It was for the turbo engines.  Not sure it worked well.  I had pondered some copper heating oil line, a flattened end, a bolt to the engine block, and a funnel in the cabin for my car.  In the early days of my worrying about detonation.

Here's the ZX sensor.

There's ideas out there - https://www.nicoclub.com/archives/diy-knock-detection.html

Edited April 22, 20204 yr by Zed Head

Thanks Zed. Do you have a part number for that knock sensor? I poked around and didn't find anything.

And I remember that Whisper 2000 thing! Back in the dark ages, us guys the engineering department would sometimes pool some money and buy stuff that we thought were funny or scams. Perpetual motion machines, X-ray glasses... That sort of thing. For entertainment.

The Whisper 2000 was one of the devices we purchased for our collection.    

This is the best I could do with part numbers (22060) ...

 

 

There are a bunch of more modern sensors out there.  Can't find the old ZX model.  As I understand them the frequency that they measure is important.  I wonder what kind of "signal"  they put out.

https://www.courtesyparts.com/?p=catalog&mode=search&search_in=all&search_str=22060

Edited April 22, 20204 yr by Zed Head

Sorry guys I missed a bit of the part number - it should be: 22060-W1502

There are a bunch of more modern sensors out there.  Can't find the old ZX model.  As I understand them the frequency that they measure is important.  I wonder what kind of "signal"  they put out.https://www.courtesyparts.com/?p=catalog&mode=search&search_in=all&search_str=22060  

 

  A knock sensor is basically an accelerometer that outputs a constant voltage wave relative to the acceleration - here is an actual plot, spot the knock ;) ...

 

 

What I understand is that each one has it’s own natural frequency and so does your engine!! So the two have to be matched. The only way I can imagine myself doing it is:

 

1. Strap a sensor with a wide range to the engine

 

2. Log a shed load of data at various RPMs and load conditions.

 

3. Set the engine up to knock / ping gently and log more data.

 

4. Perform a Fourier analysis to then take out the background engine noise from the knock reading and isolate the key frequencies and amplitudes.

 

5. Build an Arduino to connect up and perform FFT whilst driving.

 

The other way of course is to not get greedy with timing and leave it all alone!! [emoji1787]

 

 

Edit: forgot to post this - the equation below pops up a lot on various articles about finding knock frequency - not 100% the way to do it but people who tune cars and set up custom ECUs claim it has worked well for them:

 

 

Knock frequency (kHz) = 1800 / (3.14 x Piston Dia (mm))

With the tube and the funnel you can just do a FBT.  Fast Brain Transform.

Thanks guys for the part number info. I'll poke around some and see what I can learn.

I always thought knock sensors were piezoelectric elements. Hit it with a voltage, and it changes shape. Hit it with a change in shape, and it produces a voltage. And yes, they often (always?) have a resonant frequency. Was never a topic of study by me, but I'll dig a little.

1800 divided by bore circumference seems too easy. Maybe that works for old Volvos? 

Harvest the igniter piezo out of my grill lighter and bolt it to the side of my engine and measure the voltage out?

Or I could just not get greedy with the timing values. 

From:   https://www.nissanpartsdeal.com/parts/nissan-knocking-sensor~22060-w1502.html

     Part Fitment for 22060-W1502:

     Nissan 280ZX 1980-1983 L28ET

     Nissan Pulsar NX 1983-1985 E15T

From Wikipedia:

     The L28 is a 2,753 cc (2.8 L; 168.0 cu in) 12-valve engine. Bore and stroke is 86 mm × 79 mm (3.39 in × 3.11 in)

     The E15 displaces 1.5 L (1,488 cc) from 76 mm × 82 mm (2.99 in × 3.23 in)

 

So much for the 1800 / bore circumference.  Haha!

Edited April 23, 20204 yr by Captain Obvious

With the tube and the funnel you can just do a FBT.  Fast Brain Transform.

  FBT- - Brilliant!!!! [emoji106][emoji106]

 

 

 

 

  So much for the 1800 / bore circumference.  Haha! 

 

 

 Agree - I hate constants like that without explanation. Here’s a site that does it for you. 

https://www.phormula.com/KnockCalculator.aspx

 

 

 

So, 1800/(pi x 87mm) = 6.6kHz

 

Knock sensors AFAIK are made with specific resonant windows and one of them has a centre freq of 6.5khz.

 

 

 

 

I always thought knock sensors were piezoelectric elements. Hit it with a voltage, and it changes shape. Hit it with a change in shape, and it produces a voltage. And yes, they often (always?) have a resonant frequency. Was never a topic of study by me, but I'll dig a little. 1800 divided by bore circumference seems too easy. Maybe that works for old Volvos?   

Or I could just not get greedy with the timing values. 

 

 

 

 Agreed. Again from my limited knowledge - they are fundamentally Piezo Electric, crystal based devices that output mV per gramme. 

I’ve asked myself this many times, do the extra 3-4 degrees of advance really make enough of a difference in road driving to justify all this hassle and potential risk? (Notice how I brought the thread back on topic! [emoji41][emoji38])

 

 

 

Pro way:

hook a wide band mic ( piezo based) to your's engine car and run it through this software:

https://www.vb-audio.com/Spectralissime/

get the car to knock and note the dominant frequency.

Build a piezo sensor cavity tuned to this knock frequency

done.

 

Enthusiast Ways:

If you want to do it in digital, use a general purpose piezo sensor and run the signal into a processor where you filter the knock frequency

If you want to do analogue processing, send the sensor signal to a graphic EQ in your glove box and have only the knock frequency band raised.  Send this to your retarding device..

8 hours ago, AK260 said:

  do the extra 3-4 degrees of advance really make enough of a difference in road driving to justify all this hassle and potential risk?

Well put, and while it would be neat to mess around with it, I don't think I'm going to be pushing that limit that far. Haha!