Knock Sensor Mounting
#22
Local welders (even small ones) charge a minimum $20 fee to even pick up their welder. I won't ask somebody to weld it for free as I don't believe in taking free work unless I'm paying them for something else. This was in the past of course before I learned to weld. I figured others are in the same situation I used to be before I learned to weld, where they neither knew how to nor knows somebody that can. Either way, that kind of looks like *** when for $10 more, I could get a proper stud.
#24
Local welders (even small ones) charge a minimum $20 fee to even pick up their welder. I won't ask somebody to weld it for free as I don't believe in taking free work unless I'm paying them for something else. This was in the past of course before I learned to weld. I figured others are in the same situation I used to be before I learned to weld, where they neither knew how to nor knows somebody that can. Either way, that kind of looks like *** when for $10 more, I could get a proper stud.
#26
You want the knock sensor touching the block not on a huge spacer, that's going to change reaction time and frequency. I want the best protection so a custom $25 stud is worth it to me. Make some in stainless and list 25 qty on ebay. Other car models require the same stud we need.
#34
- Low change in sensitivity over the lifetime of the sensor (datasheet max -17%)
- low change in sensitivity with respect to temperature (mV/g/degC)
- ruggedized sensor with high temp/vibration limits
- most importantly, highly linear response and large dynamic range (KS4-R I just bought is 3 to 25 kHz whereas many older sensors only measure ~6 kHz. Makes closed loop knock tuning, especially if looking at a large freq. range, much easier as your threshold can be sufficiently tight throughout the measurement window of the ECU so you're not over-reacting or letting actual knock by. Ex. from newest gen KS4-R2)
So the trade off here is essentially that you need more expensive (through the lens of mid 90s-early 00s computing tech that went into the NB) broadband instrument amplifiers to bring the mV-scale signal up to the level of the ADC and much more computationally expensive digital signal processing to window the input signal with respect to cam/crank angle (i.e. what cylinder/when) and FFT or otherwise analyze/filter it's frequency components, but in turn the amount of information you have to work with is hugely more significant ( ~ 25 mV/g from 3-25 kHz vs. FWHM of ~ 1 V/g from ~6.75-7.25 kHz). I believe this Delphi sensor (originally on a Saturn?) is similar if not the same as the stock NB sensor but I could be wrong.
TL;DR: if you have an ECU with knock signal amplification and more advanced signal processing than ~MS2 era crap, buy a broadband knock sensor (bosch motorsports ones just also came on some cars OEM so low cost and are rated for the application).
I wasn't super keen on getting the sensor connector to fit down there and picked up a used (salvage vehicle [Quattraporte V6]) KS4-R (flying lead version) off eBay for ~$30 recently. No reason to go old school if the good stuff is the same price ¯\_(ツ)_/¯.
Last edited by Spaceman Spiff; 08-19-2021 at 02:36 AM.
#37
https://frsport.com/products/shift-k...tainless-steel
I found that, might be able to cut it down, glue it in place. Sit the sensor directly against the block?
I found that, might be able to cut it down, glue it in place. Sit the sensor directly against the block?
#38
From what reading I did a while ago, you could drill or ream your sensor out to 10mm, the hpacademy guys hadn't had any issues doing that. Another solution was a helicoil in 8x 1.25 apparently snugs in.
but for an easy button, I'll still take one of these when they are made cos I am inestimably lazy and am down for an idiotproof solution.
but for an easy button, I'll still take one of these when they are made cos I am inestimably lazy and am down for an idiotproof solution.
#40
Do you have the ability to measure the frequency off the sensor in realtime? If not, I'd stick with the stock resonant sensor, else finding the knock frequency is going to be a fun time. Look at the response of the plots posted already, the resonant type gives 66 times more signal at ~6K than the wideband. If 6K is the knock frequency, that's the one you want.
Haltech will let me log all frequencies off the sensor in realtime via a scope, while adding a button to temporally advance into knock, hopefully. With a wideband knock sensor I'd be able to set it's filter frequency higher or lower than 6K if needed. If it were 6K, I'd probably switch back to the resonant type, because it naturally filters noise a hell of a lot better.
Haltech will let me log all frequencies off the sensor in realtime via a scope, while adding a button to temporally advance into knock, hopefully. With a wideband knock sensor I'd be able to set it's filter frequency higher or lower than 6K if needed. If it were 6K, I'd probably switch back to the resonant type, because it naturally filters noise a hell of a lot better.