Notices
944 Turbo and Turbo-S Forum 1982-1991
Sponsored by:
Sponsored by: Clore Automotive

Datalogging the Knock sensor?

Thread Tools
 
Search this Thread
 
Old 02-10-2005, 05:47 AM
  #16  
NZ951
Race Director
 
NZ951's Avatar
 
Join Date: Apr 2003
Location: New Zealand massive
Posts: 13,778
Received 5 Likes on 3 Posts
Default

Yup, the Link records the knock values and can be plotted in a curve...
Old 02-10-2005, 06:12 AM
  #17  
hosrom_951
UAE Rennlist Ambassador
Rennlist Member
 
hosrom_951's Avatar
 
Join Date: Jan 2004
Location: UAE & Germany
Posts: 9,142
Likes: 0
Received 3 Likes on 3 Posts
Default

When you say :

Originally Posted by NZ951
the Link records the knock values and can be plotted in a curve...
Which product of the link are you referring to exactly? (the stand alone?)
Old 02-10-2005, 02:33 PM
  #18  
NZ951
Race Director
 
NZ951's Avatar
 
Join Date: Apr 2003
Location: New Zealand massive
Posts: 13,778
Received 5 Likes on 3 Posts
Default

Originally Posted by hosrom_951
When you say :



Which product of the link are you referring to exactly? (the stand alone?)
Yup, which Joe has...
Old 02-10-2005, 04:13 PM
  #19  
Claus Groth
Pro
 
Claus Groth's Avatar
 
Join Date: Jun 2001
Location: Thousand Oaks, CA
Posts: 564
Likes: 0
Received 0 Likes on 0 Posts
Default

Originally Posted by BoostGuy951
I'm sorry I didnt really follow this. Are you getting .2 volts when you knock on the block and then 0.0 volts otherwise?

I've got a knock display, but I also wanted to be able to datalog.
I got 0.2 volts while knocking and zero volts otherwise.
Old 02-10-2005, 04:15 PM
  #20  
Claus Groth
Pro
 
Claus Groth's Avatar
 
Join Date: Jun 2001
Location: Thousand Oaks, CA
Posts: 564
Likes: 0
Received 0 Likes on 0 Posts
Default

Originally Posted by hosrom_951
Claus: Did you have any logs on the knock sensor from Zeitronix but on various boost levels? maybe that would show some difference.
I didn't save the logs so nothing to show. I only tried to test in the garage by hammering on a steel bar adjacent to the knock sensor.
Old 02-10-2005, 04:21 PM
  #21  
Claus Groth
Pro
 
Claus Groth's Avatar
 
Join Date: Jun 2001
Location: Thousand Oaks, CA
Posts: 564
Likes: 0
Received 0 Likes on 0 Posts
Default

Originally Posted by Dark Lightning
Does it seem accurate and useful?
The Knocksense uses the stock knock sensor and a cable is provided to connect to the sensor and the factory connector. There is an adjustment pot on the circuit board. The object is to set it when you know that there is no knock. I did that by using extra unleaded race fuel and running on the track.

I have it set so that there is a flash of light once or twice per lap. If that changes, then I know that there is something wrong. It is basically a reference. I don't know what severity of knock the DME uses to retard timing and reduce boost .
Old 02-10-2005, 04:22 PM
  #22  
Claus Groth
Pro
 
Claus Groth's Avatar
 
Join Date: Jun 2001
Location: Thousand Oaks, CA
Posts: 564
Likes: 0
Received 0 Likes on 0 Posts
Default

Originally Posted by BoostGuy951
Yeah. I intend to give it a shot. I will report back as soon as I have a datalog with it as a variable. (which should be a while)
If you mean the Knocksense, be sure to request the Porsche wiring harness. The Volvo sensor has pins reversed.
Old 02-11-2005, 06:33 AM
  #23  
Danno
Race Director
 
Danno's Avatar
 
Join Date: Jul 2001
Location: Santa Barbara, CA
Posts: 14,075
Likes: 0
Received 3 Likes on 3 Posts
Default

The knock-sensor is basically just a microphone. It's an analog device and puts out raw waveforms that must be massaged, filtered and interpreted. I've datalogged its output with a 'scope and it's not easy to detect knock visually from the screen. Stuff that you can hear from the Knock-Link doesn't show up distinctly in the traces. And I can't imagine what inaudible knock would look like.

If you take RMS average values of the voltage over time, it will increase linearly with RPM due to the louder engine noises (valves slamming shut, cam-lobes hitting lifters, etc.). Problem here is interpreting the knock waveform within all this noise. Knock's waveform and amplitude remains the same regardless of RPM and boost. So let's say it's a 0.2v spike rising from 0v in less than 2ms and rapidly diminishing in less than 1ms. If you set a voltage-level trigger at 0.2v, it will be "on" all the time after 4500-5000rpm because of all the other noise.

Also the main combustion waveform may take 20-30ms to rise from 0 and may reach only 0.05v at idle and small throttle-openings. But at 5000rpm under full-throttle/full-boost, the combustion waveforms will be 0.3v or higher with wavelets at 1/2 frequency for the valve-train. The problem here is that a voltage-level trigger will not catch a 0-0.2v spike mixed in with noise of 0-0.3v. You need an edge-trigger with a certain rise-rate that it's looking for. This appears to be a prime candidate for DSP technology.

Here's an application I've been examining for suitability for our cars: Intersil (HIP9010/11EVAL).pdf
Old 02-11-2005, 12:58 PM
  #24  
BoostGuy951
Three Wheelin'
Thread Starter
 
BoostGuy951's Avatar
 
Join Date: Oct 2002
Location: Gulf Shores, Alabama
Posts: 1,841
Likes: 0
Received 0 Likes on 0 Posts
Default

I have a Link Knock Display. It has several little LEDs on it and one Large one that lights up to indicate knock. I'm wondering if I can solder my datalogging input wire into the circuit of the LED. I'm pretty sure the purpose of the circuit board inside the knock display is to massage and filter the signal. So I'll basically be datalogging when the Knock LED lights up.
Old 02-11-2005, 06:25 PM
  #25  
Claus Groth
Pro
 
Claus Groth's Avatar
 
Join Date: Jun 2001
Location: Thousand Oaks, CA
Posts: 564
Likes: 0
Received 0 Likes on 0 Posts
Default

Originally Posted by BoostGuy951
I have a Link Knock Display. It has several little LEDs on it and one Large one that lights up to indicate knock. I'm wondering if I can solder my datalogging input wire into the circuit of the LED. I'm pretty sure the purpose of the circuit board inside the knock display is to massage and filter the signal. So I'll basically be datalogging when the Knock LED lights up.
Check with the manufacturer on that one. I asked Boris at Knocksense using the signal to his LED and he said that I had to use an OP-AMP. Not being electronic, that went over my head.
Old 02-11-2005, 06:42 PM
  #26  
fast951
Addict
Rennlist Member


Rennlist
Site Sponsor

 
fast951's Avatar
 
Join Date: Mar 2002
Location: Atlanta
Posts: 6,885
Likes: 0
Received 37 Likes on 27 Posts
Default

I have used the Knocksense from Boris. It was on a modifed high CR 944NA.. We used a 951 knock sensor on the block connected to knocksense.. The device indicated some knock at high RPM, I'm not sure if it was engine noise or actual knock. We didn't hear actual knock...

The KLR has some interesting knock filtering algorithms.. Someone (guess who) has been asking me to show him how to log the signal after the KLR did the filtering (when KLR determines there is knock not noise). We figured a way how to log knock at the same time (RPM, AFR,..).
I think it'll be a good idea to compare the KLR knock detection with the Knocksense.. So SpecialTool, you'll have few toys to play with ..

Note: For those getting knocksense, make sure you tell Borris that it's for a Porsche as the wiring is different than his standard kit.
__________________
John
Email
www.vitesseracing.com
Old 02-11-2005, 06:47 PM
  #27  
special tool
Banned
 
special tool's Avatar
 
Join Date: Apr 2002
Location: limbo....
Posts: 8,599
Likes: 0
Received 1 Like on 1 Post
Default

John - 5 degrees advance by load okay to start????

Muuuuuhahahahahahahahahaha



Quick Reply: Datalogging the Knock sensor?



All times are GMT -3. The time now is 07:02 PM.