i have no logging abilities but i found my front corss bar hitting the downpipe. i fixed it and it runs much better.

 

Again, we are not talking about rattle issue here. Let's get more focused on knock sum decay. Even if we assume these are just rattle noise and phantom, It is only happening on 3 conditions ( fast rev, fast hard shifting and bouncing off rev limiter (2 step included))

And, stock ECU Knock decay function is giving problems!

 

Here are a couple of snap shots using DataLog Lab & EVOScan software.








Above is a 3-4 flatshift from a 1/4mile pass. You can see the knock spike to 8 counts, at the bottom center of the screen, as soon as the car starts to free rev. The timing dropped to 5* at the lowest point in the curve (white line) on that pass. compared to the other lift-shifts it looks like it hurt performance...









Above is the snap shot later on that night, after I was able to tune most of it out.

What's interesting is the fact that now timing went down to only 8* during the free rev to the next gear. total knock counts on that spike was only 2.

 

See I havent found any knock problems on pump gas shifting fast. My somewhat oft referred to 5th gear highway pull only got knock in the top of 5th at 6900-7050 only (well higher than 3 counts) and dropped back to 4 at 7100. The 2 step knock was all in the range where I had specifically reduced timing to a negative number. I should go back and log it with the timing in the map at the unchanged level and see what it did. For the most part the knock was 5 or less except for where it hit my planned timing retard.

While I understand your concern with knock sum decay, I think you might be facing other issues that not everyone else is. What do your fast shift UPSHIFT logs look like (that dont have any downshift)?

 

up shift is not bad like down shift I do get 0 knock on smooth upshift.

But, It happens when the load you are putting in next gear is significantly different. I should have phraised as " rough shifting" So, launching a car is the form of rough shift. and down shifting from 4th 2nd gear will be the form of rough shift. I believe I can reproduce this knock count on any evo or any knock sensor equipped street cars. I thought this was well known fact and I read few articles about this effect. Am I horribly misguided? .

And, any knock count above 3 will affect your timing and fuel for that gear you are in. Only way to get the promised dynoed horse power is to have 0 knock sum in the ECU register.

 

Good posting!
I am glad to see other memebers logged this shifting knock phenomenon.
I think I can reduce my knock if I do VTA with my bov. fuel will be pig rich after I let off the gas. hence, it will serve as fuel enrichment .
But, I won't able to stop free rev and 2 step knock.

 

What's even more interesting AlwaysinBoost is that the first log with lower timing has a much smoother timing advancement curve with no timing pulls then the second log with higher timing does.

Hmmmm...

 

You can defintatley get knock during quick engine transitions. A LOT of the times it is due to a few simple things.

1. The engine is moving around and hitting things. Make sure things are secure and that you have 1" clearance around your exhaust.

2. The engine is too hot. You are on the edge of your tune and at this point quick engine transistions (for whatever reason) is causing knock.

3. Hot exhaust valves (or similliar). Your running too much advance or too lean. EGT guages do NOT read in cylinder or exiting cylinder temps. They measure manifold temps where a low timing vehicle burns fuel at (heat).

Most of the time it is a tuning issue. Then some cars are posessed. Also flat shifting without a NLTS is generally bad for everything. If your doing this and a transmission failure ensues, be a man an pay for it yourself. It is not warranty claim time.

 

Are you rev matching on the down shift?

 

I don't really flat shift my car, but my shifting is very fast and smooth so I don't experience the knock you guys talk about..

I do get a bit of knock when downshifting or launching, but again, it either returns to normal within a few hundred RPM or it doesn't affect timing in a way that I could tell it was doing it.

 

NO, rev matching smooth down shift causes 0 knock on my car.

 

But, this was on race gas and I do have 1" clearance. I did quite a work to have that clearance and eliminate the rattle. I have even removed heat shield that seemed to cause bit of rattle.

So, free revving knocks are the tunning issue also?

 

well maybe I am lucky? I get between shift knock but really low counts (1-2). Are you sure about 3 counts causing timing pull? I havent been able to trace that on any of my logs, but then again maybe some of this is a difference between the VIII and IX?

 

hmm... may be 9 has less sensitive knock sensor filter map and fast decay?

3 counts are the sure way to pull timing on evo 8. if I get 3 knocks at 3000 rpm it's effect remains till 5500 rpm. 5 counts effect all the way to to 7000 rpm. And even just 2 counts ruins my peak torque.

 

Maybe I'm fortunate, but I Definitely don't get these types of knock events you guys are talking about, whenever I do, its only during part throttle and low load, and rarely affects the timing on the car. Plus the counts would have to be in excess of 3-5 for a longer period of time (on Evoscan more than one sample)

The only time I saw it happen was just after my turbo install when I would tip-in slightly lean, when I fixed that, the knock count went down to 0 or had rare 1 and 2 counts if I rolled onto the throttle..

Like I said, where I would experience the knock most frequently was in daily driving under low load where the car is generally running closed loop..

My car is a first allocation 03 car, its ECU has not been altered from when the car was delivered to me and the car doesn't seem overly sensitive to knock.. It would hit fuel cut pretty fast compared to other cars, and its boost was slightly higher than other cars I had worked on when it was stock. The one thing I can say is, evo scan (or any other useful method of detecting knock or other log data) didn't exist at the time and therefore its difficult to know what this car did stock.. But I can probably guess that it either had less sensitive, or more sensitive filters and thresholds than other later cars, because there have been almost a dozen revisions to the ECU since the USDM model was introduced.