Cylinder Skew Tuning - Ideas, Tips?

[ForcedFirebird]

EMC = Engine Master's Challenge.

It's a contest where the contestants have to use pre-1985 heads and carb. manifolds to scientifically extract as much power as possible on a brake dyno...

http://www.popularhotrodding.com/enginemasters/challenge/0901em_2009_engine_masters_challenge/index.html

Sorry, I was under the impression that they had reached nearly 3hp/ci, but wasn't that high. Still impressive to see almost 600hp from a 380ci engine (2.6hp/ci), that still equates to a 400+ hp N/A 3400.

[robertisaar]

from what i understand, EGTs also have a bit off a bell-curve issue...

[3400beretta]

Worthless? Is that why they are commonplace on the EMC motors making ~3hp/ci? Translate that to a 3400 and you would be talking over 600hp.

I can't find specfically what you are talking about, care to elaborate?

Chances are, they already know EXACTLY what temperatures they want, and they know what temp = what a/f ratio for that engine.

And yes, the more that I think about it, you could use egt's on each cylinder, combined with a single wideband. As long as all egt's are reading the same, each one would be more or less equal to the wideband reading. But you would still need some monster transmitter that is capable of reading 6 thermocouples at once, and display all that info, preferably output to datalog.

However, the OP wants to fine tune each cylinder, by changing the pulse width on each injector. You could be there all day waiting for those thermocouples to react to each minute change. I just don't see how you could get accurate enough results with thermocouples to even warrent this experiment. Like we were estimating that each cylinder should already be within 5% of the next.

[SappySE107]

Fixed

[robertisaar]

that's the same link twice....

and i think i have the second to last revision of that book...

[ForcedFirebird]

Actually, an EGT setup in this application would be pretty useless. First of all, thermocouples are rather slow to react. Second, they are not deadly accurate, and definately need to be calibrated in order to provide any satisfactory results. I doubt you could find a transmitter that is even calibratable for an automotive use. There is no such thing as *accurate* self calibration, it is impossible, a marketing gimmick. Anyone who has worked in Instrumentation would know this. And even from one thermocouple to the next, there can be huge variances.

The above pictured application is rather silly as well, since it is a carburated setup, you can't even fine tune each cylinder!

EGT's used to be a poor mans wideband, but with today's prices.. anyone can have a wideband. And for what you want to do here... wideband is definately what you need.

Worthless? Is that why they are commonplace on the EMC motors making ~3hp/ci? Translate that to a 3400 and you would be talking over 600hp.

[SappySE107]

Ideally you would have an exhaust gas analyzer on each cylinder to measure the chemical properties of each. This is probably well beyond what you need, hence the EGT is brought up. This will tell you descrepencies between cylinders, but you will need to know power output and still use a wideband as an average of values (Dyno session). You can skew the injectors then to get the temps in line but you need to know what temp = best power, and the wide band is good to make sure you don't run it excessivly lean while getting it setup. There is a specific measurement you can take from the chemical make up but its been at least 6 years since I was looking that up. I thought it was nitrogen content is 5% for ideal but thats a vague memory.

You can run EGT on carb to check consistancy, even if the carb isn't going to allow you to tune for each cylinder. You could change the valve lash or rocker ratios or maybe even spark plug heat range. Physical changes can be made to offset conditions, but this doesn't help for the OP:P

You would spend considerably more for quality wide band sensors and the controllers to keep track. Again, quality, not the cheapest possible. EGT is a lot better bang for the buck, especially once you have it setup so you can check changes over time from dirty injectors, poor valve seat sealing, excessive blow by from rings, and other physical issues.

I suggest getting some books on tuning if you want to know more in detail.

The book I am reading now.
The book I have to read next.

Google exhaust gas analyzer and maybe you can find the performance aspect. I only did a quick search and didn't find what I was looking for.

[AleroB888]

Just shoot a laser thermometer at the top of each header pipe.

[3400beretta]

Actually, an EGT setup in this application would be pretty useless. First of all, thermocouples are rather slow to react. Second, they are not deadly accurate, and definately need to be calibrated in order to provide any satisfactory results. I doubt you could find a transmitter that is even calibratable for an automotive use. There is no such thing as *accurate* self calibration, it is impossible, a marketing gimmick. Anyone who has worked in Instrumentation would know this. And even from one thermocouple to the next, there can be huge variances.

The above pictured application is rather silly as well, since it is a carburated setup, you can't even fine tune each cylinder!

EGT's used to be a poor mans wideband, but with today's prices.. anyone can have a wideband. And for what you want to do here... wideband is definately what you need.

[ForcedFirebird]

Ninja'd by Brad I was going to suggest the same.

[bszopi]

EGT would probably be better than WBO2.

http://www.stevetek.com/R-EGTprobeLocation.html

http://www.powertraincontrolsolutions.com/egt.php

[Schmieder]

honestly, you wouldn't notice much... our intakes are decent for equal flow distribution, probably less than a 5% difference between the best and worst cylinders, but i won't stick to that number since i'm just going off of general design.

as far as the datalog: i don't know how fast the OBD2 PCMs push out a datastream but consider that at 2000RPM, that's 60,000 4-stroke cycles per minute. unless you can log at 60K samples/second, you would never know which cylinder is getting which pulsewidth, unless the datastream pushes out a BPW # for each cylinder...

Unless we can
(1) Gently skew the cylinder that is in question a very specific amount and hold it there.
(2) Let it idle collecting data for a specified time.
(3) Record the difference in A/F Ratio.
(4) Repeat for remaining cylinders
(5) Compare results

Hopefully the same skewing across the board showing greater or lesser changes in A/F ratio would indicate which cylinder is richer or leaner.

Because a leaner firing cylinder would have less A/F increase per equal skewing amount.

So the variable differences should indicate which cylinder is leaner or richer.

But again, the level of accuracy would make or break such a process.


Tell you guys what, I'll try it out when I get the car running with HPT or DHP.

[robertisaar]

honestly, you wouldn't notice much... our intakes are decent for equal flow distribution, probably less than a 5% difference between the best and worst cylinders, but i won't stick to that number since i'm just going off of general design.

as far as the datalog: i don't know how fast the OBD2 PCMs push out a datastream but consider that at 2000RPM, that's 60,000 4-stroke cycles per minute. unless you can log at 60K samples/second, you would never know which cylinder is getting which pulsewidth, unless the datastream pushes out a BPW # for each cylinder...

[Schmieder]

honestly, that's about the only way to truly do it... any other method just doesn't get a perfect picture the same way one sensor per runner would...

I was hoping there was a formula of some sorts. Like data logging injector pulse variations with WBO2 readings.

If a cylinder is richer then the rest, it should have a stronger influence on total A/F ratio. Where as a cylinder that is leaner would have a lesser influence of difference in combined A/F ratio.

But at that level of data logging, considering the margins of error, I think the independent WBO2 data logging would be the best route.

I don't recall ever hearing of such a thing. Has this been done before effectively? It would be nice to research the methods used and how it was done.

What would be the result of a very well balanced, skewed, engine? Is the difference unnoticeable or will it provide a smoother idle and stronger performance?

[robertisaar]

honestly, that's about the only way to truly do it... any other method just doesn't get a perfect picture the same way one sensor per runner would...