Harlan

iTrader: (3)

I did not track load, but as you can hear from the unfiltered it was cruising at fairly steady load. I could have logged it and been able to correspond the time in the log with the time in the audio file, but that wasn't the point.

Team, if you know more about the tables in ATR I'm all ears. It looks to me as if the ecu has a -10deg modifier (max retard) and then uses the knock sensor to dial it back up to table values (plus modifiers). I'd really like to know what modes of control the ecu is in where, and what these tables are actually controlling.

logalinipoo

I thought you were trying to see if our cars are using this sensor to control the timing. I am a believer that it works, but I a curious if the ECU is actually hearing it and making a correction.

I have not had an AP for very long, and I'm still breaking in my engine so I have not had it over 5K RPM. I have logged Knock a lot and never seen it in any of my logs.

Even if you can hear it through your sensor, it doesn't do any good if the ECU does not use the data for some reason.

It might be possible to transmit a range of frequencies into the ECU to see which ones produce Knock in the logs. And cross reference that to the Frequencies you are hearing to see if the two corresponded to each other.

Harlan

iTrader: (3)

I originally wanted to know for sure what frequency knock was at so I could build some filters/analysis hardware. Didn't plan on using stock knock control because the general consensus was it didn't work. Now I believe it does work.

I did feed the ecu some tones from a tone generator. I was pulsing them, but the response though noticeable was somewhat erratic. It did show a large range of frequency response, it just didn't seem to completely believe the signal I was giving it. The best way to get a response was to quickly sweep a narrow frequency range while in pulse mode. I may set up a sound card to do further testing with a more realistic knock sound now that I have samples. It also retarded timing very noticeably when I slammed the hood. I do not believe at this time the ECU filters any frequencies.

logalinipoo

I just found your turbo build thread and found a little more of your work. It is all very interesting. I appreciate the work.

I have wondered if I could wire a second knock sensor parallel to the first and connect it to the front rotor. IT might increase the signal sent and kind of amplify any sounds they do hear, but if there is not enough power to support the second sensor it could make the signal too weak for the ECU.

Any thoughts?

Harlan

iTrader: (3)

The sensor is a piezoelectric, it doesn't take power to run it generates a voltage just like a microphone. Two of them in parallel should work, but you would need to add a resistor to prevent getting a CEL, and would probably need to adjust level out (or get control of the ECU's sensitivity table). It's a work in progress, and I need a stable version to continue my build. More to come.

olddragger

iTrader: (3)

I--too--wish it was that simple loga....

logalinipoo

It's actually Logan, If I find a knock sensor around I'll give it a try. Lets see what ebay has to offer.

olddragger

iTrader: (3)

Sorry Logan--i am lazy and try to save keystokes at times.

Harlan

iTrader: (3)

Here's what I've done with it:

I did the video tonight when I first got it working after some tweaks it's now working much better. Knock recognition is amazing, much better than I can do by ear with the unfiltered. The actual numbers it's reading out tell the story the best.

With current settings it baselines at about 300-350 and is above 1500 with a "loud knock" these numbers can be changed to anything, the important part is that when it knocks it comes well off the baseline beyond a simple margin of error. The code needs a lot of work to streamline it, but it's working great in it's current capacity.

My current plan is to make it into a dash knock light and a knock based ignition cut if it exceeds a preset value. Knock based retard is also an option, but it's trickier to code safely and I'm lazy for now.

Tomorrow I'll post an updated video, next time it will have engine sounds without me talking.

FazdaRX_8

iTrader: (1)

I have seen the knock sensor work and seen it pull timing all according to the maps in cobb, I know the sensor works and the system uses it. I have even seen it add timing.
past 7000 rpms though it just stops retarding ignition. and returns to whatever the timing map is. most of the time I see it is 5000-6500 rpms right around peak torque.

I hope you don't plan on making a tune on the edge knocking....

olddragger

iTrader: (3)

THIS IS THE MOST SINGLE IMPORTANT THING THAT HAS HAPPENED LATELY IN FURTHER DEVELOPING THIS ENGINE!!!!!!!

I bow to you Harlan--you are the man.
Now I do hope others recognize what this means.

Harlan

iTrader: (3)

Thanks for the info! Did you see the knock control working by comparing the timing accessport reported to the timing on the map? Or comparing different pids? Or the accessport knock retard parameter?

Over 7000 and below .5 load I'm guessing it goes into knock feedback control instead of the more advanced routines. I have no idea what the accessport logs as knock retard, I've never seen it log any, but I've only looked at it a couple times.

I am planning on doing stupid things, but working on this as a safeguard. It's going to work with stock knock control instead of replacing it. I've been over the edge with NA, in retrospect stock knock control is the only thing that has saved me. A staggering number of engines may have blown because people unintentionally overrode knock control.

Harlan

iTrader: (3)

Thanks OD, but please let me get it working on a car before you make any judgments! Right now it works well with recordings, when it works well on an engine then it will be time to pop the cork.

slash128

2600 you can hack phones with the RX8 knock sensor

Brettus

iTrader: (2)

This might be of interest ......

https://www.rx8club.com/engine-tunin...6/#post4455682

RX what?

N00b coming through, in lamen's terms can someone explain what the intended function of a knock sensor is. Does timing change when knock is sensed? Is it retarded or advanced? OD said he hopes others recognize what this means. I would like to recognize what this means.

olddragger

iTrader: (3)

I will try to explain . This engine when forced inducted is very sensitive to detonation. Very sensitive.
An engine "knocks" many times and the owner is not even aware of it. Some knock is not unusual nor of concern. Knock is not detonation---but if you are having many knocks--then that means you are much more likely to have detonation, a mad wife and an empty wallet.
An accurate knock counter will go a long way in identifying the parameters in which a more accurate tune can be done. It will also help identify when a tune may not be able to compensate for a bad tank of gas, overheating etc.
It is sort of like a baby monitor

logalinipoo

ok, forgive me for thread jacking.

Is the numbers for timing on stock maps(cobb AP) The advance above TDC?

Harlan

iTrader: (3)

Yes, unless it is negative then it is after TDC.

FazdaRX_8

iTrader: (1)

Yes comparing stock map to what was reported. There are sharp changes. The knock will never report on cobb. Even though it happens. Above 7krpms there is no appearant ignition retard. Infact the if your timing is 9 And it was pulled to say 4 once you hit 7k it will return to 9.

Brettus

iTrader: (2)

5k is limit for retardation on stock maps in my experience .

RX what?

Thank you

RX what?

One more thing TDC =top dead center? Would this be at the peak of the compression phase in a rotary?

Harlan

iTrader: (3)

Yup, top dead center happens on a rotary the same as on a piston the geometry is just a little different.

Prototype is on my desk, need to connect it to my car and set up the variables. Works perfect with engine sound recordings though. I'd be doing that now, but there is a cat on my lap and I can't get up.

FazdaRX_8

iTrader: (1)

7k for me jdm vs usd?