DIY Knock gauge
- Thread starter blueta
- Start date
Bobs rare 6
Member
- Joined
- Apr 11, 2004
I would be interested in a audiable knock detector without gauge. Any vendors or ideas? Thanks, Bob.
BOOSTEDGNs
Member
- Joined
- Oct 6, 2004
http://forums.corvetteforum.com/showthread.php?t=528523&forum_id=82&arch=1
I found this on corvetteforum.com dont know if of any use
Alan
I found this on corvetteforum.com dont know if of any use
Alan
- Joined
- Aug 1, 2003
www.casperselectronics.com sells both
the 2 1/16" LED gauge is hard to beat for instant in your face knock detection
the audible mounts under the dash and takes a few degrees to activate but its hard to miss as an indicator to lift off the pedal
https://www.casperselectronics.com/Merchant2/merchant.mvc?Screen=CTGY&Category_Code=8687GN_Perf
the 2 1/16" LED gauge is hard to beat for instant in your face knock detection
the audible mounts under the dash and takes a few degrees to activate but its hard to miss as an indicator to lift off the pedal
https://www.casperselectronics.com/Merchant2/merchant.mvc?Screen=CTGY&Category_Code=8687GN_Perf
hmm...after seeing that one on the corvette forum it makes me wonder why the one on gnttype has to have a circuit board fabbed up. I'm pretty sure my GTP doesn't use an ESC module so maybe i could just get one from a yard (if all gm ones are the same then that'll be a plus), and hook up power to it, sensor signal, and stick that resistor/LED on it and I got where I'm at right now without any real work 
I wish I could figure out how to make it read like the caspers though. Cause with only 1 LED on you have no idea if it's severe or not. You just know that there is a something going on.
I wish I could figure out how to make it read like the caspers though. Cause with only 1 LED on you have no idea if it's severe or not. You just know that there is a something going on.
salvageV6
Daily Driver
- Joined
- May 25, 2001
You can count the positive to negative transitions on the EST line at the ECM the yellow/black line that most KR scantools go to.
Use most any type logic counter chip to add up those transitions and output the count to a transistor switch or logic that can run a relay or relays to give you lights and buzzers at various counts.
Correlate the count to KR via direct scan or a scanmaster and set it up any way you want to.
Reset via a simple timer chip whenever you want or via a pushbutton.
It's actually simple 70's logic and electronics not rocket science.
Use most any type logic counter chip to add up those transitions and output the count to a transistor switch or logic that can run a relay or relays to give you lights and buzzers at various counts.
Correlate the count to KR via direct scan or a scanmaster and set it up any way you want to.
Reset via a simple timer chip whenever you want or via a pushbutton.
It's actually simple 70's logic and electronics not rocket science.
Pos to neg, this is because the knock sensor goes negative when knock is detected right? I'm taking it as the yellow/black one to tap into is the one from the ECS that I will have to add. I sent you an email also to see if you could draw up a quick schematic or something cause I don't have any real electronics background, but I can read a schematic pretty well. I'm sure I'll have a lot more questions when I start to buy parts Thanks!
Thanks!
salvageV6
Daily Driver
- Joined
- May 25, 2001
The line between the EST module and the ECM goes positive to negative upon KR detection, not the sensor itself, already processed pulses via the EST module going towards the ECM to be counted and further processed.
Same point most KR detectors work off of, easy to use falling edge transitions for counts.
I don't really have any schematics drawn up but if you look at the SN74143 type chip you can see how it counts nicely and can output to a 7 segment LED or other device such as a transistor that can run a relay.
They work off of 5 volts so you need a 5 volt regulated supply which can be made for about $3 with Radio Shack stuff, I don't use the ECM's 5 volt supply for aftermarket electronics I make, easy to do with a 3 leg chip and two capacitors and a heat sink if you need to draw any current.
For the input side off the EST line I use a 914 diode for isolation banded end to the EST line that only allows the ground signal input to the counter.
With no electronics background at all it could be a bit challenging, but it really isn't complicated electronics at all.
Same point most KR detectors work off of, easy to use falling edge transitions for counts.
I don't really have any schematics drawn up but if you look at the SN74143 type chip you can see how it counts nicely and can output to a 7 segment LED or other device such as a transistor that can run a relay.
They work off of 5 volts so you need a 5 volt regulated supply which can be made for about $3 with Radio Shack stuff, I don't use the ECM's 5 volt supply for aftermarket electronics I make, easy to do with a 3 leg chip and two capacitors and a heat sink if you need to draw any current.
For the input side off the EST line I use a 914 diode for isolation banded end to the EST line that only allows the ground signal input to the counter.
With no electronics background at all it could be a bit challenging, but it really isn't complicated electronics at all.
I guess I need to do some reading up on some electronics. About all know beside basic ground and power stuff (and relays) is resistors give resistance, and diodes are like electronic check vavles. I have some 914's laying around so that's good. Another thing...call me stupid but what's a sn74143?
salvageV6 said:
So if I wanted to see if the ESC is hooked up right and functioning right using the knock LED on those corvette forums, I would install the LED between the 12v positive and the signal line the LED should light up all the time UNLESS there is knock. Cause if the line is + before knock then the LED should always be lit, and then grounded so the circuit cannot be completed and the LED goes out?
salvageV6 said:
Still don't know what the SN74143 is...I tried to search it on some electronics websites (radio shack and jameco) and came up empty handed
salvageV6
Daily Driver
- Joined
- May 25, 2001
If you hook the +12 part of the LED to power and then hook the negative end of the LED to the ESC line it shouldn't light up until there is knock.
However these pulses are small in duration and may very well not be seen by the eye.
Plus you would need to make sure that the ESC module can source 10ma of current to run the diode, most likely it can but I don't have the specs. on it.
All the KR detectors and the ECM are high impedance inputs which don't draw much current from the ESC module IE. they don't load it down much with current draw.
Nobody I know puts an LED right on the ESC line to make a detector, it would be an interesting test however since 10ma isn't really a lot of current.
Make sure the diode is rated for 12 volts, available at radio shack, or has the proper resistor in line to only draw about 10ma. of current through it, 600ohms to 1K ohms works usually for most 1-3 volt forward LED's.
The resistor could be placed in series with either leg and the supply or ESC line.
To make a light LED meter you would first need to store the counts of the ESC line and decide how many of those + to - transitions you want to light up the first LED and then the second etc?
Part of the design and the reason you need to somehow correlate the counts to real knock as seen by the engine and ECM management system.
I would assume perhaps 2-3 transitions per LED. Guessing so far.
Then we can build a counter with outputs to LED's and an automatic reset of the counter after a set period of time say 15-25 seconds or so.
Sound reasonable?
I do not know what scheme caspers uses to light their LED gauge but it works just fine in my GN and is well worth the $$$.
Here is a pic. of a two digit 00-99 counts readable counter I am using on my ESC line to try to develop such a correlation.
I can only run about 10-14 pounds of boost on the snow tires however so my car rarely knocks.
I did have to make a reset feature upon powerup which is a simple timer that resets the counts in the chips for 2 seconds when power is applied.
It gets rid of ESC to ECM noise and triggering as well as engine cranking KR.
I haven't incorporated an auto reset feature after counting starts since I want to review the counts and manually reset it with the pushbutton that's on the board.
Works fine but no schmatic as of yet....
The unit is upside down in the picture and the small LED's just read out the numbers on the 7 segment displays in BCD format.
http://a6.cpimg.com/image/CA/D4/56185546-ece4-02000180-.jpg
However these pulses are small in duration and may very well not be seen by the eye.
Plus you would need to make sure that the ESC module can source 10ma of current to run the diode, most likely it can but I don't have the specs. on it.
All the KR detectors and the ECM are high impedance inputs which don't draw much current from the ESC module IE. they don't load it down much with current draw.
Nobody I know puts an LED right on the ESC line to make a detector, it would be an interesting test however since 10ma isn't really a lot of current.
Make sure the diode is rated for 12 volts, available at radio shack, or has the proper resistor in line to only draw about 10ma. of current through it, 600ohms to 1K ohms works usually for most 1-3 volt forward LED's.
The resistor could be placed in series with either leg and the supply or ESC line.
To make a light LED meter you would first need to store the counts of the ESC line and decide how many of those + to - transitions you want to light up the first LED and then the second etc?
Part of the design and the reason you need to somehow correlate the counts to real knock as seen by the engine and ECM management system.
I would assume perhaps 2-3 transitions per LED. Guessing so far.
Then we can build a counter with outputs to LED's and an automatic reset of the counter after a set period of time say 15-25 seconds or so.
Sound reasonable?
I do not know what scheme caspers uses to light their LED gauge but it works just fine in my GN and is well worth the $$$.
Here is a pic. of a two digit 00-99 counts readable counter I am using on my ESC line to try to develop such a correlation.
I can only run about 10-14 pounds of boost on the snow tires however so my car rarely knocks.
I did have to make a reset feature upon powerup which is a simple timer that resets the counts in the chips for 2 seconds when power is applied.
It gets rid of ESC to ECM noise and triggering as well as engine cranking KR.
I haven't incorporated an auto reset feature after counting starts since I want to review the counts and manually reset it with the pushbutton that's on the board.
Works fine but no schmatic as of yet....
The unit is upside down in the picture and the small LED's just read out the numbers on the 7 segment displays in BCD format.
http://a6.cpimg.com/image/CA/D4/56185546-ece4-02000180-.jpg
salvageV6 said:
Well I just read through this quick and I don't really have time to read into it (I'm on the move right now lol), but the negative side, I would assume you mean cathode? And the ESC line we are talking about is what color? The one to the ECM (black), or the one from the knock sensor (blue)? I'll read over the rest of it tonight. Thanks!
salvageV6
Daily Driver
- Joined
- May 25, 2001
The one exiting the ESC module is the yellow with black stripe.
It goes to ground upon KR detection, not for long though, doubtful you will see anything unless you are hammering near the sensor for a long while.
Remember no scantools actually do this to show KR, and if it's still hooked to the ECM it may not work well, as a stand alone test you could try it.
Normally you stick a high impedance IC chip on that line to do your counting just like the ECM does and then process the info. via computer or fixed counting chips like most KR meters to output to your leds or beepers.
You have to know how much you want to count first, hence the counter I made to test the ESC line under various driving conditions.
It goes to ground upon KR detection, not for long though, doubtful you will see anything unless you are hammering near the sensor for a long while.
Remember no scantools actually do this to show KR, and if it's still hooked to the ECM it may not work well, as a stand alone test you could try it.
Normally you stick a high impedance IC chip on that line to do your counting just like the ECM does and then process the info. via computer or fixed counting chips like most KR meters to output to your leds or beepers.
You have to know how much you want to count first, hence the counter I made to test the ESC line under various driving conditions.
Sounds good. BTW, my wires are different colors cause I got them off of a 305TBI firebird. I matched them to what you answered in my other post about the ESC hookup though. Mine are pink/black=12 volt, black=ESC line, brown=ground, and blue=Knock sensor. I'm gonna make one like yours and see how it works. I'll get back after I build it and give it a whirl. Thanks for all the help!
EDIT: When you make up a schematic for that lemme know...cause I can't really start it till then lol. And some part #'s would be great too! :biggrin:
EDIT: When you make up a schematic for that lemme know...cause I can't really start it till then lol. And some part #'s would be great too! :biggrin:
salvageV6
Daily Driver
- Joined
- May 25, 2001
Mines too complicated for a novice to build plus the chips are older and what I had lying around here to use.
How about we design a new one from scratch?
Here's what I have so far, YOU must make the counting decision to light up the LED's in sequential order via the total count.
Every 2 pulses or 3 pulse of the EST line signalling knock? Every one, 4?
The one on gnttype.org will light up with one pulse of the line and beep too from what I can see off the schematic.
I think we can do automatic turn on reset so it stays off, no lights, when powering up and cranking the car over on startup, hence no starter induced lights, also automatic reset after say 30-45 seconds once the first led lights up should be plenty of time to view it after a run or blast.
Maybe an optional switch for 3rd gear only knock so you can test it under the largest load the car sees in case you have 2-3 shift knock or false knock in other gears etc., someone might only want it working in 3rd gear anyways, easy to leave that option off if not needed.
And perhaps a buzzer for last yellow LED and perhaps a beeper for the last green one we can decide how many counts to make them beep and what color the leds will be as well, I am thinking 10 steps (LEDS) 16 may be possible.
I will try and find some easier to find parts and schemes, you make the count decision.
And don't hold your breath on the timeline either.
I've got lots of other projects going too.
How about we design a new one from scratch?
Here's what I have so far, YOU must make the counting decision to light up the LED's in sequential order via the total count.
Every 2 pulses or 3 pulse of the EST line signalling knock? Every one, 4?
The one on gnttype.org will light up with one pulse of the line and beep too from what I can see off the schematic.
I think we can do automatic turn on reset so it stays off, no lights, when powering up and cranking the car over on startup, hence no starter induced lights, also automatic reset after say 30-45 seconds once the first led lights up should be plenty of time to view it after a run or blast.
Maybe an optional switch for 3rd gear only knock so you can test it under the largest load the car sees in case you have 2-3 shift knock or false knock in other gears etc., someone might only want it working in 3rd gear anyways, easy to leave that option off if not needed.
And perhaps a buzzer for last yellow LED and perhaps a beeper for the last green one we can decide how many counts to make them beep and what color the leds will be as well, I am thinking 10 steps (LEDS) 16 may be possible.
I will try and find some easier to find parts and schemes, you make the count decision.
And don't hold your breath on the timeline either.
I've got lots of other projects going too.
As stated, and especially using tapped signals that normally go to the ECM, use a high impedance input so nothing gets sourced improperly or causes interference with existing use. You can use a simple logic buffer or general purpose opamp in voltage follower configuration. If you need negative swing, even though you could just re-ref the signal, use a highly filtered charge pump. I see some circuits drive Zener's to limit swing... be carefull on that and be sure your not sinking current from the ESC into the input of your circuit.
If I were making one, I would make it all on a PIC cycling at 10-20MHz; should be waaay fast enough for instruction set to check signal input. With a uC you could easily build in the calibration and drivers in one unit, in the code. Buffering of the analog signal would be handled at the ADC. Most units have resolutions of 8bits for the ADCs... many 18fxxx series has res. of 10b on ADC lines. More than enough IMO. That would all be for using the knock signal directy, which is what I would do since it would be interchangeable for any application. Besides, code is always easier to change than hard parts.
If you want to use the existing ESC module the thresholding and filtering is done for you already (this is actually difficult to set w/o proper testing due to engine resonances, etc...); the calibration on how long it is held low I currently do not know.. A brute force calibration method would be to simply use a scanmaster, DS, DFI etc... to check your readout, be it numeric (seven segment(s) or just LEDs... I will recommend you just do seven segments as there is no added complexity as drivers already exist to decode them (see 74LS47). You will need a counter, however, to step the display. The counter will be "stepped" by a timer... could be 555 or whatever... the timer in monostable operation will act as a latch with reset and trigger used... the 'on' duration will determine how quickly it will step through your bar graph... You can reset the timer by the reset pin to begin the next iteration. This needs reset at the end of the period. The latching effect will be enabled by the trigger input. That should work... again I really dont know much about the sensors as I really haven't had the need to build one of these... I may do one on a uC for kicks... practice never hurts!
Finally, the actual values on your diplays are indicative of whatever the ECM is pulling off the timing. Otherwise you'd be reading the small voltages created by the resonate frequencies the peizo transducer is generating based on engine vibration. I think the general rate for the 'low time' on a post ESC knock signal is 4deg pulled/second. That is also not set in stone as there are other techniques to curve it if other things are not happening... One problem may be that different ECU's in different vehicles don't use that convention, requiring the timing to be recalibrated. That's kind of what the issue is... the knock sensor isnt what's calculating how much timing to remove, that is done after the sensor and is specified by the ECM ultimately. Saying that, a scanmaster is like 200 bucks. However, it is always more neato-bandito building it yourself... you just have to figure out if the time investment is worth it.
Of course, you will all scour this for errors i'm sure... it's getting late over here. Electronics and AM's usually don't mix well.
If I were making one, I would make it all on a PIC cycling at 10-20MHz; should be waaay fast enough for instruction set to check signal input. With a uC you could easily build in the calibration and drivers in one unit, in the code. Buffering of the analog signal would be handled at the ADC. Most units have resolutions of 8bits for the ADCs... many 18fxxx series has res. of 10b on ADC lines. More than enough IMO. That would all be for using the knock signal directy, which is what I would do since it would be interchangeable for any application. Besides, code is always easier to change than hard parts.
If you want to use the existing ESC module the thresholding and filtering is done for you already (this is actually difficult to set w/o proper testing due to engine resonances, etc...); the calibration on how long it is held low I currently do not know.. A brute force calibration method would be to simply use a scanmaster, DS, DFI etc... to check your readout, be it numeric (seven segment(s) or just LEDs... I will recommend you just do seven segments as there is no added complexity as drivers already exist to decode them (see 74LS47). You will need a counter, however, to step the display. The counter will be "stepped" by a timer... could be 555 or whatever... the timer in monostable operation will act as a latch with reset and trigger used... the 'on' duration will determine how quickly it will step through your bar graph... You can reset the timer by the reset pin to begin the next iteration. This needs reset at the end of the period. The latching effect will be enabled by the trigger input. That should work... again I really dont know much about the sensors as I really haven't had the need to build one of these... I may do one on a uC for kicks... practice never hurts!
Finally, the actual values on your diplays are indicative of whatever the ECM is pulling off the timing. Otherwise you'd be reading the small voltages created by the resonate frequencies the peizo transducer is generating based on engine vibration. I think the general rate for the 'low time' on a post ESC knock signal is 4deg pulled/second. That is also not set in stone as there are other techniques to curve it if other things are not happening... One problem may be that different ECU's in different vehicles don't use that convention, requiring the timing to be recalibrated. That's kind of what the issue is... the knock sensor isnt what's calculating how much timing to remove, that is done after the sensor and is specified by the ECM ultimately. Saying that, a scanmaster is like 200 bucks. However, it is always more neato-bandito building it yourself... you just have to figure out if the time investment is worth it.
Of course, you will all scour this for errors i'm sure... it's getting late over here. Electronics and AM's usually don't mix well.
Sure, lets (well you mostly ) design a new one from scratch. I'm willing to learn this and actually going to sign up for a class or 2 at the local community college on some electrical courses. This stuff really interests me but I gotta get up to par with the terms and the "why" of some of this stuff. I have learned a LOT more then I knew in the last couple days looking around on the internet. With that said...
As far as counts. I looked on my programmer for my GTP and couldn't find anything to monitor counts so I could compare it to retard. If we use like 10 LED's I would imagine something like the first 4 be green, next 3 be yellow, and the last 4 be red. Something like 1* or knock retard for each LED. I would imagine this wouldn't be spot on using the counts but something close to that. So if anyone has any #'s comparing counts to retard chime in The 3rd gear activation sounds good. Cause if it doesn't reset for all that time the fasle KR when my car shifts sometimes could be hiding some real kr. Maybe a way to bypass it if possible and a manual reset??
And Phil... After reading your post I'm totally lost I got some stuff but man I'm tired (and too new at this) to understand all that different language hahaha. BTW, I did have a scanmaster but I sold it go get my powr tuner. I wish I still had it cause it was nice to monitor without lugging around a laptop all the time. Scanmasters are freaking expensive for the 3800's too. I think it was over $300 when I bought it
One more thing. Since I'll probably have to do a bunch of soldering on small circuit boards is there a soldering gun you guys recommend for small stuff? I'm sure it's about time to get a new one anyway as my $4 one from harbor freight has seen better days I'm so tired I have don't know if I am really making sense right about now. Had a long day with the g/f
) design a new one from scratch. I'm willing to learn this and actually going to sign up for a class or 2 at the local community college on some electrical courses. This stuff really interests me but I gotta get up to par with the terms and the "why" of some of this stuff. I have learned a LOT more then I knew in the last couple days looking around on the internet. With that said...As far as counts. I looked on my programmer for my GTP and couldn't find anything to monitor counts so I could compare it to retard. If we use like 10 LED's I would imagine something like the first 4 be green, next 3 be yellow, and the last 4 be red. Something like 1* or knock retard for each LED. I would imagine this wouldn't be spot on using the counts but something close to that. So if anyone has any #'s comparing counts to retard chime in
The 3rd gear activation sounds good. Cause if it doesn't reset for all that time the fasle KR when my car shifts sometimes could be hiding some real kr. Maybe a way to bypass it if possible and a manual reset??And Phil... After reading your post I'm totally lost
I got some stuff but man I'm tired (and too new at this) to understand all that different language hahaha. BTW, I did have a scanmaster but I sold it go get my powr tuner. I wish I still had it cause it was nice to monitor without lugging around a laptop all the time. Scanmasters are freaking expensive for the 3800's too. I think it was over $300 when I bought it One more thing. Since I'll probably have to do a bunch of soldering on small circuit boards is there a soldering gun you guys recommend for small stuff? I'm sure it's about time to get a new one anyway as my $4 one from harbor freight has seen better days
I'm so tired I have don't know if I am really making sense right about now. Had a long day with the g/f
salvageV6
Daily Driver
- Joined
- May 25, 2001
I found some chips in my stock that should be easy enough to use and source from an electronics store like in the NTE replacement semiconductor line many of those shops use.
Radio shack carries some small ciricuit proto-boards you can solder stuff to and decent enough irons, I use the one with the handle and screw in elements and tips. On the bench I use a professional Weller with digital readout I found in a dumpster.
The problem is the EST line is not an ECM processed line, it's the output of the conditioning circuit of the knock sensor just uncounted KR pulses so you can't relate to degrees of KR or how much timing is pulled by the ECM unless you monitor the ECM itself and it's counters. Direct Scan can do it as well as the scanmaster. It's processed in the ECM by that computer but the bad news there is it's slower output than what we will be tying to the EST line.
But we can make the LED thingy probably using 2 counts per LED with a scheme to add or subtract parts to alter this count, that part will be on paper, with 10 LEDs.
Gives you a count of 20 EST pulses for the last red LED should be plenty of KR or banging of stuff going on in the motor on that last LED.
For all the options used we may be talking up to (6) 14 or 16 pin chips, and the price might exceed $40 or so for everything excepting the soldering iron.
Good news is it should be as good or better than anything out there.
If it works.
Good news is I will be trying it before writing it all down.
Radio shack carries some small ciricuit proto-boards you can solder stuff to and decent enough irons, I use the one with the handle and screw in elements and tips. On the bench I use a professional Weller with digital readout I found in a dumpster.
The problem is the EST line is not an ECM processed line, it's the output of the conditioning circuit of the knock sensor just uncounted KR pulses so you can't relate to degrees of KR or how much timing is pulled by the ECM unless you monitor the ECM itself and it's counters. Direct Scan can do it as well as the scanmaster. It's processed in the ECM by that computer but the bad news there is it's slower output than what we will be tying to the EST line.
But we can make the LED thingy probably using 2 counts per LED with a scheme to add or subtract parts to alter this count, that part will be on paper, with 10 LEDs.
Gives you a count of 20 EST pulses for the last red LED should be plenty of KR or banging of stuff going on in the motor on that last LED.
For all the options used we may be talking up to (6) 14 or 16 pin chips, and the price might exceed $40 or so for everything excepting the soldering iron.
Good news is it should be as good or better than anything out there.
If it works.
Good news is I will be trying it before writing it all down.
