tkklemann

iTrader: (6)

Hey all:

What is the correlation of knock voltage to knock count (sum)? When I look at the two data logs from a pull, it does not appear as though the voltage can be used to determine the knock sum because there seems to be no correlation between the two. It seems as though the ECU is using something else to calculate or spit out a knock sum count.

Basically, I was told by a local tuner here, in a shooting the breeze conversation, that he is tuning (The stock ECU) by using knock voltage from the sensor to tune, and that the knock sum isn't really a good indication of what the ECU/motor is really doing. Now, I don't really agree with this because the ECU react's when it sees a knock sum, not a spike in knock voltage, at least as far as I can tell from looking at my logs.

So, in looking at my data logs, during one tuning session on my car, I had a spike of 8 counts, and it reduced my timing and octane flag, but the voltage of the knock sensor during that knock ccount logged never rose above 1.1 volts. About 30 seconds later, you can see I did a 5th gear merge into traffic and see knock voltage's of 2.0v up to 2.1 volts (Not sustained, just spikes) but no registered knock counts (sums).

So, with all that being said, is there any real correlation here? And, how is it wise, knowing that the ECU is seemingly not reacting to voltage, to tune a car based only on the voltage and not the knock sum output?

On another note, I have the following mods, and was wondering what kind of boost levels using 93 should I be shooting for? I have a Defi, and I am getting readings on it around 26psi (But I know they can be off). Also, what kind of timing values should I have at those levels of boost? some more info:

Hitting loads of 240-260
RPM, Timing, Then AFR
3K 2* 11.2
3.5K 1* 11.2
4K 3* 11.2
4.5K 4* 11.1
5K 5* 11.1
5.5K 6* 11.0
6K 8* 11.0
6.5K 11* 11.0
7K 14* 10.9
7.5K 16* 10.9

I am currently working on the AFR's in the higher RPM's at the moment too...

Mods:
Injen Intake and Hard Upper I/C Piping
AMS Lower I/C (Turbo Outlet) Pipe
Injen Intercooler
Hallman MBC
Ported Turbo
Manifold, High Flow O2
3" All the way back, no cat
PTE680's, Perrin Fuel Rail, FPR
Mini Battery
1 Shocker Sticker on Trunk
Ground Kit (Irrelevant)


With all that, what should I be shooting for boost wise? I am going for 26 with spikes to 27..

(Pretty soon it's going to be all irrelevant because my cams, valvetrain, belts etc, etc. are going in soon.)

l2r99gst

iTrader: (2)

I would never use knock voltage to tune by. That's the biggest downfall of every piggyback and standalone that doesn't have a true knock algorithm like the stock ECU.

Knock voltage is only one variable that goes into calculating the knock sum and determining whether or not there is real knock. Bez explained some of the knock routine over on aktivematrix if you want to read the thread, but tuning by knock voltage is just foolish. I've been saying that for years and now that we can log knock sum directly from the stock ECU, in my opinion there is no better way.

Mitsubishi may or may not have it spot on, and the routine may need some adjusting with different mods, but they developed algorithms, maps, filters, etc, for our exact engine. They created the frequency maps and everything that goes into determining a knock sum. Knock sensor voltage tells you little alone. The RPM, crank angle, frequency, voltage, change in voltage, etc all come into play. It's not just simple knock sensor voltage.

As you mentioned, you didn't even see much of a correlation. If I were you, for now, I would stick with tuning by the knock sum. That's the best that we have at this point and in my experience over the years with DSMs and now Evos, it's pretty good.


Eric

tkklemann

iTrader: (6)

That is exactly how I have been tuning (By knock sum, not voltage). I was taken aback when a local tuner told me that myself tuning by knock sum was not a good method, and the better way was to go by the voltage.

With that said, the tuner confused me more because of my prior knowledge of there being no correlation of the two.

Thanks for helping, that's exactly the answer I was hoping/looking for.

Any comments on the second part?

Jeff_Jeske

iTrader: (66)

The engine has a noise floor that gets higher as the RPM and load increase. There will be more noise at 7500 than at 3000 rpm and the knock sensor will pick up this noise in the form of voltage. The ECU knows what is considered acceptable and what isn't. The ECU runs the out of spec noise through an algorithm of some kind and produces a count of knock.

Voltage is only one part of the formula.

EricJ@AMS

iTrader: (1)

Now if someone could only find a way to adapt the knock sensor and formula from an Evo or 2g and program them into a 1g! The 1g knock system is total garbage.

iTune

iTrader: (3)

Most OEM ECU's(especially Honda's) have whats called a knock level(voltage) and whats called a knock threshold. When the knock voltage goes over the knock threshold you get a knock sum(or whats called a knock count). But, here is the kicker.....the threshold changes according to several factors, such as RPM, TPS and engine load, among others. The ECU determins what the threshold should be according to its knock threshold protocols. The hard part is figuring out what the actual threshold is(logged) and where the threshold is changing per the ECU and why/when. After you establish the threshold during the engine's various states, you can see how voltage crosses over the threshold, resulting in a knock...

If in fact this ECU does have a knock threshold, it would be beneficial to log it, along with knock sensor voltage. This would give you a better picture of the knock condition.

CJ

tkklemann

iTrader: (6)

Any comments on the second part of my question?

Mr. Evo IX

iTrader: (5)

26 to 27 psi is too much for 93 octane. Most people tune 21-23psi on 93 octane.