess what the question is. The most likely question as far as I understand is

Why are the cylinders 6&7 in the driver side bank that are fed with long runners from the passenger side plenum filling better at mid range than the cylinders 2&3 [was incorrectly 5&8] in the passenger side bank that are fed with long runners from the driver side plenum?

There are a number of hypotheses why this is happening, most (all?) of which have already been mentioned in this thread:

1. The runners 2&3 [edit, was incorrectly 5&8] are blocked by the plenum walls and the plenum cover. This should impact steady state flow and if it is the issue one should be able to diagnose it with a flow bench. With the covers on, 2&3 should flow less than 6&7 on a bench. If this is the problem, installing a plenum spacer on the driver side plenum, opening up the areas around the 2&3 bellmouths, and restoring the plenum volume with epoxy would be the way to correct the problem. See more in this post: https://rennlist.com/forums/928-foru...l#post10455177

2. The driver side cams are timed to produce a better filling at the mid range rpms than the passenger side cams. Furthermore, as the block heats it expands. This expansion advances driver side cams a little and passenger side cams more (and is by the way the only reason why the cold cam timing is set slightly differently between the banks). One should be able to figure out whether the knock patterns between banks as a function of advance grows or shrinks with block heat expansion in a way that is consistent with this hypothesis. I say "one should", because I am not going to try to twist my brain into a pretzel trying to figure that out! ;-)

3. There is some plenum resonance effect that makes the passenger side plenum fill the driver side bank better, especially cylinders 6&7 with long runners. Since the flappy is open at mid range in your test runs so far, the manifold works as a single plenum manifold with pulses mostly cancelling out. Therefore, the plenum resonance effects are relatively unlikely. However, it might be worth it to run the who rpm range with the flappy closed and to see if it is still the case that the passenger side bank fed from the driver side plenum can take much more advance before knocking.

To make progress on this with the car running, I think there are a couple of things to do to help diagnosing the issue:

A. Create a knock probability table, by cylinder and by rpm range from data logs. Do this separately with flappy open and flappy closed for the entire rpm range.
B. Separately measure the AFR from the different bank exhausts by rpm range. Do this separately with flappy open and flappy closed for the entire rpm range.

From those two data sets we should be able to make some educated guesses about cylinder filling at each rpm.

I know this is not very helpful, since it amounts a shot into the dark that involves a lot of work.

PorKen

Note it is the center cylinders that are the 'long' ones, 6&7, 2&3. The four corner cyls, 1458, are the noodly 'short' ones.

The pax/right plenum has the 6&7 horns on the outside. In the left plenum the 2&3 horns are on the inside. This might have something to do with the filling? 6&7 have effectively even longer tubes, including the plenum distance to the feed passage.