So I'm helping 3dman1 tune his ERL 2.3L 4B11T stroker. We had the car pretty nicely tuned (450 whp) on the stock bottom end and were seeing only occasional minor knock after crossing past 7300 rpm. He decided to be proactive and had the motor rebuilt with forged internals (ERL 2.3L stroker). He got the car back together just before xmas, and driving on the 2.0L tune, he was observing massive knock (maxing out the knock sum to 36) under certain conditions on 92 octane. Fast forward to this evening when I finally had time to start looking at the tune. After dialing in a new set of injectors (FIC1680s) on E85 (E85 WOT AFR dialed to ~11.5:1 to start), we set off to try to decipher the knock issue.

Sure enough, when doing some freeway pulls, we would see the knock sum shoot through the roof (36 knock sum) very rapidly. However, we quickly realized that it wasn't always happening. After a bit of testing, we found two things that seem to set off the knock sum:

1) Going WOT below 3000 rpm. This make it much more likely for the knock sum to shoot up, usually as soon as the throttle was pressed.

2) A rapid increase in accelerator pedal position would seem to set off the knock sum.

Or to put it another way, starting a pull after ~3300 rpm and gently rolling into WOT would yield a knock free run nearly all instances. Clearly, with E85 the knock we are seeing isn't real, so the question is whether its engine noise or something else. I'm leaning somewhat towards the idea that its a combination of a minor increase in engine noise and some logic in the knock sum calculation that is sensitive to change in throttle position.

I'm interested in hearing what amount of knock other people are seeing with built motors.

 

MAP Stage 1.5 2.0L w/ their ultimate duty rods and .020 over JE pistons, I have not touched the knock tables and no major knock like that in any situation, ever.

One thing that Shane@DB Performance did while tuning my car was hook up an extra knock sensor to the passenger motor mount and then ran a wire into the cabin, hooked to some fancy headphones he wore during most pulls. I don't think we ever saw a knock count greater than 3. I can get more info on this setup if you want, might be very useful in this case.

 

We have built about 6 4B11T's at our shop in the past 8 weeks..however all engines were built with stock displacement of 2.0L and we have not seen any unusual knock...I have not had to adjust the stock knock tables for any of the cars except for one..where knock shot up to 36 counts and that as well happened occasionaly...I played with the knock sensitivity tables to yield no positive results were the knock occasionally appeared and disappeared...bottom line, it was a faulty knock sensor..which was replaced with a new one and the problem was resolved...the knock sensor plug was crushed somehow and caused inconsistent signals to the ecu causing the phantom knock to appear.

I apologize in advance for any spelling mistakes..typing on the forum from the iPhone in bed isn't fun

Keep us posted mrfred.

Regards
Moosa

 

yes that is a great way to determine if there is real detonation/knock occuring...mostly refered to as the "old-school" method...

The problem is not only to determine whether or not it is real knock...an experience tuner can always find a way to determine that however it is having to trick the ecu to prevent it from pulling timing when it sees this phantom knock is what proves challengiing and in most cases prevents you from making power.

Regards
Moosa

 

Shane@DB plans to take his headphones on the road with me to check the spool up knock.

I know he had my top end tables modified for my FPREd setup, but for my black I am not sure what he has adjusted yet. I don't think we have really ever seen "real" knock on my motor. It's been straight E85 the whole time, I am guessing engine noise.

 

update: 3dman1 had the knock sensor overtorqued, and after retorquing to something closer to factory spec (no good torque wrench handy until tomorrow), we are seeing more normal knock response. still getting some big spikes though, so we're going to try a new knock sensor.

 

That is great! Hope that works fir yiu guys..

Would you be able to share the stock torquing spec for the sensor? Would be interesting to know.

 

15 +- 1 ft lb

 

good luck.

MrFred its the same knock algorithm as CT9A (basically) so you can log the usual suspects and fiddle with the multiplier to decrease sensitivity in required areas...

 

The knock sensor was swapped out yesterday with a known working sensor and the same problem was still there.

 

Yes, but its still much better with the sensor properly torqued, correct?

 

Are sure you have no metal contact anywhere on the chassis like an intercooler pipe or exhaust hangar? I found a hanger contacting all the way in the back near the muffler be the culprit on one car that had random high knock count spikes. In another there was a lower intercooler pipe touching on the core support area and not bolted.

I am sure you probably already gave it a once over, but with random spiking like that, I would go over everything very closely to rule out all possible external sources. Otherwise this could dog you for a while and be a major head ache.

 

It appears that we've got the knock whipped now. Looks like it was two things:

1) Overtorqued knock sensor causing knock everywhere in the rpm range.

2) Compressor surge lighting off the knock sensor at 3200 rpm or below. We killed the surge by zeroing out the exhaust cam retard. We'll slowly add a bit back after we get a bit further down the tuning road.

We also tweaked the knock multiplier a bit (changed from 16 to 20) at 2500-3000 rpm, but we feel pretty confident that the majority of the issue below 3000 rpm was compressor surge. We'll try putting the values back to stock after we get the tune cleaned up a little more.

 

Another question on over torquing the sensor:
If it has been over-torqued, is it damaged?
In other words, is just re-setting it correctly sufficiant to rectify the issue?

 

The sensor looks ok. 3dman1 tried droppinbottom's sensor with the same result.