From the recent threads posted with logworks data logs used for tuning, I have noticed that the knock voltage plot trace differ significantly from what I see on the xede knock buffer signal plots. Even the voltage values do not correlate with the signal values I see with the Xede, values being distinctively higher. I was under the impression that the y axis scale was 0 = 0v, 100 =5v. But from reading more into the subject I am now confused at what is the correlation between the knock buffer signal and knock voltage.

I have 2 questions 1) what voltage values does the Knock buffer values correlate to if any? 2) How safe is using knock voltage values with out amplification and low sample rates 12.5/sec of the LM-1/LMA-3 for tuning?


From the Innovatemotorsport forum:



#1 10-07-2004, 12:31 AM

BJai02



knock counts



--------------------------------------------------------------------------------



Is there a way to log the knock activities with LM-1 by reading the knock sensor voltage? I am not sure how the knock sensor voltage work, anyone? My car is an Evolution 8, and it has a factory knock sensor.

__________________

Autronic EVO8







#2 10-07-2004, 06:09 AM

Stephen Max



The short answer is no. The knock sensor produces a very high frequency voltage when it senses knock, and the sampling rate of the LM-1 is not fast enough to pick it up with any certainty.





#3 10-07-2004, 07:17 AM

klatinn

Administrator Join Date: Dec 2003

Posts: 3,276



Hi,



See this thread for details why the LM-1 does not interface with knock sensors: http://www.innovatemotorsports.com/forums/showthread.php?s=&threadid=474 which is basically the answer bellow.



Innovatemotorsport Forum Post by Klattin

Hi,

Knock sensors are nothing but heat resistant microphones. The signal from a knock sensor is an audio signal recording engine noise. The knock detect circuit in an ECU first of all looks at that audio signal ONLY during about 5 deg ATDC to about 70 deg ATDC, where knock is most likely. The characteristic audio signal for knock has a specific frequency (pitch) that the knock detect circuitry filters out. This pitch (some frequency between 5 kHz and 20 kHz) is dependent on bore size, average gas temp in the chamber and comb. chamber shape. The knock detect circuit is factory tuned for that. On many OEM ECU's the knock circuit is switched off above ~4000 RPM, because the knock signature signal gets buried in other engine noise.
Engine mods like boring out, turbo's added, different boost, different fuel and so on, can change the characteristic knock frequency with the result that the ECU will listen to the wrong frequency and will not detect knock at all with it's existing sensor. I can also pick up other engine noise as knock.
There are products out there that attempt to detect knock from existing knock sensors. They typically require you to program a threshhold amplitude level for the knock signal by RPM, without regard to the characteristic knock frequency of the engine. That works only very marginally, because if the knock sound level is above the engine noise level at higher RPMs, you can hear it already with your ears before that circuit can detect it and you might already have engine damage.

Regards,
Klaus



More on the topic

http://www.innovatemotorsports.com/forums/search.php?searchid=89943

 

I played around with the innovate logging software yesterday and monitored the knock signal. Yes, it's pretty useless due to sampling rate limitations. It's basically a crap shoot as to whether it will actually be able to record a knock event. What is the max sampling rate of that input channel, btw?

 

The individual channel rate may be real time but AFAIK the Innovate products/software spits out engine data at 12.2 samples per second.

http://www.innovatemotorsports.com/s...s_Manual_2.pdf page 7-11 .

 

Here is some more reading on the subject...http://home.netcom.com/~bsundahl/kno...unds.htm#Knock
the more I read into this the more I'm conviced that reading knock voltage (LM-1)without the proper sampling rate and/or intephase device, i.e. knock buffer, is like pissing in the rain.
http://home.netcom.com/~bsundahl/kno...rtb11_1_10.wav

 

Yep.. absolutely useless as far as knock detection goes.

Shiv

 

Hey SHiv,

How are people posting up knock with XEDE SMART? Is it with donour's program. I have seen graphs before, but I haven't yet explored how to do it. Apparently, knock info from the LM-1 is useless.

 

pyxede can be used, and will do all data conversions. I use commslog and convert the raw csv data by using Excel and formula AN0In *100/255. Then I graph out the results

 

Since most of that is gibberish to me, at some point I think I'll need some help from someone local.

Well, you gotta learn sometime, right?

 

With an excel macro it would be easy to do, but I don't know how or have the patience and/or time to learn right now.

Any one here good with excel macros?

 

What JorgeT was trying to say is yes. You can use my program (pyxede) to log knock and AFR. You can also use Xmap, but that requires you to do some arithmetic on the result in order to get the correct units.

d

 

thanks for clearing that up

 

Sure thing.

Excel macros for what?

d

 

A macro to do the arithmetic and graph of a csv datalog file made with Commslog6, similar to the one I posted above. I like to look at data using the same scale and values as used on the xede smart tables.