AFRs swinging at light load in CLL.

Post Reply
dr_jones
Posts: 162
Joined: Wed Jun 08, 2016 3:24 pm

AFRs swinging at light load in CLL.

Post by dr_jones »

Hi - I have a situation currently where i'm seeing constantly AFRs swinging around 5 points or so at steady light load and rpm (say around 3K rpm and 800mb for instance). Looks a little like narrow-band AFR cycling but this is under wide-band control.

I can't immediately identify if this is intentional but I suspect not - would I be right in thinking this it related to the proportional & integral gain setup? Or is it more likely related to the Service Interval Cal (which for the records I currently have not populated with 'real' data).

-Ed.
RICE RACING
Posts: 448
Joined: Mon Feb 24, 2014 1:08 am

Re: AFRs swinging at light load in CLL.

Post by RICE RACING »

This will help you.
https://www.youtube.com/watch?v=JEpWlTl95Tw

The service interval needs to be set up to suit your engine and location of sensor
Same for the gain tuning of the P and I for various MAP and RPM points

In my experience in looking at allot of these from various sources I am yet to see one that works well, they either are way too slow never reaching target or unstable with over and under shooting wildly. It does take some time to make it work properly (read my post on getting familiar with the system and costs passed onto customers), once set up correctly though its excellent.

Below are some parameters I log and display when looking over it to see what is happening.
RIGHT click and view image to see full size
Image

p.s. One particular car I took on the fuelMltCll was so far out that the poor ECU would be battling constantly to meet the target Cll, while I could not personally verify it the customer stated to me after I had set it up the case temperature of his S12 went from being able to 'fry an egg on it' to just being warm after a ~20 minute drive.
http://www.riceracing.com.au
https://www.youtube.com/riceracingdonmega
Real turbo road cars fast > reliable > durable
Water Injection Specialist
dr_jones
Posts: 162
Joined: Wed Jun 08, 2016 3:24 pm

Re: AFRs swinging at light load in CLL.

Post by dr_jones »

Thank you - that's very helpful!
dr_jones
Posts: 162
Joined: Wed Jun 08, 2016 3:24 pm

Re: AFRs swinging at light load in CLL.

Post by dr_jones »

So I've begun looking lookign into tuning the CLL control and have come up against something unexpected when gathering data to populate the Service Interval

These logs are taken with CLL disabled and I've used the global multiplier @ 1.5x over a base of 1.0 to step modify the fuelling:

At 1K rpm I get exactly what I'd expect - the only question is where on the falling lambda reading am I best to measure to from the step up in fuelFinalPri1? Here I've used the moment the sensor detects the falling AFR - but of course i could use the point it levels out or anywhere in between...
SI-1K.JPG
SI-1K.JPG (107.42 KiB) Viewed 22024 times
At 3K rpm I see something very unexpected - a sudden increase in AFR as the global multiplier is increased prior to the expected drop - any ideas??
SI-3K.JPG
SI-3K.JPG (104.68 KiB) Viewed 22013 times
Last edited by dr_jones on Tue Nov 29, 2016 4:01 pm, edited 1 time in total.
stevieturbo
Posts: 1321
Joined: Sun Dec 07, 2008 2:04 pm

Re: AFRs swinging at light load in CLL.

Post by stevieturbo »

Without units or scalings etc...it's impossible to see what is happening on a screenshot.

IMO look for the time difference between fuelfinal changing substantially, the a corresponding change in what is seen at the lambda sensor and that should give some sort of starting point for that load site or approx load sites.

Obviously you'll need to be logging these things quite fast, say 100Hz. Doing it at say 10Hz would be a waste of time.

Also if it is oscillating badly....reduce the amount of correction it can actually do until you have things under better control
dr_jones
Posts: 162
Joined: Wed Jun 08, 2016 3:24 pm

Re: AFRs swinging at light load in CLL.

Post by dr_jones »

Ah - my mistake - have now updated the plots to show AFRs at least!

Still I'm very confused about what would cause the lean spike on the 3K rpm plot when we have closed loop disabled and the step change was an addition to the fueling.

I haven't tried to repeat the experiment but will do so and at an increased logging rate - however I expect I'll see the same just in more detail all other things being equal.

This is actually shown better at the trace taken from an initial 4.5K rpm the lean spike is very visible between the initial 12.0 AFR and the subsequent 11.1 ish AFR (obviously I need to dial in the base fueling!).
SI-45K.JPG
SI-45K.JPG (118.26 KiB) Viewed 22006 times
stevieturbo
Posts: 1321
Joined: Sun Dec 07, 2008 2:04 pm

Re: AFRs swinging at light load in CLL.

Post by stevieturbo »

If the lean spike does not correspond with a relevant change in injector pulse, or more air going into the engine, maybe it is a misfire ?

Presumably the engine does run ok ? or was the lean spike actually felt too ?
RICE RACING
Posts: 448
Joined: Mon Feb 24, 2014 1:08 am

Re: AFRs swinging at light load in CLL.

Post by RICE RACING »

Transient fuel is inadequate thus 'spike leaner than target' upon rapid change in map and tps, simple way to see this is do a math channel for 'derivative tps'
where you have a rapid acceleration change in tps this rapid transient effect on air density needs a matching transient additional fuel, which you will find under that sub menu, lots of engines have differing demands dep on where injectors are located etc.... it needs to be calibrated to suit.

Example below of one that goes richer than target with CLL 'auto' disabled.
Image
http://www.riceracing.com.au
https://www.youtube.com/riceracingdonmega
Real turbo road cars fast > reliable > durable
Water Injection Specialist
dr_jones
Posts: 162
Joined: Wed Jun 08, 2016 3:24 pm

Re: AFRs swinging at light load in CLL.

Post by dr_jones »

Thanks chaps!

This was taken with the car stationary - I can't say I heard a misfire or felt anything other that the rpm drop as the AFRs went down - so the transient fueling sounds correct.

TBH I was planning on addressing transient fueling after I'd finished the initial fueling calibration - however it looks like I should likely address this now or at least prior to attempting to further populate the SI table.

-Ed.

P.S. I'm starting to realize just how useful these maths channels are!
Post Reply