GTR Forum banner

1 - 11 of 11 Posts

·
Registered
Joined
·
2,043 Posts
Discussion Starter · #1 ·
Theres a few rumors flying around at the moment (I am aware of the Axe-grinding source of said rumors) as to the knock detection methods I've been using in my ten years of mapping cars - so to clarify, in no order - heres the list of things I use, and "might" use in future...

Current detection methods...

1. Experience, having provided mapping to over 1000 turbo cars over the years (600+ SAABs, many hundreds of Skylines (300+ at a guess), and various Supras, Subarus, Mitsubishis and some odd ball stuff), ignition figures are usually pretty clearly wrong, when they are wrong. Experience is a great teacher, and a great adviser. If the numbers in the map seem wrong, something is at work that requires investigation (in the case of the GTR its usually people too lazy to align the static ignition (controlled by the CAS) with the ECU's idea of what the ignition actually is, rendering all of the mapped ignition numbers meaningless).

2. Old-school det cans, copper tube + a megaphone or headset. Its tried and tested to be effective for 99% of applications, traditional resonance det cans are a great way to hear it all at source, but can give you ear-ache!

3. Electronic det cans, pretty much performs the same as above, though electronics provide the amplification, and often boost the frequencies you need to hear (between 4.5Khz and 7Khz as a general rule) and cancel those you dont, making audible knock recognition far easier.

4. Oscilloscope, provides a visual feed of "noise" from the knock sensor(s), this isn't frequency specific (very much like PowerFC for example) but is a very good guide, if you see it going crazy, something is making some noise. Shielded cable for the knock sensors is essential for this to be useful as the wiring can be very sensitive to electronic interference.

5. High sample rate Spectrum Analysis, this takes the oscilloscope one step further by showing a breakdown of the frequencies within the "noise", knowing the det frequency of the engine allows you to look for specific noise in that frequency band and very accurately find and eliminate knock. Hugh Keir did some work with this a while back and published his results, very interesting stuff and very accurate indeed.

6. The Exhaust-Ploom - this is the one that seems to have got people talking, so to give a little more info... Watching the tail pipe obviously isn't knock detection, what it is however is a (seemingly) fairly accurate way to see when knock is ABOUT to occur on a turbocharged car. I discovered it accidentally while I still worked at Abbey, we had installed an exhaust camera to watch for exhaust smoke, as being focused on my PC most of the time meant I didnt often look behind me - so we stuck a camera in to see any problems with turbos or heavy engine smoke etc, and placed a monitor in the front of the cell where I could see it easily. While mapping a PFC I noticed that just (split seconds!) before a noteable knock event on the PFC, "something" was clearly occuring at the tail pipe - a jet of deep black smoke. I reduced the timing, the knock event went away in the PFC, and so did the jet of black smoke. I tried to repeat the effect and sure enough, you could make it produce a fine jet of black smoke momentarily before the det event was recorded, and heard. From then on I used this as a "guide", along with the oscilloscope. It proved accurate and repeatable on 99% of cars.

Future knock detection...

Ion Sensing - this was (I believe) pioneered by SAAB for use in their "Trionic" engine management system, it has no knock sensors, but instead uses the Ion Current around the spark plug tip to detect knock on a per-cylinder basis, with spectacuar accuracy, it pre-detects knock and in most cases stops it, before it even starts. Ion Sensing control modules are now available as external boxes, so in theory, could be integrated into existing engine management systems that feature a +5V analogue input.

Today I will mostly be using...

Whatever knock detection systems I need, I do not depend on "just one", and will use all the methods available to me (more now thanks to my new employers) - my reputation lives and dies on my ability to do my job (like anyones), so it is of course in my interests to do it as best I am able, with all the tools available to me.

Anyone that would like to discuss my methodology or knock detection techniques can feel free to PM me, please do not make any trade or work enquries in this thread however, as neither I or my employers sponsor this forum, and as such it is unfair to those that do if we gain free advertising from it.

Best wishes for 2010 to you all btw :)

Dan
 

·
Registered
Joined
·
1,329 Posts
by far the most interesting thing ive ever read on this forum, thanks dan.
 

·
Registered
Joined
·
384 Posts
very interesting Dan. After recently having had a serious knock problem its good to understand how it is measured/detected. Thanks,

Ali
 

·
Registered
Joined
·
9,657 Posts
Dan! Long time no speak & glad to see your still around. Echo everyone's thoughts on the interest of you post & you know you have my vote for your work fella :thumbsup:
 

·
Registered
Joined
·
834 Posts
Interesting!

You are the second person I've heard use the black smoke as an idicator of det. I've seen it demonstrated on a classic mini before. Although without detcans to confirm it!
 

·
Registered
Joined
·
2,043 Posts
Discussion Starter · #7 ·
Thanks for the interest and positives chaps :)

Speed merchant: Hope all is well with you and yours sir!
 

·
Registered
Joined
·
2,674 Posts
Interestingly I heard Abbey use method 6 through Dave Walker at Emerald about 3 or 4 years ago. Have you got any videos you can post? I am interested to know if you see 6 on a fresh engine as I am guessing the smoke is carbon being dislodged from the combustion chamber as cylinder pressure increases?

I do knock calibration for OEM manufacturers and use method 5to capture the data to process (with 3 +4 as additional monitors) recording 4 knock sensor outputs and all cylinder pressures relative to TDC. We record at 0.001 CA. Recently we use scripts to then process the data.

My aim is different through I am deliberately trying to run as much det as possible while we record data to process to identify the knock filter & gain values at every speed and load. Its a fine line between recording sufficient knock events and melting or snapping the ring lands at high speed and load which is where 1 comes in.
 
1 - 11 of 11 Posts
Top