AFR transitions - best way?

coyoteboy · Post by **coyoteboy** » Thu Jun 15, 2006 3:12 am

I've been tuning away for a while now, getting some amazing torque gains through the mid range and top end power seems un-ending. I'm running 11:1 roughly at WOT and 150Kpa+ but watching the logs and looking at the AFR tables i see my AFRs in the logs go through a danger zone before reaching 11:1. 14 or 13:1 at 3Krpm (spool up time) on the way to 150. Obviously the AFRs cant jump in steps, that would be silly, but is it a case of 'by feel' tuning to see whether those moments passing through lean with fairly high boost, or should it be 'by X Kpa you shouldnt be seeing leaner than 12' for example? I'm getting no knock/det and the denso ECU isnt pulling timing.

J

Bernard Fife · Post by **Bernard Fife** » Thu Jun 15, 2006 6:44 am

CB,

You should not use the WB sensor to tune transient conditions. Bruce has information indicating that such sensors have a respnse time of about 0.3 seconds, which is much to slow for things like accel enrichment and other changing conditions.

And yes, AFR can jump in steps, it only takes two revolutions (on a 4-stroke) of the engine for all cylinders to react to a large change in fuelling, such as what occurs in accel enrichment. 2 revolutions takes just 0.066 seconds. So you would expect an accel event to look nearly like a step-function, given the resolution of a datalog (and ignoring the response curve of the sensor).

Lance.

coyoteboy · Post by **coyoteboy** » Thu Jun 15, 2006 7:05 am

I have some data somewhere saying that the worst case response time for the bosch wideband that comes with the LC-1 is 100ms. And after the gasses have been mashed through a turbine, through a wastegate and thrown round some bends it would undoubtedly blur any distinct change anyway so it will be an average.

Trimming all that down in an edit:

Some info on the sensor mixed in with the 'sampling rate' of the LM-1 there.
http://innovatemotorsports.com/forums/s ... =transient

I have a feeling i made that very hard to understand. What i mean, in escence, is that if the VEs for 3000rpm locked on a dyno are right at 150Kpa constant load, and *wrong* at 3000rpm 150Kpa on the way to 7000 during a drag run (or even light throttle) - how does one correct this, assuming its a comparatively long time since the MAP altered and so the MS does not understand it is accelerating. .

J

keithmac · Post by **keithmac** » Thu Jun 15, 2006 11:21 am

Maybe RPM based accelleration enrichment is the way to go?

coyoteboy · Post by **coyoteboy** » Thu Jun 15, 2006 2:38 pm

Im using it already, but its still only active shortly after the MAP reading or TPS changes. If I put my foot down hard on the throttle and leave it down you get no MAP variation past 3Krpm and no throttle variation - hence no AE. I could set the AE to last for ages but that makes normal driving impossible.

J

mops · Post by **mops** » Thu Jun 15, 2006 4:06 pm

i'm not sure whether i'm understanding.

you are saying that at stable 3krpm 150kpa afr's are stable 11:1
but when you are durnig 'normall' acceleration conditons, i.e.

say 2.5krpm to 6krpm at WOT, the afr's are less ?

I'm experiencing similar issues in NA engine, but not to that extent. Basically I noticed that the every ve bing afr are different if I:
1. very slowly accelerate through it (either slow rpm or map changes - slow engough to aviod AE) - leaner than constant
2. constant cruise on one point of ve table - constant basically
3. deccelerating through that ve point slowly produces richer afr's than on constant cruise.

it is therefore difficult to tune engine at the edge of misfires, because slow throttle movements cause it to miss...

coyoteboy · Post by **coyoteboy** » Fri Jun 16, 2006 12:34 am

Yes mops, thats exactly what im finding. Hard on the throttle my AFRs go from 11:1 to 13:1 (scary at 15psi) in any one bin. I realise theres a lag in the wideband readings but this clearly falls outside that issue when i scan my latest logs.

But yes, even when de-activating AE and decel enrich I get changing AFRs like you do, only it seems slightly more so. Maybe if i do mine under light accel I might find lesser difference. I might post to the LC-1 group and ask there.

J

mops · Post by **mops** » Fri Jun 16, 2006 2:38 am

i dont think wideband lag is an issue. it responds very fast.
program it to instant response and you wil see how afr needle jerks.

with oscilloscope, there is a way to even see afr per each cylinder with only 1 sensor ! - that really illustrated how fast wideband response is.

bottom line is that i think wideband had pretyt much no lag (at least for car tuning purposes).

I hope it will be possiblt to tune it out with x-tau.
because right now in stable conditions my afr's are stable (+-0.2), but in situations like gear change (release throttle then apply throttle) afr's are all over the show pretty much....

coyoteboy · Post by **coyoteboy** » Fri Jun 16, 2006 3:13 am

It may respond very rapidly but it wont be correct at 5ms. It takes time for it to respond to step changes in AFR and although it may jump 10:1 to 11:1 it could well be still on its way to 12:1 in actual fact. Its not a delay where it holds one value then waits a while and then reports the next, its a delay where it shows some movement towards the correct level but takes time to settle. Time we dont really have between engine cycles.

Either way it should, according to the specs on the LC-1 forum, be fast enough to give a good indication.

Jedrik · Post by **Jedrik** » Fri Jun 16, 2006 9:40 am

I dont use AFR readings to tune VE unless my TP is not changing during the test interval. Otherwise you should be tuning AE/DE not VE.

For me, acceleration EGO spikes are resolved within ~1/2 second.
I can not imagine this short period of reduced cooling being damaging as long as it only happens intermitently.