I'm glad to see some interest in my humble det detection exercises.
For those who want to make a mechanism like this electric knock detection kit, I tried to arrange this post as a step by step guide - with explainations and illustrated by photoes. Those who already made one up, I hope, will also find this post interesting. Feel free to comment, make suggestions and advices. For native english-speakers - excuse my grammar and my spelling. I'm still learning.
OK, here goes
1)
A little theory
There are two wide-spreaded and commonly used types of knock sensors in automotive industry - resonant and non-resonant, the latter is also known as wideband. Resonant sensor is constructed and adjusted such that when it picks up a vibration of a specific (narrow) range of frequencies, it begins to oscillate in its natural resonant frequency, generating output signal. I'm no expert at this, but you can read more on this if you ask google about "resonance force sensor", "resonance strain sensor" etc. What is sufficient for us to know here is that the signal it generates has about 30000+ Hz frequency (depending on application and internal structure) so it can not be heard by ear even if it has enough amplitude.
The oher type is non-resonant, "wideband", what means that it translates any noise it can "hear" and at the frequency similar to its input vibration, so theoretically we can connect our headphones directly to it. The main concern here is the amplitude of a signal - it may be insufficient (it may also be too big), so we want to have the ability to amplify it (and to separate the freq band we are interested in hearing from other mechanical noise - e.g. valvetrain and normal combustion sounds). To make this kind of sensor practical in automotive use, Bosch actually offers us not only this piezoelectic sensor, but also a programmable electronic filter to build into ECU's architecture, to pick a det among other noises, and to transform a sound stream into a signals recognisable by CPU [0,1].
2)
From theory to practice
What we need (and what this whole system is centered around) is the standard round bosch-style knock sensor. Actually it is a piesoelectric sensor, that generates electric pulses when it is subjected to a force. The same exact principle is used in microphones, so we can say that bosch knock sensor is a microphone.
Well, so much for the words, let us see some pics.
Here is the sensor:
Here is the plug side. There are two types of this exact sensor available - with two wires - like this - and with three wires and a little extention cable, shielded - used on porsche, some VW's etc. If you are not going to make shielding work, stick to this type. It is much easier to access tight underbonnet areas without any "tail".
Here is side and bottom view. This bottom flat surface is actually sensor input and should contact similarly flat solid surface of an engine.
For testing purposes I borrowed a stock bosch-like JECS sensor off a VQ, or so I've been told. It's been tried, but sounded crappy, a lot of extraneous noise, so if you decide to go second-hand one, i strongly advise against it. The new knock sensor is not that expensive. I ended up buying a new bosch one, P/N of an item without a "tail" is 0 261 231 006.
Here is a plug for it, audi made:
Yep, 100% OEM VW-audi group one
Stock nissan plug for it is crap and has very inconvenient lock to use if you plan to connect and disconnect your sensor often, which is obviously the case with the system like this. Note that the plug is standard type, it does not have to be off a knock sensor. This particular one is off an injector harness.
Ok, so we've got a sensor and a plug. Now we need an amplifier. You can use your old radio, player - whatever you like. I've found an old cassette player.
The coolest thing is that it is also a dictaphone, so it has a microphone input, also a 3.5mm jack socket
Now it is time to connect out sensor to the amplifier. I used 3m of 3.5mm jack extention cord:
I bought two of these, because I was not sure if 3m should be enough. But if it should, i didn't want to sit wrapped in wires, so two 3m cables for me. One has sensor plug soldered to it:
Now it is time to have the sensor connected to an amplifier. We do not want to write some noise to a tape and to listen to it later. Well, maybe we do, but to be honest, I can not remember, when it was the last time I saw an audio tape.
Anyway, we need to be able to hear some engine noise instantly, the same moment it comes from an engine. So the idea is to connect the microphone input directly to the head output of the player. This operation may require the head to be dismantled, but it was not necessary in my case - the head was connected to a board with wire. This wire will be connected directly to the mic input.
Grey wire here goes from the head to a board. Two other, green and red, from mic socket to the head. If you look carefully at this pic, you'll see "mic" stamped on a board. This is the place where 3.5mm socket is mounted.
Better seen here:
At this stage I have been advised to include one more adjustment component in the schematics. To be able to adjust signal volume, we will put a trimming resistor inline. It will work as a voltage divider.
If principle here seems inclear to you,
this link may help you remember Ohm's law.
I've been told that 100 kOhm one would do the job. Here is the theory (scheme):
And here is the realization:
Adjustments can be made by rotating the adjusting screw with a screwdriver.
This is actually it if you can not wait to listen to some knock (not necessarily of your engine). But I myself just wanted to be a little more autonomous, so to speak. So I've spent some time being busy with
3)
Optimizing the package
By optimization I mean that we already have a working knock detection solution, it just can be tweaked a little. That is, I just don't like batteries. So I looked at the player once more and found a socket for power supply:
.
Later that day I found an AC charger for a really old motorola cell phone. Its plug fitted to a power socket perfectly:
I also had a car charger for my current SonyEricsson, which i decided to sacrifice for a noble cause of powering my knock detector. But player needed ~3 volts, when SE charger delivered full 4.7 volts. So two inline diodes were used, the voltage drop on each is ~0.8 volts. After the soldering it looked like this:
Hidden under a black shrink-wrap are two slim diodes.
4)
Lab testing
Here is how complete kit looks like in my case:
Now it needs a little adjustment.
Insert a couple of batteries in the player. Connect the sensor and headphones. Press "play". Rest sensor against some solid surface (not necessarily flat). Try a tap in your bathroom. Turn on water. Listen to the hissing noise, place stock player adjusting knob in middle position. Turn adjusting screw on the trimming resistor until the volume level is comfortable, but preferably on a low side, because your tap is quieter than your engine.
5)
Conclusion
Now the system is ready to use. All you need to do is to find a hole on the block, preferably near the combustion chamber, with some flat surface around the hole - sensor will pick more useful sounds from this surface. Factory knock sensor location works the best, but you might need to make an adapter with internal M8 thread for the sensor. Bolt your sensor in there, trying to mount it solidly, otherwise it will fail to pick up all the noise, but not overtighten the bolt. Put an extention cord in any convinient way to a vehicle cabin, plug it into the player, plug phones and power source, put phones on. You're now ready to listen to the car. Start your laptop, plug consult wire into the consult port and into laptop, load nistune, start the car, and open primary ignition map. Now you just need to push the "play" button and start to TUNE.
6)
Additional notes
Since this was made 4 mounths ago, I may have forgotten some minor details. Those that I remember are:
- When I first switched it on, I tried to listen to everything.
water plumbing, gas plumbing, TV, my desktop, walls, floor. It works perfectly on solid surfaces, but it is not by any means a stethoscope - I did not hear anything from human body
- So far I listened only to CA's and RB's. On these det sounds like a "tick" - something between a spilled rice and a nail that is being hammered to a block. Hard to describe, but it can be clearly heard and recognized
- RB25 is equipped with resonance-type sensors from factory. Their output frequency is ~44kHz, or so I've been told. As far as I know, CA18DET is not equipped with a knock detection in its factory form
- I'm not sure on this one, but I think that detonation sound depends on the block material - I mean, vibroacoustic qualities of aluminum and cast iron are not the same and this may affect the way ping sounds like. I guess SR's and VQ's sound differs from RB's and CA's
- Phormula web site states otherwise - they have a frequency calculator that asks only for your cylinder bore diameter. Prior to building my KDK, I tried to learn as much as internet could provide me with on detonation phenomenon. Besides a couple of really informative web pages, that can be found using right words in google, I highly recommend reading SAE papers on this - 900488 in particular, and a some others. Nissan also holds three or four patents on knock detection and recognition. All these sources mention factors that affect det sound frequency - chamber temperature and bore diameter are the most important.
- I thought about sound filtering but, as I said before, I just rely on my brain in doing all the filtering more than i should on electronics. Nevertheless, if you decide that filtering is the way to go then the basic idea is to read bosch whitepaper on their sound processing unit first. Then you can try to find out what car has a cylinder bore diameter similar to yours (and is equipped with bosch-style knock sensor). You then buy a ECU off of that car wreck and desolder sound processing module. Now you can even roll your own digital knock monitor
- I'm pretty certain that I can NOT detect preignition with this kit. Also I'm certain that none of existing electronically filtered vibration-sensor-based
systems can. Cylinder pressure and temperature monitors are best at this. I've been told and I read that preignition is a very silent event that can very hardly distingushed among all other combustion noises. It is, nevertheless, an engine killer just like det.
Sorry for such a long post