Civic858

Has anyone ever heard or seen this device. Looks good and only $129, but can it be used on a Honda/Acura.
http://www.turboxs.com/more_info.php?ID=212

EJ1 wilcox

Not bad for $129.

I'm interested in hearing if anyone is running one of these as well

awdgsx96

I've seen a couple people with the homemade "det-cans". Lets see if somebody can pull one of those threads up, I'm sure this will be twice as a good as a homemade device

Legion

Blah, they are trying to bank on consumer ignorance.

Even a simple factory knock sensor will detect knock at any RPM. The sensor is a simple transducer that converts noise to voltage; any abrupt change in noise (knock) will create a voltage spike, and the ECU will read the spike as knock. The ECU constantly monitors the knock sensor's output and knows what normal voltage outputs should be. It senses knock anytime there is an abnormal noise.

Their little graph showing the "competition's" knock sensor is bullshit - no one makes a sensor that shitty. Even their "less linear" curve is crap.

Bottom line, that thing is a piece of crap. Knock detection shouldn't be linear! No wonder it's only $100. They probably make them out of $5 of digikey parts.

Oyvind Ryeng

It's my understanding that the human ear (preferably connected to the engine block via a det-can), combined with the multi-teraflop parallell processor that is the brain is the better knock detector by a long shot. This is because the human psyche' knows from understanding what knock sounds like, knows what VTEC-solenoid clicking sounds like, knows what VTEC-lock pins getting pushed into place sounds like, knows what slight piston slap sounds like and knows what slight valvetrain noise sounds like, but the knock sensor, on the other hand, can only differentiate between "some kind of noise" and "there's not alot of noise right now", thereby, in effect, giving it an incredibly low "signal-to-noise-ratio" of sorts. This was semi-recently discussed for hours on end in the large DIY det-can thread.

mike93eh2

so basically its a worthless pos?

Legion

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by mike93eh2 &raquo;</TD></TR><TR><TD CLASS="quote">so basically its a worthless pos?</TD></TR></TABLE>

Well it sure sucks compared to OEM Honda/Acura knock sensors. But if you have a pre-OBDII car that didn't come with a knock sensor, this is better than nothing. Although it still won't throw a CEL or pull timing, all it will do is flash an LED light. Pretty corny.

nowtype

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by Semtec &raquo;</TD></TR><TR><TD CLASS="quote">

Well it sure sucks compared to OEM Honda/Acura knock sensors. But if you have a pre-OBDII car that didn't come with a knock sensor, this is better than nothing. Although it still won't throw a CEL or pull timing, all it will do is flash an LED light. Pretty corny.</TD></TR></TABLE>

How can you say this so definitively? I read up on this unit about 6mo-1 year ago and the reviews were pretty good overall. Someone tested it versus a det can and when he hear detonation, the light flashed. So again, why is this garbage?

Legion

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by nowtype &raquo;</TD></TR><TR><TD CLASS="quote">

How can you say this so definitively? I read up on this unit about 6mo-1 year ago and the reviews were pretty good overall. Someone tested it versus a det can and when he hear detonation, the light flashed. So again, why is this garbage?</TD></TR></TABLE>

Because they posted this:



Linear reading of a non-linear output? Doesn't make sense to me.

EE_Chris

The green and blue lines on the graph aren't showing actual KS output - that's teh red line. It's showing the threshold of when knock will be detected. It's showing how the 'competitors' KS box uses a single trip point to say "knock detected". It's showing that at 3k and 4krpm the 'competitors' KS box would have never picked up that knock because the threshold had to be set high enough that high rpm noise didn't falsely trigger it.

I'm not saying it's all hype, but it seemed your interpretation of the chart was a hair off.

mike93eh2

on a pre odb2 motor...where do you mount the knock sensor? should you tap one of those extra holes on the back of the block and just screw it in there.

EJ1 wilcox

I think this would be really useful.

Instead of wearing det cans all the time I could stick this on my dash and watch for the light.

Even after I'm tuned what if I get detonation from a bad tank of gas? Or because of some other factor?

I'm not going to drive around with det cans on all the time. So I definately think this could be a useful device. You see the light starting to flash, you let off the gas. Pretty simple. Since my OBD1 motor doesn't have a knock sensor this would be a nice addition and safeguard.

khaosman

I'd rather spend the money on a J&S Safeguard, which definitely senses the knocking and does something about it! (retard timing)

adseguy

The only problem is that knock frequency is an exact equation. Look it up.

It's dependent on bore size and such. So without entering these specs in somewhere the device cannot functional optimally.

I used Matlab to make a script using the Fourier transform to isolate frequencies and amplify them. I found some sound bits on the internet that had light knock on them used for the same purpose. I found the engine size plugged in the equation and My "meter" showed me that knock was evident. I haven't done this on my car since well....I don't knock nor do I care to even light knock...I just don't care engouh.

Cliff notes: These things aren't complete garbage, but people who rely on them too much like DSM people relying on their stock sensors do are dumb. Use your own ears and find out for yourself.

alpha

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by adseguy &raquo;</TD></TR><TR><TD CLASS="quote">Use your own ears and find out for yourself. </TD></TR></TABLE>



Be suprised what you can hear through properly setup det phones

PrecisionH23a

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by Semtec &raquo;</TD></TR><TR><TD CLASS="quote">Blah, they are trying to bank on consumer ignorance.
</TD></TR></TABLE>

For the most part.... yes. I doubt an experienced tuner would run one of these considering you can measure knock via most EMS's.

CivicDXRaceCar

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by adseguy &raquo;</TD></TR><TR><TD CLASS="quote">The only problem is that knock frequency is an exact equation. Look it up.

It's dependent on bore size and such. So without entering these specs in somewhere the device cannot functional optimally.

I used Matlab to make a script using the Fourier transform to isolate frequencies and amplify them. I found some sound bits on the internet that had light knock on them used for the same purpose. I found the engine size plugged in the equation and My "meter" showed me that knock was evident. I haven't done this on my car since well....I don't knock nor do I care to even light knock...I just don't care engouh.
</TD></TR></TABLE>

f=900/(pi*r) Each knock sensor is made to a certain design frequency. So you just have to find one around that calculated frequency. I don't understand why you went the the hassle of a fourier transform just to have a knock sensor that resonates at a certain frequency anyways!? Above about 5000rpm the thing is close to useless anyways so what's the big deal?


Modified by CivicDXRaceCar at 12:35 PM 5/11/2007

adseguy

Noise is the problem. That's why I went through all that trouble. Band pass filters and such wern't cutting it. If you can target a frequency and put it in frequency domain it's better then time domain.

CivicDXRaceCar

i think i follow. you mean noise on top of the signal from the knock sensor? so what your doing was signal conditioning right?

adseguy

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by CivicDXRaceCar &raquo;</TD></TR><TR><TD CLASS="quote">i think i follow. you mean noise on top of the signal from the knock sensor? so what your doing was signal conditioning right?
</TD></TR></TABLE>

Yes and Yes

blundar

Adrian really has the right idea here...

A couple things:

This is somewhat oversimplified. Knock is a sound, and like any sound resonating in a pipe, it will have harmonics. If you had sufficient DSP resources, you'd probably want to include at least the second and third harmonics in your bands. The other thing to think about is low-frequency energy. The amount of energy at low frequencies (i.e. below 1000hz) is GREATLY increased during periods of detonation, regardless of RPM. Also, when integrated closely with the ECU, windowing can be used to limit the amount of time before/after sparks are fired in order to reject a large portion of engine noise based simply on the fact it occurs at a time when knock would not happen.

See http://www.vems.hu/wiki/index....ction for a great discussion.

Actually, not quite true. There are essentially two kinds of sensors - flat response and bandpass. Flat response sensors have a (relatively) flat gain across the frequency range. These are generally best used with a digital knock monitoring solution. Bandpass sensors have a bandpass filter that responds with a higher gain to certain frequencies over others. Someone (BigMoose perhaps?) did some spectral analysis on a bunch of common sensors. I seem to recall that one of the GM vette sensors was pretty damn flat in its response.

Any experienced tuner would know better than to rely on measuring knock via most EMSs. Since this is HONDA tech, I'll stick to the EMSs most commonly found on Hondas:
-Crome/Uberdata/eCtune/Hondata/Neptune - joke. The stock honda knock board is a joke for anything other than an OEM motor with OEM components. Totally ineffective for anything other than making the check engine light come on. Anything based on this hardware platform that doesn't replace the knock board is subject to its limitations.
-AEM - Ineffective. Voltage + background noise table. Configurable, but primitive.
-PowerFC - Ineffective. Many false positives. Seems to be a pretty basic voltage vs. RPM threshold. Primitive
-Autronic - Don't know, can't comment.

So what conclusions can we draw? Any experienced tuner that relies on the knock detection facilities present in "most EMS's" is relying on inherently flawed systems. Outside of South Carolina, most experienced tuners know better.

Designing a proper acoustic knock detection system would ideally require close integration with the stock ECU, closer than is possible with say a J+S safeguard. Knowledge of the bore of the motor would be extremely helpful for configuring the system. The use of a flat-response sensor would probably be a good thing. A series of filters applied to the signal (by something like MATLAB, a DSP or a TPIC8101) would need to be used to get decent results.

integ401

How would the knocklink compare to the TurboXs knock lite? I've been searching on google to determine which is better. To me it seems neither is effecient because of the different sound levels that the engine makes.

Also from searching i found this thing called the knockblock. http://www.linkecu.com/product...Block
It has a Headphone jack to plug in and it hooks up to the knock sensor i believe. Also it seems like it can be used to datalog with say if you have the Innovative LMA3. It has different sound frequencies as well. Anyone try this out or know anything about it. I've been searching forever with very little luck

John_at_J&S

From the KnockLite thread on the Subaru forum:
http://forums.nasioc.com/forum...nt=76

Honda S2000 tuner sent this data log, comparing the AEM knock log to the J&S bargraph output:
http://www.jandssafeguard.com/....html

Team RS2

sounds like that can only be used on utec tuned motors, utec deosnt exist for honda/acura applications