| Author | Topic |
|---|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 04:50
|
 |
Having read the following:
http://www.eng-tips.com/gviewthread.cfm/lev2/6/lev 3/35/pid/71/qid/54276
and also having a pretty good idea about the problems with generic plug-n-play knock detection systems, im a little concerned about jumping in and building something like the follwing:
my main concern (obviously) being how well tuned such a device would be for the specific harmonics of my particular engine (7M 83.5mm bore)
that said, i dont see many alternatives out there that would be any better (MSD included).
Does anyone have any thoughts on the application of the above design, or are there any resources available to help me build something better??
Or, am i just stuck with an imperfect solution, which ill just have to accept the shortcomings of?
cheers
ed
|
|
|
I supported Toymods
Banned User
Location:
Brisbane
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 05:41
|
|
I think that design sux  All engines are different and make knock sensors produce completely different signals. I looked into that same circuit a few months back, I also looked into making my own using a micro but in the end it just wasn't going to be reliable.
Its just not a simple thing to interpret the knock sensor signals. If it were, all aftermarket ECU's would.
But then what would I know? Not like I work with electronics or anything 
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 05:45
|
|
would it 'kind of' work??
im after something to switch on the water injection system, and obvioulsy knock would be the ideal trigger.
cheers
ed
|
|
|
Registered:
June 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 05:46
|
|
Have seen this diagram before and have done extensive searchin on the topic. I think this is the best you will find. There may be other of similar design.
If you are using it for tunning purposes which is the only use for it really. It would be good to have the head phones as discribed by auto speed also as it can filter on frequency.
I haven't built it because I have no use for it yet, but for $10 worth of components its worth trying. Link have a system but again without going into the specific harmonics associated with a specific engine you are attempting the generic solution. I have also heard of people using the knock module found on V6 commodores again these are design for a specific application. Go the $10 route. And some headphones with a micrphone and frequency filter.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 05:52
|
|
| roger wrote on Tue, 22 July 2003 15:46 |
I have also heard of people using the knock module found on V6 commodores again these are design for a specific application
|
any more information on the v6 module? which model? is it a separate box, or integrated into the ECU?
for tuning, yes, i will be using the mic/filter/amp/headphone thing as described on autospeed. this unit, however, i want to use as a trigger for a methanol/h20 injection system.
realistically i guess, it doesnt have to be 'accurate' in terms of absolute knock detection, but it must offer some kind of threshold detection in order to switch on the injection system in a timely manner...
cheers
ed
[Updated on: Tue, 22 July 2003 05:53]
|
|
|
Location:
Kita-Ku, Sapporo, Japan
Registered:
January 2003
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 05:58
|
|
there are a few designs around on the net (and simple jaycar type circuits), but they all have the same problem (as do OEM ones) in that the definitions of the 'knock signature' is often completely different in different motors, and depends on engine size design, and even the geometry of parts the signal reverberates thru to get to the sensor.
this signal will also change in both frequency and pattern with rpm.
the second problem (and the other main issue) is the filtering of the noise that 'is meant to be there' such as valve noise, piston slap etc.
to mimimise general noise, some OEM circuits only sample the knock sensors input signal in a certain time window (when knock is most likely to be). after that, it is a matter of creating a bandpass filter that works for the knock signals you want to detect.
if you want it to respond to the specific noises in your motor, then about the only way, short of trial and error, is to hook up an oscilloscope to a knock sensor and record both the normal engine noise at different RPM and different throttle, and then do the same again with a hint of knock, and then compare the two traces to see what the knock signal is comprised of. when you know what it's frequency is, then you can pretty closely design a bandpass filter for it, and then maybe you can just adjust the gain and detection limit for that freq range.
it's very do-able, but you have to know what frequency and amplitude the knock signals are first.
(yup i realise that this is pretty much just repeating that thread, but this is what i've come to the conclusion ar the probs)
i have a few knock sensors (the type that you put a bolt thru) and will be doing some of this testing when i put my MS on (when i finish my thesis)
fwiw, the soon to be released 'ultramegasquirt' will have knock sensing with software adjustability, and also 'ion sensing' stuff.... so finally it will be an overall engine package. i haven't been following it too close, but maybe do an archive search thru the yahoo Megasquirt list for knock sensing and ion sensing
Cya, Stewart
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 06:10
|
|
stewart,
i agree with every single thing youve just mentioned.
but its such a pity i simply cant be bothered to define the natural harmonics and knock characteristics of the engine 
if i were using the detector for something 'critical' yeah, id go to the trouble, but it will only function as a fancy switch in the current setup. if push comes to shove, id be inclined to drop the knock as a trigger all together, and just use absolute throttle position...
is there ANY cedit to that circuit posted above? is there anyway of modifying the detection characteristics of that circuit?
PS i know extremely little about circuit design etc. i know what stuff is, what it does, and how to put it together, but have no idea how designs come together, or how components 'really' interact...
cheers
ed
|
|
|
Location:
sydney
Registered:
March 2003
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
|
I supported Toymods
Banned User
Location:
Brisbane
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 06:39
|
|
Ed, that circuit will work fine for what you want to do. But it will also trigger the water pump when ever you sudden throttle movements(when backing off, accelerating hard and the like). It will also trigger when changing gears.
A true knock indicator will be linked to the throttle position, monitor revs, manifold var and pressure.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 06:52
|
|
Shane, can you comment on the following...
from BOZZ on PF's ::
| Quote: |
That knock sensor offers no adjustability, only offers output should it pick up a signal on the input which is greater than 0.7v (1N914 diode forward bias voltage). The rest of the circuit is to keep the LED/buzzer triggered for a time period so you know knocking has occured.
|
and my reply in PF's ::
| Quote: |
interesting...
so, to clarify, ANY input signal, regardless of frequency, if it produces an aplitude over 0.7v, will trigger the output of that circuit (with prologation of the output for sensible LED/buzzer triggering).
so it may NOT be specific for the harmonics and knock characteristics of my engine, but it will signal the occurance of 'big' ie >0.7v detected events...
(just say) i want to use this circuit (or similar) as a triggering device for a h20/methanol injection system. im not after accurate knock detection per se, rather, im after something that will trigger the injectors at a reasonable threshold...
to my mind, assuming the 0.7v threshold is 'appropriate', this circuit could provide such a threshold trigger...
what could i do to modify the circuit in order to perhaps crudely 'tune' it more specifically to my engine ie, raise of lower the 0.7v threshold etc. id image id just change that diode..
|
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 06:58
|
|
and actually...
if there is some delay in the switching off of the output in response to an input spike over 0.7v, is there some way of increasing that delay, ie making the LED stay on for like 3-5 sconds after knock has occured??
id imagine its just a case of swapping in a bigger capacitor or something? or is it the IC555 timer ??
thing is, i was going to use a relay with a capacitor to switch on the injector, and keep it open for 3-5 secs, but if this circuit can perform an adequate delay function, thatd be cool 
cheers
ed
|
|
|
I supported Toymods
Banned User
Location:
Brisbane
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 07:01
|
|
I agree Sorry I dont understand. What do you want me to comment on exactly?
I dont want to get into this too much right now because i'm waiting for someone to arrive here.
But a knock sensor is very basically a pezio electric microphone. The sensitivity can be adjusted in the circut but all you would be doing is changing the sensitivity across everything.
I dot know really what you want to do, but would setting the injection to switch at a predetermined RPM be any use?
|
|
|
I supported Toymods
Banned User
Location:
Brisbane
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 07:03
|
|
| ed_ma61 wrote on Tue, 22 July 2003 16:58 |
is there some way of increasing that delay, ie making the LED stay on for like 3-5 sconds after knock has occured??
|
This is really easy to do with the above circuit, you can make it 3-5 minutes if you like.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 07:08
|
|
| Cool1 wrote on Tue, 22 July 2003 17:03 |
This is really easy to do with the above circuit, you can make it 3-5 minutes if you like.
|
ok, thats a good thing.
secondly, is what bozz said correct re:
| Quote: |
only offers output should it pick up a signal on the input which is greater than 0.7v (1N914 diode forward bias voltage)
|
ie, ANY signal over 0.7v will trigger this circuit ??
|
|
|
Location:
Kita-Ku, Sapporo, Japan
Registered:
January 2003
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 08:10
|
|
you can easily make high and low pass filters using resistor and capacitors
http://www.phys.unsw.edu.au/~jw/RCfilters.html
try various combinations on the input signal.
if the signal is too low you could put a small gain (maybe x10) on the input to bump it up before filtering.
1N914 info
http://www.fairchildsemi.com/ds/1N/1N914.pdf
isn't the 0.01uf cap providing filtering after the diode?
Cya, Stewart
|
|
|
I supported Toymods
Banned User
Location:
Brisbane
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 09:00
|
|
ed_ma61 wrote on Tue, 22 July 2003 17:08secondly, is what bozz said correct re:
[quote title=Quote: |
only offers output should it pick up a signal on the input which is greater than 0.7v (1N914 diode forward bias voltage)
|
ie, ANY signal over 0.7v will trigger this circuit ??
[/quote]
Yes thats correct. The 0.7v is the spec of the diode. You could bring this down to as low as 0.5v using a germanium (spelling?) diode, or raise it by using another "cheaper" diode.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 10:01
|
|
| oldcorollas wrote on Tue, 22 July 2003 18:10 |
you can easily make high and low pass filters using resistor and capacitors
|
ive got a funny feeling that this wont be necesary:
i get the impression that individual knock sensors themselves are 'tuned' to a specific frequency, for particular use with a given engine. as such, its conceivable that the sensor itself will only output a high amplitude signal when stimulated within its specific restricted freq range (corresponding to the 'knock zone')...
thus, a circuit such as the one above might be adequate in simply being trigered whenever one of these high amplitude signals makes it through from the sensor... negating the need for an interim filter
damn, i wish i knew the first thing about electronics...
cheers
ed
|
|
|
Location:
sydney
Registered:
March 2003
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 12:14
|
|
i ran into this web site few yrs ago
they have some interesting devices u can make
http://minimopar.net/perf/eepc.html
it shows u how to make devices turn relays on at certian voltages so u can use ma sensors and shgit like that
hoipe it helps
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 12:37
|
|
hey thanks. thats cool
cheers sideshow
ed
|
|
|
Location:
Melbourne
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 12:41
|
|
Keep it basic Ed! switch on the H2O >2000RPM >90% throttle
OR
I was even thinking of a 5m ecu to run the water injection! just hook up a MAP sensor in place of the AFM and then you can fire upto 6 water injectors based on load/air temp/coolent temp and engine speed with a pot on the map and air temp sensor inputs for fine tuning.. just an idea and you allredy have the 5m ecu in there!
If only you could trace the knock signal in the 7m ecu!!!
Who do we know with the smarts todo that hmmmmmmm
Allan
|
|
|
Location:
sydney
Registered:
March 2003
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 12:43
|
|
no probs
the tps controller is prob what u r lookin for
when i look at the drawings theyre bit to complicated for me
i can do some tiny electronics and i have a book which might help
i can do most car electronics but when it comes to circuits like those its abit out of my league
i would have to sit down and learn the basic so i can understand it first
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 12:55
|
|
| Allan wrote on Tue, 22 July 2003 22:41 |
Keep it basic Ed! switch on the H2O >2000RPM >90% throttle
|
allan - oh common! when do you want H20 injection? when the thing is going to start detonating. the best way to tell when the thing is going to start detonating?: with a detonation detector - then just let the injectors overrun for 5 seconds before resampling the engine state, and firing the injectors again...
ill just set a low threshold on the detector input, and the spray will fire before the engine starts making too many odd noises. sounds pretty simple to me
i have a funny feeling that ill run out of water quickly if the thing just ran @ 2000rpm and 75% throttle or so  i need some 'real' indicator for turning the things on, and konck seems like the perfect trigger
and the circuit is simple enough.
as for having another ECU run the injectors, interesting idea - but remeber, most ecu's wont work unless ALL sensors are working properly.. and thatll be a bitch to hook up
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 13:02
|
|
actually -
you know what i just realised 
guess who has a ACIS valve which is switched on by the STOCK 7m ecu @ 4000rpm and ~75% throttle
might just get that to switch the injectors too
ill still get the knock detector going though, that way i can see if its being effective
30deg BTDC here we come
cheers
ed
|
|
|
Location:
Melbourne
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 13:12
|
|
|
ACIS?
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 14:26
|
|
ahhh CRAP
the VSV is switched off at >60% TPS and 4200rpm...
shitcan that idea
im becomming less certain that the knock sensor is a desirable primary trigger. sure, it'd be good to have running as a secondary override trigger just in case, but i think either a MAP based or an RPM + TPS based system would be better...
cheers
ed
|
|
|
I supported Toymods
Location:
Brisbane, QLD
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Tue, 22 July 2003 17:22
|
|
Whoa! this is a bit of a long thread...
Ed,
If you want an ideal knock sensor solution, what you would want is a DSP (digital signal processor) hooked up to your knock sensor and an interesting algorithm (for example, a FFT [fast fourier transform] that would convert the discrete knock sensor input samples into it's frequency spectrum and look for strong signals out of the engines characteristic frequency band at any given RPM). Look at Motorola web site and some of their automotive industry application notes for DSP products. However, this is a costly and time consuming approach.
Contrary to popular belief, most knock sensors are just simple piezo electric pickups. The high pass filters are implemented in the ECU. And most ECU designs tend to have a simple high pass (or low pass) filter followed by an integrator (AC to DC converter if you must) and then a comparator (voltage difference sensitive trigger of sorts). The simplest setup is to find a cutoff frequency for the filter (say the frequency, for example at which 95% [variable number] of the power concentration in the frequency spectrum is divided at a high [close to redline] RPM, yet lower than the knock frequencies at lower RPMs). Then the output of the filter (both the low and high component) are integrated seperately, and then run through a comparator. The comparator is triggered whenever the power concentration of the frequency spectrum is significantly different from 95:5 (as per our example), biased to the high frequency side (i.e. potential knock).
Most simple circuits you find on the net are of this nature (with different cutoff frequencies).
IMO, the engine RPM/AF ratio/Intake air temp and AFM (or MAP/TPS) signal presets can be used to create a "potential knock zone" trigger succesfully, by marking conditions that are most likely to cause knock. The knock sensor trigger can be a secondary "shit, it's all bad" trigger.
Sorry if I repeated what's been said already... it's too early in the morning and the thread is just too long...
Cheers
|
|
|
Location:
Brisbane
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 00:14
|
|
I'm another fan of the "keep it simple" philosophy. What exactly is wrong with running water injection constantly under high-load conditions? Sure you'll go through a fair bit of water, but so what? Use a sufficiently large reservoir and you'll be right. Remember the ability to run more advance isn't the only performance advantage of water injection, IF your engine is tuned for it.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 00:17
|
|
| humble wrote on Wed, 23 July 2003 03:22 |
IMO, the engine RPM/AF ratio/Intake air temp and AFM (or MAP/TPS) signal presets can be used to create a "potential knock zone" trigger succesfully, by marking conditions that are most likely to cause knock. The knock sensor trigger can be a secondary "shit, it's all bad" trigger.
|
humble,
thats basically what i went to bed thinking...
please, feel free to make a black box that processes RPM/AF ratio/Intake air temp and AFM (or MAP/TPS) signals 
alternatively, a simpler solution might be to have it based on a threshold of RPM and throttle position. ie over 3000rpm and 70% throttle (thanks allan) with the kncok sensor as a secondary override for the, as you say "shit, it's all bad" conditions...
[Updated on: Wed, 23 July 2003 00:19]
|
|
|
Location:
Melbourne
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 02:19
|
|
Ok what about the ammount of water from what i have read about the research done in the late 1800's is the mixtures should increase before water is injected and then as water injection begins fuel should be leaned to end in a final ratio of 1:1 fuel and water???? its been awhile sence i have looked into this feel free to correct me! (like you need my permission to shoot me down in flames)
Whos willing to put there engine on a dyno while we get the calibrated garden hose out??
Allan
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 03:02
|
|
me thinks a megasquirt could come in use here (thanks oldcorollas)...
that way i can pulse the methanol/h20 in, with increasing duty cycles according to MAP and rpm etc... and we can derive a whole 'fuel map' for its delivery on the dyno 
as for decreasing fuel... hmmm doubt it - getting to the point of a whole aftermarket tuning system, which as we all know, is WAY outta my league.
cheers
ed
|
|
|
Registered:
June 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 03:11
|
|
Bloody electronics!!!
Gives me a headache, not to mention eye strain after reading all that!!
Just a thought, I'm guessing you want to run the 7M on pump gas, but why not run higher octane fuel?? (100+ & 98 mix??)
Or mix additives with your fuel to give the engine constant running conditions.
http://www.turbofast.com.au/racefuel13.html
Another 2c.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 03:20
|
|
| E30-323ti wrote on Wed, 23 July 2003 13:11 |
Just a thought, I'm guessing you want to run the 7M on pump gas, but why not run higher octane fuel?? (100+ & 98 mix??)
Or mix additives with your fuel to give the engine constant running conditions.
|
yeah, itll be run on 98RON BP or Mobil as standard
i just want to tweak it even more
Looking at the megasquirt ecu - things are suddenly getting terribly complicated for a simple 2 injector water spray!! i might as well run the whole engine off the thing for the effort in setting it up 
'crude' is a word that comes to mind on this one. keep it simple...
cheers
ed
|
|
|
Registered:
June 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 03:34
|
|
Ed,
Were/are you planning to run this hipo monster on the stock ECU anyway?? Thus the issues with detonation??
I've tried to look at the megasquirt stuff before, just to painful to figure it out.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 23 July 2003 03:40
|
|
| E30-323ti wrote on Wed, 23 July 2003 13:34 |
Ed,
Were/are you planning to run this hipo monster on the stock ECU anyway?? Thus the issues with detonation??
|
yep
theres nothing ive done that the stock ecu shouldnt be able to handle with a bit of tweaking. detonation will most likely be an issue for me though, considering the compression and timing advance id like to run (+25deg BTDC).
Ive always planned on using some kind of h20/methanol injection system to allow me to achieve 11:1 comp/r with big timing advance, its just come time for me to figure out exactly how im going to do it.
|
|
|
Location:
Lost in the K hole
Registered:
May 2002
|
|
Re: Knock Sensor Circuit Design ...
|
Wed, 30 July 2003 11:48
|
|
UPDATE
ok
i built the circuit, and the sucker works (thanks shane! )
hehehe
BUT at the moment, it takes a HEFTY whack with a ballpein hammer on the deck of the block (on the cyl above where the sensor is mounted) to get the thing to trigger.
i have no idea how powerful a shock detonation puts through the block, but i suspect at the moment the circuit isnt sensitive enough, so im going to swap the 1N914 diode (0.571v forward bias) for a germanium diode (lower forward bias volatge) and hopefully capture things at a lower threshold (0.45v or so?)
i might put the two diodes on a rotary switch so i flip between the two as i figure how this thing works.
anyway, its looking 'not bad' as a basic indicator of what the sensor is hearing...
cheers
ed
|
|
|
Location:
Canberra
Registered:
July 2003
|
|
|
