The Knock Box

[jblaine]

http://www.theknockbox.com.au/

* Registered users of the site do not see these ads.

[cdvma]

Neat. No price listed so I assume its too much for me

[happasaiyan]

haha, i thought this was going to be a post about your car

[CMJ]

Originally Posted by cdvma

Neat. No price listed so I assume its too much for me

I remember reading that the cost was over 1k.

[jugger]

actually i rememeber ~$400

[ride5000]

i like the noise cancelling headphones!

[gpatmac]

350lbs or about $635.

http://www.thor-racing.co.uk/KNOCK_B...LNK_KNKB_K.asp

[colby]

I've been thinking about making a product like this that incorporates a FM modulator so you can just listen through your car stereo / or FM headphones. I'm curious how much interest there would be. You could also do some cool stuff interfacing to a PC via the sound card input and using FFTs and normalization to a recovered tach signal to see the knock events in time / energy plots referenced to TDC...

[SaabTuner]

Originally Posted by colby

I've been thinking about making a product like this that incorporates a FM modulator so you can just listen through your car stereo / or FM headphones. I'm curious how much interest there would be. You could also do some cool stuff interfacing to a PC via the sound card input and using FFTs and normalization to a recovered tach signal to see the knock events in time / energy plots referenced to TDC...

Knock sensors are just a form of microphone, so I see no reason you couldn't hook it directly up to your car's stereo with the right adaptor. You might need to use a pre-amp of some sort if the noise floor on your stereo isn't very low, or if the signal is very weak.

As for the normalization and such, most modern ECUs already do all that. They have a "knock window" referenced to TDC as well as a specific set of frequencies. Good ECUs will have a RPM/Load "noise" map, which amounts to a normalization map, though it doesn't continuously renormalize; it's just set once at the factory.

-Adrian

[TheMadScientist]

Originally Posted by colby

I've been thinking about making a product like this that incorporates a FM modulator so you can just listen through your car stereo / or FM headphones. I'm curious how much interest there would be. You could also do some cool stuff interfacing to a PC via the sound card input and using FFTs and normalization to a recovered tach signal to see the knock events in time / energy plots referenced to TDC...

I have used both FM modulators and wired adapters with my TXS TunerPRO. The FM modulators pick up on electrical noise in the car and you hear that noise as det. Once I switched to a wired adapter that "noise DET" was gone. There was one other guy around here that ran into the same problem. I would stick with a wired adapter unless you want to chase random "noise DET".

TMS

[colby]

Originally Posted by TheMadScientist

I have used both FM modulators and wired adapters with my TXS TunerPRO. The FM modulators pick up on electrical noise in the car and you hear that noise as det. Once I switched to a wired adapter that "noise DET" was gone. There was one other guy around here that ran into the same problem. I would stick with a wired adapter unless you want to chase random "noise DET".

TMS

Good point. The idea behind using the FM modulator was to simplify the installation for people utilize their existing stereo. Sound like from your experience that is too much of a comprimise. It would still be interesting to see how well a PC could analyze the knock sensor signal to detect really low level knocking using some decent signal processing techniques.

[gpatmac]

I don't know if this is what you were thinking, but, though I've never really used it like this, I've experimented with hooking one of those cassete shaped thingies (what folks use to play their MP3 players through their car stereo) into my TUNA.

Seemed to work just fine, but I prefer putting headphones on (when I'm tuning in the passenger seat, of course.)

[bboy]

I have a post somewhere on here about hooking up a knock sensor to your PC and using simple (usually free) sound processing software to filter out other engine noise. The valves in my engine tap at about the same frequency as knock, but you can still hear it.

I just bought a Bosch knock sensor at the junkyard and wired it to a mini-audio jack (radio shack has one's that come with alligator clips), and plugged it into my computer. Set the filtering software to "play through," filtered out all sound except 5-10 kHz, and it worked great. You can either play it through head phones or right out of the speakers of the computer.

The valve noise is atrocious--one person told me it's the buckets rattling on top of the valves that are making all that racket.

[Jon [in CT]]

Originally Posted by bboy

The valve noise is atrocious--one person told me it's the buckets rattling on top of the valves that are making all that racket.

Hence, the concept of a "knock window" where you only listen during the period when knock is likely to occur, which happens to coincide with a period when the valves aren't closing (at least on 4-cylinder engines). According to p. 9 of http://focus.ti.com/lit/an/spra039/spra039.pdf:

Detection Strategies
Knock detection strategies use the output of a signal conditioning stage to compare with a reference to determine the presence or absence of knock. Most systems today use windowing to isolate periods during the cylinder’s firing cycle for analysis when knock is possible. There is a window from approximately 10° to 70° after top dead center (ATDC) of the piston’s cycle when detonation is most likely to occur for the firing cylinder. The detection algorithm is run only during this window for that cylinder. By eliminating possible false trigger sources, such as valve closing, the detection algorithm is more robust. The time this window is active varies with engine speed from 20 ms at 500 RPM down to 1.25 ms at 8000 RPM. Tracking changing engine speed to calculate this time variation requires hardware or software overhead for implementation.

That TI document, BTW, is a very good introduction to knock detection.

BTW, I've read at least one Subaru patent that adjusts the knock window based on IAM (e.g. if IAM is low, the knock window is shifted farther from TDC).

The J&S Safeguard system tries to employ a "knock window" but, unfortunately is unable to identify TDC or crank angle degrees ATDC. Their "window" is described in a post at http://www.eng-tips.com/viewthread.cfm?qid=54276 but it doesn't necessarily exclude valve closing noise:

Hi to all. My name is John Pizzuto, and I'm the designer of the J&S SafeGuard.

For a background on knock sensors, read SAE Technical Paper Series 900488, Combustion Knock Sensing: Sensor Selection and Application Issues

Authored by Steven M. Dues, Joseph M. Adams, and George A. Shinkle: Delco Remy Div, General Motors Corp.

In the mid '90's, a GM engineer sent me a list of GM sensors, showing part numbers, resonant frequencies, parrallel load resistance, and mounting style.

At the time, they were producing sensors in these frequencies: 5.2kHz, 6.0kHz, and 7.0kHz.

A chart from the SAE paper shows that the fundamental frequency of a 75mm bore is about 7kHz, a 94mm bore is about 6.0 kHz, while that of a 110mm bore is about 5.2kHz. The chart shows a fairly linear relationship of frequency to bore diameter.

The paper terms the GM sensors to be "broadband resonant", with "bandwidths approaching one thoundsand hertz".

The detector in the J&S unit is in software, running on a 68HC11. The unit controls dwell and timing as well.

Software sets up a knock window, since knock is expected to occur around TDC. The window opens 32° after the ignition trigger, and remains open for 44°.

A load qualifier is used to disable the detector under light engine loads and decel, where piston slap can cause a large interfering signal.

The unit uses a proprietary algorithm to develop a threshold, against which the signal is compared.

Not ideal.

[Jon [in CT]]

The classic knock detection method is emodied in the TI chip TPIC8101 - Knock Signal Interface. It is a reincarnation of the Harris HIP9010 chip (do a search, it's no longer available). I've seen a Subaru patent that essentially implements this processing in the digital domain.

[Jon [in CT]]

And then there's the idea that you can discern knock by its "signitaure" instead of its relative loudness. That idea is espoused by http://researchnews.osu.edu/archive/knock.htm and you can see its embodiment here:

where the "backslash" lines represent knock and its declining frequency.

And then there's National Instruments' approach to knock detection, described at:
http://www.ni.com/svix/engine_knock.htm
http://www.ni.com/pdf/csma/us/collins.pdf

[Jon [in CT]]

I have to confess that my responses here were due solely to:

Originally Posted by colby

It would still be interesting to see how well a PC could analyze the knock sensor signal to detect really low level knocking using some decent signal processing techniques.

It's not at all clear (to me, at least) that any PC can provide a realtime answer to the knock question. However, it seems very clear that a PC could be programmed to analyze (post process) a knock sensor's output and decide whether or not knock was present.

And I'm sure any useful analysis would empoy DSP and FFT.

[bboy]

I think you could design a meter that eliminates almost all the valve noise if you could isolate the frequencies.

Interesting web site. I agree that knock has a certain change in pitch that makes it distinct from all other noise, hence even if you are hearing the valves click away, knock sticks out. The human ear is great at detecting the change in pitch (frequency). I'm sure you could get a computer to do it, but it won't be easy.

[cdvma]

A PC can do it in realtime, there was a guy who made his own filter in LabView for his Porshe tuning days.

So far I've used a stock knock (not the stock one, another stock one) sensor spliced into a headphone jack and pluged into an amp (laptop) and then plugged into headphones for listening. I use a set of Bose noise canceling headphones (QC2s and yay for employee discount otherwise I wouldn't have them). So far so good and soon I'll have the TXS Tuner at my door. The bonus of the raw audio is you can record it with a laptop and use the spectrograph as shown above. That works well but I've found our engines to be VERY noisy with the spectrograph and you have to mess with it to clean it up. I have had more luck just listening than trying to analyse.

[Hitokiri]

I am currently using labview for knock detection. send me a pm if you are set up with labview and want to check it out. You will need the signal processing toolset to use it however which is annoying.

Todd

[bboy]

Originally Posted by cdvma

I have had more luck just listening than trying to analyse.

Same here.

[John at J&S]

Todd: Is your J&S still connected? It would be interesting to see how it compares to the labview.

The J&S employs a knock window. The fact that it doesn't precisely know where TDC is not important. The window opens near TDC, and remains open for 44°, which is long enough to reliably detect knock.

Here's an AEM data log from an S2000 Honda, showing the knock volts vs J&S detector output. Looking at the log, it's difficult to tell when the AEM actually makes a detection.

http://www.jandssafeguard.com/Giles/AEMknocklog.html

Giles's Gilbert's site http://www.holyrollerdyno.com/

He spent a year fine tuning the AEM maps on his dyno, before he installed the J&S. As soon as he tested it, he emailed back "at last, a real time knock detector that I can rely on".

[jblaine]

US Dealer @ Motec:

Hi jeff,

Thanks for your interest in the knock box.

the pricing for the knock box is
racer net $1537.26
plus tax if you are in Ca. or NC.


units are in stock

[jblaine]

Originally Posted by gpatmac

350lbs or about $635.

http://www.thor-racing.co.uk/KNOCK_B...LNK_KNKB_K.asp

Not the same product.

Knock Box vs. Knock Block

[John at J&S]

Here's another knock detector:

http://www.thielert.com/en/automotiv...ment/knock.htm

I have a price sheet from a few years ago, and the unit with cable and sensor was 3700 Deutch Marks ($2300).

It has a digital output to signal a knock event, and an analog output proportional to knock intensity, for connection to an ECU as a timing modifier.