Help Analyzing S52 Dyno Graph

[ian e30 318is]

Did you check the camshaft range of operation with VANOS actuated?

https://www.youtube.com/watch?v=angrrw9TaRc

I did not, but wow that is ingenious! I’ll have to give that a try ASAP! Thanks for the link.

[dude8383]

Did you check the camshaft range of operation with VANOS actuated?

https://www.youtube.com/watch?v=angrrw9TaRc

[pazi88]

Yes if the vanos wouldn't work correctly it will throw dtc. But not necessarily if the cam(s) are degreed incorrectly. Non working vanos will cause loss of low end torque. Not high end.

[ian e30 318is]

so at 6100 rpm the AFR is the same but there is a huge difference in power. something else must have changed.

my thought regarding the OP was cam timing

I have a question for everyone involved in this thread. If my cam timing was physically not correct (off by a tooth) or VANOS was the culprit (not actuating), wouldn’t the engine idle/run terrible, throw codes, and not make power up to 5,500 RPM per my original graph?

I talked to a BMW tech and had him come look at the engine with the valve cover off and everything looked appropriate at TDC.

I’m thinking that the air filter I currently have installed is too small to allow the correct amount of air into the engine per a surface area calculation, so I’ll get another on order that is much larger. My intuition is that the engine may not be receiving enough air since the AF mixture at WOT is fairly rich.

[digger]

The red line is first pull with only 21.5lb injectors added. Blue line is after I trimmed the fueling. It would have made more power if the right of the AFR graph was at stoich at the rev limiter, but as I said, I like to run the road racing cars a little richer than for all out power.

You should keep the AFR's on the mid to high 13's with a little dip at peak torque. Timing on most 24's I have tuned didn't gain MBT after 27°, but this one actually kept gaining torque until 33°, but left it at 30 peak.

You don't need a 3.5" MAF. They don't make more power than a 3". An LS1 has a 75mm MAF (2.9") and makes power well into the 400's. Your engine is restricted at the valves. Even a BBTB doesn't show peak power gain, just a tiny bump in the curve at throttle opening/transitions, naturally, as it "gulps" a little more air earlier than a stock TB. Having a MAF larger than the throttle opening is moot.

so at 6100 rpm the AFR is the same but there is a huge difference in power. something else must have changed.

my thought regarding the OP was cam timing

[ForcedFirebird]

As stock s52/m52 go even to 12.5:1 AFR. Not over whole rev range but still. It's even clearly seen in the stock tune and not engine problem. But it's true that the modern engines run close to 13.5 AFR for obvious reasons.

I have a stock s52 .winpep file somewhere. Pretty sure it rarely dips close to 12.5, if at all. The pent roof design of the 24v isn't as efficient as the m20, the m20 will run over 14afr in areas at peak performance, but the 24v make that up in flow and cylinder filling.

[ForcedFirebird]

The result is an approximately parabolic curve of lambda vs. torque with the peak around .9 lambda. Between ~ .7 and 1.05 there is never really more than 5% difference in torque between the highest (~.9) and lowest (~.7 and 1.03) points on the graph. However, once you get outside this range the slope does start to increase more dramatically and torque starts to drop off very significantly.

I stand corrected. Went through the dyno and .bin files today and the TQ didn't change much more than you said. It was closer to 8% going from 11.x to 13.3afr, the power was really gained after MBT tuning. ;)

[pazi88]

yeah, I disagree. Modern BMWs do not run 11:1. The richest they go is about 0.8, and it's only to cool the cats at high load - which hurts power quite a bit. There's just no reason to go that rich if you care about power. Saying that 11:1 would only lose 5hp is also demonstrably wrong. I lost 5hp just by going from 13.5:1 to 13:1 on my N52.

As stock s52/m52 go even to 12.5:1 AFR. Not over whole rev range but still. It's even clearly seen in the stock tune and not engine problem. But it's true that the modern engines run close to 13.5 AFR for obvious reasons.

[ForcedFirebird]

I agree it's a huge loss up top. I may need to test the fuel pressure as you mentioned. I have a fairly new Euro HFM, and the tuner is aware that it is not the 540 MAF. The VANOS is also broken in as far as I know. I have driven the car for a few hundred miles now.

Would be interesting to know if he has the proper MAF calibration tables for the Euro unit. This is yet another reason I don't like the larger MAF's. The stock 3" can handle more than enough power for an NA car. The graphs I posted is using stock MAF and Ford 19lb injectors (rated at 2.5bar, running at 3.5bar).

I still don't think that the OP's issue is caused by AFR, but data logs are need to help here. I don't want to venture too far off topic but I wanted to address a couple points to try to provide some clarity on the air fuel ratio discussion.

I disagree about your take on power and AFR's. I have experienced many times as part of my living is made by putting cars we build on the chassis dyno for fine tuning. Also, HP/WT restricted racing classes need the sheets for proof in racing sanctions. I do agree without logs, we don't know what's happening.

I can't speak to you specific scenario as I don't know what else may have changed in your cal or dyno setup. I'd also point out that it looks like you are measuring lambda at the tailpipe with the dyno sniffer, where readings aren't particularly accurate. O2 sensors tend to read leaner further away from the engine. Additionally, there could be a small exhaust leak or numerous other factors.

The car has mid length headers, 2" drue dual exhaust with an x-pipe 3' from the collector where there's an on-board wide band, the dyno 02 AFR's matched in my case - well, within .2-ish AFR's. I have been using the same dyno for about 8 years now, never had mismatched AFR's to a degree that causes any concern.

I'm a calibration/controls engineer at an OEM and I have access to tons of data to support my claims. I'm not trying to start an argument or anything, just trying to help with some useful info. All production engines have lambda sweeps performed at steady state conditions during the mapping process, with all other parameters held constant. These tests are done under perfectly controlled conditions in engine dyno cells with calibrated lab grade equipment, something that could never be properly replicated with a vehicle on chassis rolls. The result is an approximately parabolic curve of lambda vs. torque with the peak around .9 lambda. Between ~ .7 and 1.05 there is never really more than 5% difference in torque between the highest (~.9) and lowest (~.7 and 1.03) points on the graph. However, once you get outside this range the slope does start to increase more dramatically and torque starts to drop off very significantly. Once you get down to anywhere below .65 you are around the combustion stability limit of gasoline and you will be way down on power and/or misfiring or having partial burns. I can't share propriety information, but there is plenty of information available online showing similar curves.

I don't have any papers showing education, but I do have long time real world experience. My shop builds a lot of engines and heads, and we document before/after results. I approach this very scientifically and back up my theories with proven results.



This isn't an argument, it's conversation. :)


Sure, .7-1.05 lambda (~10.2:1-15.3afr) doesn't change the torque much, but based on the OP's curve, the peak torque was close to being reached, then drops off more than it should.


Not sure why the fact it's a chassis dyno, that it must be that far out of calibration. The numbers don't lie. I can isolate the pulls and explain what was done at each one, and you can clearly see the numbers. The winpep files are here at the shop and could put together the runs where the fuel curve was first flattened, then mean brake torque was found with the timing curves. In fact, I saved each .bin file and named it after the dyno pull, so I can match the dyno curves with the fuel/spark files.

My suspicion based on your dyno graph would be that your o2 sensor is reading leaner than it should. So while your o2 reading tells you that you're going from .75 to .85, you are probably going from something like .65 to .75, in which case, the big jump in power make much more sense since you are on a part of the lambda curve that is quite steep. If you had just swapped from the S50 17.5lb injectors to 21.5lb injectors with no calibration changes you should expect to be running about 22% richer than stock, so this makes sense.



That is just not correct. Most places in the US pump gas is now E10, which has a stoich AFR of around 14.1. 11:1 would be .78 lambda, perfectly within the combustion stability limits of pretty much any modern engine. In fact nearly every modern engine with close coupled cats will enrich down to around 11:1 on the factory calibration to control cat and or turbine inlet temps when under sustained high loads.

OK, sure it can run there, but power will be down based on my experience with the chassis dyno. 11.1 isn't uncommon on an above atmo engine, that's a whole different conversation.

[nando]

I didn't say they will not run. I said pump fuel doesn't burn efficiently at that ratio.

Anyway, not to derail the thread - I also agree that it's not the *primary* factor for the low power, and certainly not for that big drop off at 5500rpm.

personally, I would be looking at cam timing / vanos.

[R3VM3UP]

I didn't say 5hp, I said 5% torque at any given rpm, i.e. 5% power.

Naturally aspirated engines often won't see much below .8 lambda in most cases, as there is no reason since exhaust temps typically do not dictate it.

Boosted engines on the other hand, even DI engines, frequently go below .8 under sustained high load in my experience. This is due to a combination of factors, the biggest being that they are more knock limited, and run less spark advance, resulting in higher exhaust gas temperatures. I brought up this point mainly to say that engines will run at 11:1.

I don't think we are necessarily in disagreement here. I should have clarified that it's not common for NA engines to run that rich.

[nando]

yeah, I disagree. Modern BMWs do not run 11:1. The richest they go is about 0.8, and it's only to cool the cats at high load - which hurts power quite a bit. There's just no reason to go that rich if you care about power. Saying that 11:1 would only lose 5hp is also demonstrably wrong. I lost 5hp just by going from 13.5:1 to 13:1 on my N52.

[R3VM3UP]

I still don't think that the OP's issue is caused by AFR, but data logs are needed to help here. I don't want to venture too far off topic but I wanted to address a couple points to try to provide some clarity on the air fuel ratio discussion.

The dyno graph I posted says otherwise. Look at the power difference when leaning it from 11 to 12.5 at 5500rpm. ;) In your thinking I went from ~.75 to .85 lambda at 5500 RPM, which is 10% difference from lambda. In most of the formats I have tuned, 10% adding/subtracting fuel is a BIG difference specially when OBD1 scales fuel at 0-255 as it's binary hex.

I can't speak to your specific scenario as I don't know what else may have changed in your cal or dyno setup. I'd also point out that it looks like you are measuring lambda at the tailpipe with the dyno sniffer, where readings aren't particularly accurate. O2 sensors tend to read leaner further away from the engine. Additionally, there could be a small exhaust leak or numerous other factors.

I'm a calibration/controls engineer at an OEM and I have access to tons of data to support my claims. I'm not trying to start an argument or anything, just trying to help with some useful info. All production engines have lambda sweeps performed at steady state conditions during the mapping process, with all other parameters held constant. These tests are done under perfectly controlled conditions in engine dyno cells with calibrated lab grade equipment, something that could never be properly replicated with a vehicle on chassis rolls. The result is an approximately parabolic curve of lambda vs. torque with the peak around .9 lambda. Between ~ .7 and 1.05 there is never really more than 5% difference in torque between the highest (~.9) and lowest (~.7 and 1.03) points on the graph. However, once you get outside this range the slope does start to increase more dramatically and torque starts to drop off very significantly. Once you get down to anywhere below .65 you are around the combustion stability limit of gasoline and you will be way down on power and/or misfiring or having partial burns. I can't share propriety information, but there is plenty of information available online showing similar curves. If you have a copy of Heywood's Internal Combustion Engine Fundamentals, there is also a plot of this on page 831.

My suspicion based on your dyno graph would be that your o2 sensor is reading leaner than it should. So while your o2 reading tells you that you're going from .75 to .85, you are probably going from something like .65 to .75, in which case, the big jump in power make much more sense since you are on a part of the lambda curve that is quite steep. If you had just swapped from the S50 17.5lb injectors to 21.5lb injectors with no calibration changes you should expect to be running about 22% richer than stock, so this makes sense.

at 11:1 pump fuel hardly even burns. We're not talking about E85 here.
though.

That is just not correct. Most places in the US pump gas is now E10, which has a stoich AFR of around 14.1. 11:1 would be .78 lambda, perfectly within the combustion stability limits of pretty much any modern engine. In fact nearly every modern engine with close coupled cats will enrich down to around 11:1 on the factory calibration to control cat and or turbine inlet temps when under sustained high loads.

[pazi88]

Sorry I missed that you were referring to ignition and understood they are different. I agree that tuning via email isn't ideal, but what choice do I have with an OBDI setup? Wouldn't the knock sensors throw a code if pinging was an issue? (not trying to be facetious)

Tuning remotely by the logs is fine. I even do that too. But in that case you must have logs recorded? Or if not and the tune is made by guessing, which is not so good...

This is from the hubs. The dyno reads about 10% low from what I was told, but still seems pretty low for the mods. I would love to data log if I kept the setup OBDII, but hindsight is 20/20.

From hubs you get about 15% drive train loss so it's now close to 250hp from crank. Which isn't far off, but that dip shouldn't be there. But without logs, it's impossible to say if that is tune related problem or some mechanical problem in the engine. the 12-11 AFR is too fat but it doesn't cause sudden dip in torque curve like that. It should just cause the hp output to be low through everywhere where it's too rich. And still not too much. maybe 2-5hp.

I personally have strange dip in torque curve at bit lower rpm which came there agfter driving the car for summer: https://pazi88.kuvat.fi/kuvat/Projek...918_122255.jpg

That is clearly seen in logs so that the engine just doesn't get normal amount of air in that point. So it's mechanical problem that I have not yet figured out.

[e30davie]

Ive no idea on obd1 how you tune etc.I'm just a home gamer on megasquirt. Yer hopefully the knock sensors kick in if required. But im not sure how much authority they actually have. Whos yo say that when it detects knock it takes 15deg out, and thats what you are seeing on your graph?

It sounds like you are aware but unless you tune the thing on the dyno you are really just stabbing in the dark. And i think your dyno curve is a testament to that.