Knock Detection with variable sensitivity over RPM

A forum for discussing the MegaSquirt related (but non-B&G) board development, assembly, installation, and testing.

Moderators: jsmcortina, muythaibxr

Locked
dfabro
MS/Extra Newbie
Posts: 15
Joined: Sat Apr 19, 2008 5:05 pm
Location: Brazil

Knock Detection with variable sensitivity over RPM

Post by dfabro »

Hello to all. (Sorry by my English)

This article is about a circuit capable of detecting detonation with excellent precision. It was tested in VW 1.8 Turbo Engine with the MS-Extra1 pcb v3.0.
This is an enhancement and much better of the existing knock detection circuit.

SPECS
RPM range: 2000-7200RPM tested.
Overall sensitivity adjustment according to the noise of each engine.
Knock Detection with variable sensitivity over RPM.
A 4.8kHz high-pass filter and a 7.4kHz low-pass filter of 40dB per decade.
Four LED's to indication of actual knock level.
A 0-5v analog output signal (for Datalog)

BASICS:
The knock detection as become increasingly important in the years. The principle of finding out if the engine is knocking or not is expected to be simple, but it can become much more complex when the subject is analysed with greater depth.
As all Engine Tuners know, the loss of many engines, particularly the high-
performance and boosted, can be attributed to knock. And as every Vehicle manufacturer knows, the control of knock is directly linked to longevity, economy and emissions of an engine.
Generally speaking, the power of an engine is limited to the amount of air that it can accept, and the ability to burn the air/fuel mixture on the threshold of detonation. The control of Ignition Timing (point) must always be done within the limits of detonation so that consumption and performance are optimized. If the curve is too late, the consumption greatly increases, and performance decreases. The occurrence of knock, even in sub-audible levels, can still be very harmful to the engine.
Image
This circuit can detect the presence of knock long before a professional can hear. When you can hear the engine "knock" at a level higher than the noise of the engine, it’s already too late.
The benefits are the optimization of the combustion process, meaning excellent performance, maximum use of the power without compromising the longevity, with consumption and emissions always under control, thus, maximizing the potential of your vehicle and protecting your investment. Due to all these benefits, it becomes an essential tool, when the target is the maximum in performance, safety, economy and durability.

WHAT IS KNOCK?
The start of the burning mixture of fuel in a gasoline engine must be controlled by the ignition of the spark plugs. However, in many situations, the high temperatures and pressures that the mixture is subject during the end of the compression time, can lead to a self-ignition sudden and uncontrolled. For this spontaneous combustion and unwanted, there is the name of Knock,
that should be avoided at all costs, because if it occurs with high frequency can seriously damage the engine.
When something is not within the specifications - very high compression rate, time of ignition well advanced, spark plug incorrect or poor quality fuel, for example. The flame spread so disorderly, which brings suddenly the pressure and temperature in the combustion chamber. There will be a spontaneous combustion, a second flame on another point of the chamber
This front of flames moves up to ten times faster than normal, causing huge pressures and creates a metallic noise, which is similar to that of balls of gude within a cup. It is the detonation or the common vocabulary "knock".
The detonation causes damage to produce a very rapid movement of the combustion gases, that beats against the extremes of the chamber and generate heat. If the detonation persists for some time, raising the temperature of components can lead to a drop in the head, a hole in the piston or the cylinder casting.
Image

The effects caused by the detonation phenomenon depends directly of its duration and
intensity. While moderate, the knock does not produce significant changes in performance and engine durability, but when intense and/or for prolonged periods, can damage it.


AS THE CIRCUIT WORKS :
The noise generated by knock causes a vibration in the structure of the engine (block and Head). The Circuit receives electrical signals from sensors installed on the engine (knock sensor). The task of detecting the Detonation is to capture the sound of knock above the background noise of the engine, through a filter that allows the passage of a specific frequency range, corresponding to the frequency of the sound of Detonation. By filtering the noise of the engine, it allows detection of the knock as soon as possible. The level of background noise specific and the changes in RPM of each engine, make the process of filtering a difficult task.
Thus, for an efficient detection system, it’s important to vary the sensitivity throughout RPM range. If you have only one setting of sensitivity, when the detection of knock is sensitive to the noise of an engine at high RPM, it will not be able to detect knock at low RPM. The circuit conducts an increase of sensitivity at low RPM, noticing any appearance of noise across the engine RPM range.
The sensitivity will gradually decrease in accordance with the increasing speed of the engine (adjustable). If contrary, the circuit will detect false detonation, which is very harmful, so as not detect it. The dyno test showed that the circuit is able to detect the precise level of knock from 2000RPM until 7200RPM, in a 4 cylinder engine with 22 psi of boost, using ethyl alcohol (ethanol) with octane equivalent of 93oct. Alcohol from sugar cane (Brazil).
The circuit indicates four levels of knock, according to the intensity and duration. Small knock for long periods can be harmful.
Image
The stage of filtering has a low pass filter and a high-pass filter in series, which will only allow the passage of the knock resonant frequency.
The circuit reads the RPM of PCB3 output JS3(tach-out). According to the frequency of this signal, it generated a DC voltage (Vref), that is RPM directly proportional.
This voltage is used as a reference for determining the detector sensitivity.
You can connect two diodes with R29 and R26, and linking them in the knock circuit, I believe that this will not change the signal sent to igniter. the frequency-voltage converter, must be calibrated so that the tension of reference is 1.20v at 1500RPM and 3.7v at 7000RPM. to facilitate, you must have a stimulator.
The diodes should be singled out for the knockcircuit, because it needs positive pulses.

KNOCK SENSOR:

The Knock Sensor is usually set in the engine block or in the head and has a piezo-electric crystal as sensor element. This material when subjected to mechanical deformation, generates electrical voltages on its surface. Thus, the sensor is able to capture ("listening") the vibrations caused by the parties that move in the engine and also by the detonation phenomenon, turning them into electrical signals.
The installation of the knock sensor must be made on a completely flat surface and free of impurities. The torque applied on the screw for fixing the Knock Sensor is between 2.0 to 2.5 Kgf.m.
The location and the tightness of the knock sensor should not be changed after the process of adjusting the circuit sensitivity.
The knock sensor should be positioned in the block highest possible (the closest to the head).
Try not to assemble near the camshaft, valve train, rocker arms, rods, That is to reduce the background noise.
Keep the lead of the knock sensor away from the discharge, or use thermal insulating and not pass near the ignition system (spark plug, distrubutor and coils).
Use only knock sensor type non-resonant.

The illustration was copied from an article made by a Brazilian university professor.
Source Article NGK Spark Plug (2004).

ABSTRACT
Fernandez, B. O. (2006). Considerations on other jobs of the knock sensor for
the electronic control of Otto cycle engines using spectral analysis. Dissertation
(Master) – Escola de Engenharia de São Carlos, Universidade de São Paulo, São Carlos-
São Paulo, Brazil, 2006.
This dissertation analyses the suitability of using the knock sensor as a feedback element for the electronic control of internal combustion engines.
The proposed approach consists of using the knock sensor, originally installed by the engine maker in order to eliminate the spontaneous combustion effect, to sample the mechanical vibration produced by the engine. This vibration, which results from the moving parts and the natural oscillation of the system in combustion, produces an output in the range of audible sound. This research contemplates using the spectral variation of that sound to estimate information about an engine operating with different fuel blends.
http://op6wta.bay.livefilestore.com/y1p ... f?download
According to him: In a traditional Otto cycle engine, the mixture of air and fuel is carried by the combustion chamber and compressed by the piston movement. The burning of the mixture begins with the ignition spark plug at some moment near the TDC (when the volume in the combustion chamber is minimal).after that, the pressure wave spreads by the combustion chamber. The last gas to be consumed by a flame front receive the name: Final gases and are usually found away from the ignition point.The final gases are heated and compressed by the piston movement and the pressure waves, starting to react homogeneously.If reach a certain condition, the gas will come into spontaneous combustion. If reach a certain condition, the gas will come into spontaneous combustion. Under normal conditions, the flame front consumes all the mixture, rather than a spontaneous ignition can occur and thus observe a normal combustion.Otherwise, if the final gases start the combustion before the end of the flame front, the ignition explosive results in schock waves in the combustion chamber.
Image
Image

The Mystery of Detonation
By Robin Tuluie, Jun. 23, 2000
Detonation usually happens first at the pressure wave's points of amplification, such as at the edges of the piston crown where reflecting pressure waves from the piston or combustion chamber walls can constructively recombine - this is called constructive interference to yield a very high local pressure. If the speed at which this pressure build-up to detonation occurs is greater than the speed at which the mixture burns, the pressure waves from both the initial ignition at the plug and the pressure waves coming from the problem spots (e.g. the edges of the piston crown, etc.) will set off immediate explosions, rather than combustion, of the mixture across the combustion chamber, leading to further pressure waves and even more havoc. Whenever these colliding pressure fronts meet, their destructive power is unleashed on the engine parts, often leading to a mechanical destruction of the engine.

In this case, the pressure waves generated by flame front, was that caused the second flame. Without the ignition of the spark plug, this would not happen.
The pre-ignition occurs when the air-fuel mixture in the cylinder (or even, to enter the cylinder), starts the flame, before the spark occurs. This pre-ignition is caused by another source of ignition, other than generated by the spark plug.

When you retard the ignition advance, this results in reducing the intensity of pressure waves, and that it does not coincide with the TDC, this hampers the occurrence of detonation.
The pre-ignition can occur regardless of the progress of ignition.

In an attempt to better the process of detection by piezo-electric knock sensor , I implement a band-pass filter. As the knock sensor non-resonant plays virtually all audible frequencies of engine noise, a filter is necessary to separate only the resonant detonation frequency.
The resonant frequencies excited by the presence of knock depend on the geometry of the combustion chamber and the speed of sound in the cylinder charge. These resonant frequencies are typically estimated by assuming an acoustic model for the combustion chamber. For a homogeneous gas filled, acoustically hard walled ideal cylinder, the resonance frequencies are given by Draper’s equation.
A document of SAE Technical Paper Series 900488, shows that the fundamental detonation frequency in a cylinder of 75mm is 7.6 kHz, one of 94mm is approximately 6.1 kHz, while a piston with 110mm is 5.2 kHz. The document shows a reasonable linear relationship , the frequency with a cylinder bore. The following equation can be used to estimate the resonant frequency of detonation for a particular engine:

Image

Where the resonant frequency is measured in Hertz and the cylinder radius in metres

According to this equation, the knockcircuit allows detection reasonably accurate, for engines with cylinder of 75mm to 120mm.

I am preparing a PCB and developing a project with HIP9011
I have to do more tests, but until now, the unit was very efficient.
grateful for the opportunity to explain my work. I am available to answer any questions.
Last edited by dfabro on Mon Apr 28, 2008 9:49 am, edited 12 times in total.
dfabro
MS/Extra Newbie
Posts: 15
Joined: Sat Apr 19, 2008 5:05 pm
Location: Brazil

Files

Post by dfabro »

AbeFM
Super MS/Extra'er
Posts: 875
Joined: Wed Dec 05, 2007 1:40 pm
Location: San Diego, CA
Contact:

Post by AbeFM »

OH! I really like the idea, this is just what I want. I will have to look at it when I've had a little sleep, but so far I like it.

Do you just run the output to the MS, and let it decide (based on RPM) what the knock is, or just have it all from your circuit?
2000 VVT Miata turbo, MS3Pro

Contact me if interested in a MS-II 2nd gen NB Miata PnP board.
dfabro
MS/Extra Newbie
Posts: 15
Joined: Sat Apr 19, 2008 5:05 pm
Location: Brazil

Vref

Post by dfabro »

Hi
The circuit reads the RPM of PCB3 output JS3(tach-out). According to the frequency of this signal, it generated a DC voltage (Vref) directly proportional to the RPM.
This voltage is used as a reference for determining the sensitivity of the detector.
You can connect two diodes with R29 and R26, and linking them in the knock circuit, I believe that this will not change the signal sent to igniter. the frequency of a voltage converter, must be calibrated so that the tension of reference is 1.20v at 1500RPM and 3.7v at 7000RPM. to facilitate, you must have a stimulator.
The diodes should be singled out for the knockcircuit, because it needs positive pulses.

Already available, a 0-5v analog output signal, proportional to the level of detonation, especially developed to datalog.
I am preparing a PCB, soon
so they made all tests
dfabro
MS/Extra Newbie
Posts: 15
Joined: Sat Apr 19, 2008 5:05 pm
Location: Brazil

Only to help ......

Post by dfabro »

What are the possible detonation prevent solutions?

After identifying the knock occurrence with Knockcircuit, you can make some steps to avoid them. Through MegaLogViewer, you can change the ignition time or Richening up the mixture (for example) only in the rpm/map range where the knock is occurring. Thus, the engine operates with the highest performance across of useful RPM range.

Below is some solutions in order of priority in knock prevent:

*Reduce the Spark Advance - As the spark advance increases, the knock tendency also increases.The engine rpm and manifold vacuum/pressure are the main factors that determine the volumetric efficiency, and this which sets the ideal spark advance.The spark point should occur in the exact moment, so that the maximum cylinder pressure happens between 15 and 20 degrees ATDC. We must reduce the advance just enough to stop knock, and only at the point where it is occurring (RPM, Map and Temp).
Image
By Lars Eriksson.
Three different pressure traces resulting from three different spark advances. The different spark advances are:
SA1: spark advance 32.5 degrees (BTDC)
SA2: spark advance 22.5 degrees (BTDC)
SA3: spark advance 12.5 degrees (BTDC)
The optimal spark advance is close to SA2, With too early ignition timing the pressure rise starts too early and counteracts the piston movement. and this is a detonation provocative.

*Increasing fuel octane - The octane determines the fuel ability in resisting the self-ignition, that is, their ability to wait for sparks to then start the flame (combustion).

*Reducing the dynamic compression rate - There must be a balance between the cylinder VE and the compression rate. In an boosted engine, the pressure in the manifold should be consistent with the compression rate, avoid the combustion chamber overheating, it causing knock.

*The decrease in volume of the combustion chamber caused by charring, is a knock generator too.

*Richening up the mixture, AFR below 14.7:1 or lambda = 1. By Klaus Allmendinger - Richening up the mixture results in a slower burn, moving the pressure peak later where there is more leverage, hence more torque. Also the pressure peak is lower at a later crank angle and the knock probability is reduced. The same effect can be achieved with an optimum power mixture and more ignition retard. Optimum mix with “later” ignition can produce more power because more energy is released from the combustion of gasoline

*Decrease the manifold air temperature

*Increase humidity admission air

*spark plug with adequate heat graduation

*Making more efficient the engine cooling system

*Combustion Chamber High Turbulence

etc...
Last edited by dfabro on Tue Apr 29, 2008 7:56 am, edited 5 times in total.
cguenther
Helpful MS/Extra'er
Posts: 47
Joined: Sat Mar 03, 2007 5:41 am
Location: Germany

Re: Vref

Post by cguenther »

dfabro wrote:I am preparing a PCB, soon
so they made all tests
Very interesting, please keep us updated!

If a PCB is available - ill take two!

Best regards
Carsten
thedge
Experienced MS/Extra'er
Posts: 177
Joined: Tue Jun 28, 2005 1:54 pm
Location: Canada, Eh?

Post by thedge »

Sounds interesting!!
1986 Porsche 944 Turbo
2.7 liter MID sleeved big bore, Wossner pistons, Vitesse Stage 2 turbo, coilovers, various weight deletes, and loads of other stuff.
wes kiser
Super MS/Extra'er
Posts: 1402
Joined: Tue Jan 03, 2006 2:49 pm
Location: Charlotte, NC

Post by wes kiser »

Any idea when a PCB will be available?
86 Rx-7, swapped to 2.3 ford turbo (BW EFR 6758), ms3/ms3x sequential fuel /waste spark, ls2 coils
88 Tbird 2.3t, Microsquirt Module (PIMP), TFI ignition
masterx81
Master MS/Extra'er
Posts: 776
Joined: Mon Oct 25, 2004 7:36 am
Location: Asti - Italy

Re: Knock Detection with variable sensitivity over RPM

Post by masterx81 »

Hi dfabro!
How are you with your project? A really good knock system (with windowing) is one of the leaks of the ms system...
Enrico
Opel/Vauxhall Corsa GSi MS2
Subaru v4 EJ20 MS3
KDW
MS/Extra Newbie
Posts: 32
Joined: Thu Jul 10, 2008 11:54 pm

Re: Knock Detection with variable sensitivity over RPM

Post by KDW »

Ay updates on this? Is the circuit provided above tested? :D
dfabro
MS/Extra Newbie
Posts: 15
Joined: Sat Apr 19, 2008 5:05 pm
Location: Brazil

Re: Knock Detection with variable sensitivity over RPM

Post by dfabro »

Hello to all
Thanks for asking
The electronic circuit is ready, tested and approved. It worked perfectly in all of chassis dynamometer tests, with several engines, including a forged piston engine.
Sorry to delay, rs
But everything is right now.

Videos by a friend:
http://br.youtube.com/watch?v=H5SqJ9vk-aM
http://br.youtube.com/watch?v=utd1DkndCIs
http://br.youtube.com/watch?v=y3UNT2dHLQg
http://br.youtube.com/watch?v=oBAzNeGJUDY
http://br.youtube.com/watch?v=I2QAR90RVk8

The PCB with components:
http://cid-86e44207bc6c1984.skydrive.li ... _KNOCK.JPG
I am available to help them to build the circuit, or sending a PCB ready.
KDW
MS/Extra Newbie
Posts: 32
Joined: Thu Jul 10, 2008 11:54 pm

Re: Knock Detection with variable sensitivity over RPM

Post by KDW »

Great! :D So the PCB is already available? How much for the PCB?
Any automotive knock sensors will do for the circuit? I have a Miata Knock sensor
DR_EVIL999
MS/Extra Newbie
Posts: 19
Joined: Thu Jun 12, 2008 1:42 pm

Re: Knock Detection with variable sensitivity over RPM

Post by DR_EVIL999 »

Is there a componant list?
richyvrlimited
Master MS/Extra'er
Posts: 568
Joined: Mon Jun 26, 2006 1:03 pm
Location: Warrington, NorthWest England

Re: Knock Detection with variable sensitivity over RPM

Post by richyvrlimited »

Just to add that I'm 100% interested in this and want a PCB/parts list too :)
2003 MX5. Coldside MP62
-
MS3, RTC, & Knock board, Release 1.2. LC-1 Wideband.
DR_EVIL999
MS/Extra Newbie
Posts: 19
Joined: Thu Jun 12, 2008 1:42 pm

Re: Knock Detection with variable sensitivity over RPM

Post by DR_EVIL999 »

I still really want one of these, Whats going on???
vinister
Experienced MS/Extra'er
Posts: 159
Joined: Sat Jul 19, 2008 11:04 pm

Re: Knock Detection with variable sensitivity over RPM

Post by vinister »

works on my breadboard ;) haven't tested in-car yet.

I'll take 2
Trev16v
Master MS/Extra'er
Posts: 420
Joined: Tue Apr 08, 2008 1:30 pm
Location: Oxfordshire, UK

Re: Knock Detection with variable sensitivity over RPM

Post by Trev16v »

Many thanks for posting this design, along with the very detailed information.

I'm just building the circuit right now to put into my MS1 next to my own Bosch ISV current drive board :D
Golf MK2 G60 Syncro with 16VG60 on MSNS-E

MSDroid :: MegaSquirt tuning on Android

Image
boost junkie
Master MS/Extra'er
Posts: 802
Joined: Tue May 11, 2004 8:58 am
Location: dallas texas usa
Contact:

Re: Knock Detection with variable sensitivity over RPM

Post by boost junkie »

Does the 0-5v signal work with the MS1 or MS2 code for knock retard?
poliux
MS/Extra Newbie
Posts: 4
Joined: Fri Sep 05, 2008 2:16 am

Re: Knock Detection with variable sensitivity over RPM

Post by poliux »

Hello dfabro :D i've built your knock cirquit, tach signal is from x2 (from megasquirt I board v2.2), knock sensor is Hitachi from audi 200 With KG engine.V ref is about 1,25V at 1500 rpm, at 7000 is about 4 volts. (using stim) .both timer cirquits are working, but OP amp isnt :( , i cant get any response i.e no leds are lit when knock sensor is hit with a metal stick. I would like to ask, are there any other ways to test OP amp (lm324) stage 1,2,3? stage 4(led control) is working :shock: ....
By the way, the pcb, you'we posted, is differnt from the knock cirquit in the knock .zip file, so maybe i'we missed something important :cry:

Thanx

Poliux, Lithuania
dfabro
MS/Extra Newbie
Posts: 15
Joined: Sat Apr 19, 2008 5:05 pm
Location: Brazil

Re: Knock Detection with variable sensitivity over RPM

Post by dfabro »

poliux wrote:Hello dfabro :D i've built your knock cirquit, tach signal is from x2 (from megasquirt I board v2.2), knock sensor is Hitachi from audi 200 With KG engine.V ref is about 1,25V at 1500 rpm, at 7000 is about 4 volts. (using stim) .both timer cirquits are working, but OP amp isnt :( , i cant get any response i.e no leds are lit when knock sensor is hit with a metal stick. I would like to ask, are there any other ways to test OP amp (lm324) stage 1,2,3? stage 4(led control) is working :shock: ....
By the way, the pcb, you'we posted, is differnt from the knock cirquit in the knock .zip file, so maybe i'we missed something important :cry:

Thanx

Poliux, Lithuania
Hi, sorry to delay.
The tach signal should be a square wave, can not be the output of ignition.
You could hit, the Vref is correct, there is that
The active filters did not work very well in some cases, So I did just a passive filter.
New circuit is attached.

You can test that will work, I assure you.
Locked