Knock sensors resonance / frequencies
Discussion
Far as I can see knock sensors are just microphones, so are they mechanically tuned to the specific knock frequencies? I don't see any reason to do this as you would simply tune out the knock signal electronically in the ECU given that the sensors are generically more sensitive to higher knock audio frequencies anyway. The reason I ask is I want to test for knock on a 3.9 Rover v8, and I have a sensor for a 2 ltr vauxhall laying around- in both cases its near 500cc per cylinder, so the frequencies should be similar.
Some sensors are narrowband and would need to be off a similar/same bore diameter car
Others are wideband, and the ecu does any filtering
http://phormula.co.uk/KnockCalculator.aspx
Others are wideband, and the ecu does any filtering
http://phormula.co.uk/KnockCalculator.aspx
In the "early days" the power of EMS systems was limited (in terms of computational cycles and the width of their data buses) As such, they simply couldn't cope with doing lots of high speed sampling and data processing on the knock signal and also running the engine. To negate that, several specific AISC (Application Specific Integrated Circuit) were developed to move the knock signal processing outside of the main ecu's processor, so all it had to do is get the knock data straight from that ASIC. Unfortunately, this added cost, and OEMs HATE cost ;-) So, some volume manufacturers moved towards tuned knock sensors, that have a narrow frequency response gain, meaning much less processing is required by the ECU.
Unfortunately, these needed careful calibration, and any change in engine condition (like wearing / increased clearances on the cam/lifters for example) could result in the system frequency changing enough for the knock system to either false flag knock, or worse, simply miss it completely.
Fast forwards 10 years,a nd we have processing power coming out of our ears, so most modern ECUs have gone back to wide frequency response sensors, and do the entire signal processing in their DSP (Digital signal Processor). In effect, they use a fast Analogue To Digital Convertor (ADC) to sample the raw knock sensor voltage at over twice the maximum frequency of interest (as demanded by our good friend Mr Nyquist) and process the whole signal in the digital domain (rather than in the analogue domain)
So, long story short, most knock sensors you will find will be "wide frequency response" sensors, with a fairly flat v/g response from around 3Khz to say 18Khz.
Unfortunately, these needed careful calibration, and any change in engine condition (like wearing / increased clearances on the cam/lifters for example) could result in the system frequency changing enough for the knock system to either false flag knock, or worse, simply miss it completely.
Fast forwards 10 years,a nd we have processing power coming out of our ears, so most modern ECUs have gone back to wide frequency response sensors, and do the entire signal processing in their DSP (Digital signal Processor). In effect, they use a fast Analogue To Digital Convertor (ADC) to sample the raw knock sensor voltage at over twice the maximum frequency of interest (as demanded by our good friend Mr Nyquist) and process the whole signal in the digital domain (rather than in the analogue domain)
So, long story short, most knock sensors you will find will be "wide frequency response" sensors, with a fairly flat v/g response from around 3Khz to say 18Khz.
Typical Bosch sensor:
You can log knock sensors into the sound card of your laptop and FFT it to see the frequency spectrum with and without knocking events (44Khz sampling rate is enough to FFT up to around 20Khz)
(you may need to buffer the sensor with an op amp, and/or reduce it's output voltage (using a resistive divider) to match the sound cards ~ +-1v input range)
You can log knock sensors into the sound card of your laptop and FFT it to see the frequency spectrum with and without knocking events (44Khz sampling rate is enough to FFT up to around 20Khz)
(you may need to buffer the sensor with an op amp, and/or reduce it's output voltage (using a resistive divider) to match the sound cards ~ +-1v input range)
Thanks Guys, Ive bought one of the Knockteck detectors from Extra EFI, that appears to have been tested on the RV8, and they recommend the sensors that go with the later 4 to 5 ltr block used on the Sagem and Thor ECU systems. , but I dont know if these are "narrow" band- Id suspect not as one sensor seem to cover a range of engine capacities. Mines a 92 3.9 pre knock detection engine. The Vauxhall sensor I have certainly is not tuned as if you listen to the audio you can clearly hear the engines lower frequency mechanical components. Ill strap it on and proceed with caution, thanks.
If you are bolting the knock sensor to the engine and listening with your ears with some form of amplifier, unless the amp has lots of dedicated filtering you are always going to hear lots of other engine noises.
Your ears/brain will be the best filter in these cases
If using your ears, the basic mechanical stethoscope type devices are almost the best, although overall sounds tend to be quieter, but there isnt as much background noise with them, so knock when you know what to listen for is fairly easy to detect at a very early onset.
Your ears/brain will be the best filter in these cases
If using your ears, the basic mechanical stethoscope type devices are almost the best, although overall sounds tend to be quieter, but there isnt as much background noise with them, so knock when you know what to listen for is fairly easy to detect at a very early onset.
blitzracing said:
Thanks Guys, Ive bought one of the Knockteck detectors from Extra EFI
As those devices are not "windowed" they have a poor signal to noise ratio, but will pick up gross knock, ie the sort of knock you can hear with your ears over the mechanical noise of the engine. Without speed and load dependant thresholds they will almost certainly miss the extremely damaging knocking that occurs at high rpm and high load. My recommendation is to use the device to find where gross low speed high load knocking occurs (say WOT @3000rpm) and not rely on it for setting higher rpm ignition angle. of course, unless you are running an very high CR engine then your engine will almost certainly be MBT at peak power rpm anyway! Slightly off topic, so apologies.........I have an Evo 8 and fitted a VVT head ( and turbo and ecu ) from an Evo 9 about 3 years ago.......the previously quiet aftermarket knock monitor I've had in the car for years instantly began to go nuts on high rev gearchanges, even though the engine itself has been shown to have no knock ( separate sensor connected to headphones worn when mapping on the road ). Is it likely that the way the VVT works is causing the knocklink to give false readings? Does the way it operates generate significantly more noise in the head than on a car without it?
VVT is a bit of mare for knock systems, because the point where the valvetrain noise occurs is not fixed to a specific crank angle (which is kinda the point of VVT!)
What will be happening is that at high rpm when you lift off, the VVT will be swinging the cams to one end, and the valve noise will be creeping into the normal knock window (which on a non VVT engine never has valve events in it). Of course, that means the system will now be responding to that false knock, and as retarding the ignition doesn't remove the "knock" you'll be getting very retarded ignition......
What will be happening is that at high rpm when you lift off, the VVT will be swinging the cams to one end, and the valve noise will be creeping into the normal knock window (which on a non VVT engine never has valve events in it). Of course, that means the system will now be responding to that false knock, and as retarding the ignition doesn't remove the "knock" you'll be getting very retarded ignition......
Thanks for the reply.......I'm assuming that as I throttle off to change up, the knock sensor can hear the mechanical noise of the VVT closing, and is having a fit because of it. If that's the case, it's good and bad news I suppose. The good being that the knock sensor is working, the bad that it more or less renders it pretty redundant! It's already been backed off from maximum sensitivity ( where it was set before the mods ), to where it is now at halfway.......as it is, on maximum attack, high rev changes, all 5 lights come on.
Any way around it? The sensor itself is mounted on the inlet manifold, I'm wondering if putting it somewhere on the block, away from the VVT would be better, but don't know how much heat they can cope with. It's an aftermarket, old style knocklink, with the usual bosch sensor.
Any way around it? The sensor itself is mounted on the inlet manifold, I'm wondering if putting it somewhere on the block, away from the VVT would be better, but don't know how much heat they can cope with. It's an aftermarket, old style knocklink, with the usual bosch sensor.
I've got the impression that it isn't on occasion..........it feels like it's pulling timing sometimes on full throttle in high gears, but because it's a pretty random occurrence, it's hard to pin down quite why it's happening. It doesn't necessarily coincide with knock monitor activity. I don't understand enough about how remaps work to know if the parameters of the standard fit knock sensor are altered to take into account different ignition settings?
As it stands the car should really have another remap, it's just had an ETS intercooler and larger ic pipes which have reduced inlet temps by a remarkable amount, and a few other mild tweaks, including a cop kit, although the latter shouldn't really affect anything map wise, to be fair.
I boroscope it now and again to physically look at the top of the pistons, there never seems to have been an issue with det, so I'm more interested in the phenomenon than concerned with it. The car has used oil since the modifications were done, which I understand can make an engine more likely to be prone to det.
As it stands the car should really have another remap, it's just had an ETS intercooler and larger ic pipes which have reduced inlet temps by a remarkable amount, and a few other mild tweaks, including a cop kit, although the latter shouldn't really affect anything map wise, to be fair.
I boroscope it now and again to physically look at the top of the pistons, there never seems to have been an issue with det, so I'm more interested in the phenomenon than concerned with it. The car has used oil since the modifications were done, which I understand can make an engine more likely to be prone to det.
Edited by Heaveho on Monday 31st March 17:18
Gassing Station | Engines & Drivetrain | Top of Page | What's New | My Stuff