Desensitize the knock sensor 94/95 LT1

[roachjuice]

I’ve been chasing some false knock for a couple of months now.
Ever since I installed some headers it’s pulled timing 4000+ at WOT.
I’ve knocked it down to only pull 4 degrees so it’s helped out a lot.
But what table(s) does one adjust to desensitize the knock sensor?
I thought it was the fast attack knock rate table but it doesn’t seem to be helping.
Im running 91+ octane. Timing tables are stock. I’ve actually taken 1 degree out at 90kpa and 4000rpm.
AFR is 12.7 @ 4000 and creeps to 13.0 @ 5400. All at WOT.
I’m thinking about just zeroing the table out at WOT 4000+
I know it’s not real knock. I’ve actually lowered the timing live with EEHack and had the same result.
If someone needs a log I’ve got a couple.
Thanks.

[NomakeWan]

There is no table for it. Knock Detection is handled by a separate PCB entirely, in what's known as the "Knock Module." it's attached into a slot on the underside of the PCM, under a cover.

This module's job is to take the raw output from the knock sensors, filter them based on its internal values as to which are "real knock" and which are "noise," then if I recall correctly will pull a pin low to tell the PCM that knock is occuring. When and how long this pin is pulled low is then used by the PCM to retard timing.

At present there is no real good way to fix it. Some will say that the LT4's knock module is less sensitive to false knock because of the switch to roller rockers in the LT4 requiring a new knock module program to compensate, but I have not seen any empirical data comparing the two modules against one another. In addition, these knock modules are now going for over $300 on eBay as they have been out of production for many years.

If you are handy with an oscilloscope, you could hook up to the knock sensors themselves and try to figure out which frequencies represent real knock on your car versus which don't, then tear apart another knock module (one less expensive than the LT4, haha), reverse-engineer its functionality, and build your own. I'm actually looking into doing that personally.

Or you could just give up on the electronic spark control system altogether and set maximum knock retard to 0 across the board. If you know your car isn't knocking, then the system is only hurting you.

Best of luck no matter which path you choose.

[spfautsch]

Have you thoroughly checked that the exhaust system isn't making contact with the body somewhere? Possibly due to driveline torque loading?

I've dealt with all sorts of false knock. The roller rockers and pushrod guides seem to have been my primary source, but since going to shaft mount rockers it seems to be completely cured.

Long story short, the knock sensors on these cars are extremely sensitive and LT4 knock modules are very scarce so I'd eliminate all possible sources before investing in one. I once had a cracked exhaust tip causing resonance that was causing severe false knock.

[roachjuice]

The passenger side header likes to tap the floor a little bit lol.

[vilefly]

Well, I am not sure will work, but I know what won't work. An inline resistor to the knock sensor won't work. There is a 5v reference voltage sent to the 4K impedence of the sensor. The ecm expects to see this voltage drop to 2.5v when connnected. This means there is a 4k resistor in the ecm series with a 5v supply, which has a tap that heads to an input to the processor. (voltage divider) Any extra resistance will set a code, so it gets a little tricky messing with it.

Perhaps a mechanical extension on the knock sensor might dampen things (or let it resonate more, dunno). Not sure how close it is to the headers, though. I can't remember if the knock sensor uses pipe thread or not.

14762

[steveo]

you don't really have control over when knock counts occur but you definitely have a control over how they're interpreted

if it occurs in a narrow-ish RPM band i'd reduce ignition advance in that region to safe-ish levels and zero that area in the max knock retard vs RPM table.

if it occurs in a wide area, then disabling the entire system is usually what i'd do.

as far as a hardware solution, i'd suggest passive filtering on the sensor itself rather than hacking the knock module. if you can identify the frequency, a notch filter isn't that hard to build.

headers setting off the knock sensor isn't really a problem i've had unless they're smacking the frame, usually valvetrain changes do it to me.

[vilefly]

Passive filtering. I suspect this would be how to wire it up.

14763

[roachjuice]

you don't really have control over when knock counts occur but you definitely have a control over how they're interpreted

if it occurs in a narrow-ish RPM band i'd reduce ignition advance in that region to safe-ish levels and zero that area in the max knock retard vs RPM table.

if it occurs in a wide area, then disabling the entire system is usually what i'd do.

as far as a hardware solution, i'd suggest passive filtering on the sensor itself rather than hacking the knock module. if you can identify the frequency, a notch filter isn't that hard to build.

headers setting off the knock sensor isn't really a problem i've had unless they're smacking the frame, usually valvetrain changes do it to me.
around my area we usually only have 91 octane. i found a station with 93 so i filled it up. im running the stock timing tables too.
the car has 209k on it but doesnt make any noises. ill see if i can post up a log.

[roachjuice]

https://drive.google.com/open?id=1bUu3BibCI8TgkwHWVKMc5n6jkdX6OL7K

hopefully this link works.

[roachjuice]

Well now EEHack won’t open after clicking on a log and windows wanted to open EEHack. No clue.

[roachjuice]

anyone run across this issue? ive tried uninstalling and reinstalling

[spfautsch]

Try a system restore to a snapshot before the problem.

[roachjuice]

Try a system restore to a snapshot before the problem.
lol shit

[vilefly]

Hmmm. same thing happened to me months ago. I will have to try that. I had switched computers earlier.

[spfautsch]

That'll be $49, each.

vilefly you seem tech savvy enough to reload winblows. That fixes all software problems, always.

[vilefly]

heh. I'll just have to call off christmas.

[steveo]

why the hell?? i do not understand this bug at all. please provide reproducable steps if you can so i can fix it

[NomakeWan]

Usually "fails and won't let me reinstall" I use the forbidden technique of "uninstall as best I can, delete the directory if still present, then use a registry cleaner like CCleaner to delete the orphaned registry entries." This tends to let me reinstall whatever program was giving me trouble, though to be honest the last time I had that issue was decades ago with a virtual drive, and I've never had such an issue with EEHack.

Your mileage may vary, but cleaning the registry of orphaned entries is a lot easier (and, in my experience, way more reliable) than using system restore. ;)

[spfautsch]

I've been sitting behind a telephone supporting hundreds of windows machines for the better part of 24 years. My mileage never varies. When in doubt I wipe it out (windows installation).

Note that I do not use it on any of my personal machines. Windows runs just fine in VirtualBox on any linux distro.

When my "registry" gets fouled up and eehack won't launch (never had it happen, completely speculative) I just delete /home/spfautsch/.config/EEHackB/fbodytech.com.conf and let the program create a new config file.

[vilefly]

What I had done is allow windows to associate log files with EEhack. Then, I double clicked on a log file, and EEhack does this.
14766
then this.
14767
Perhaps I should post this in a matching section for the subject.

[roachjuice]

Mine was working perfectly.
I tried to open a log without actually having EEHack open. When I clicked on the log, it asked to open EEHack. So I proceeded.
Then threw the error up at me and has done so ever since.
I’ve even tried to download the new beta you have and still the same thing is happening. No clue honestly.

[roachjuice]

I've been sitting behind a telephone supporting hundreds of windows machines for the better part of 24 years. My mileage never varies. When in doubt I wipe it out (windows installation).

Note that I do not use it on any of my personal machines. Windows runs just fine in VirtualBox on any linux distro.

When my "registry" gets fouled up and eehack won't launch (never had it happen, completely speculative) I just delete /home/spfautsch/.config/EEHackB/fbodytech.com.conf and let the program create a new config file.
I’ll try this. Thanks.

[NomakeWan]

I’ll try this. Thanks.
Note that this is for a Linux machine, not a Windows machine. He uses Linux for his EEHack machine.

Registry cleaners like CCleaner will identify that there are registry entries associated with a program/directory that no longer exist (such as the .eehack file association, and any other entries the program created upon installation) and clear them out. This way you can specifically target changes EEhack made, rather than resorting to trying to restore the system to an earlier point in time (something of a carpetbombing approach to a specific issue with a specific program, not to mention unreliable in my experience) or nuking your entire Windows installation and starting over (far more reliable than system restore but a total waste of time and effort for such an unbelievably minor issue).

Just sayin'.

[roachjuice]

Note that this is for a Linux machine, not a Windows machine. He uses Linux for his EEHack machine.

Registry cleaners like CCleaner will identify that there are registry entries associated with a program/directory that no longer exist (such as the .eehack file association, and any other entries the program created upon installation) and clear them out. This way you can specifically target changes EEhack made, rather than resorting to trying to restore the system to an earlier point in time (something of a carpetbombing approach to a specific issue with a specific program, not to mention unreliable in my experience) or nuking your entire Windows installation and starting over (far more reliable than system restore but a total waste of time and effort for such an unbelievably minor issue).

Just sayin'.
i ended up running ccleaner. didnt work.
i reverted back too a previous windows save. eehack works again. i dont keep anything on this pc as its from 2008 lol. all i do with it is tune and datalog and it barely does that. thanks again for the help.

[NomakeWan]

I know it's irrelevant since you already used system restore with success, but did you uninstall EEhack, remove its installation directory, then use the registry cleaning function in CCleaner to remove the orphaned entries? Just for my own personal curiosity.

As I said I haven't run into these issues with it before, but hey, if I do, it'll be good to know.

[spfautsch]

When I get a moment to play with a winblows machine I'll dig through the registry and possibly have a suggestion to fix this involving deleting the file association subkeys for .eedata. I'm relatively sure it's related to the DDE transaction that happens when explorer feeds a flie list to the associated executable.

Of course, the obvious solution to this one is not to associate .eedata files to anything. DDE is the brainchild of insane coders from the 80s.

[roachjuice]

I know it's irrelevant since you already used system restore with success, but did you uninstall EEhack, remove its installation directory, then use the registry cleaning function in CCleaner to remove the orphaned entries? Just for my own personal curiosity.

As I said I haven't run into these issues with it before, but hey, if I do, it'll be good to know. yes I did. It still did the same thing. Even though ccleaner found files EEHack left. I saw them on the list. Oh well. It works now. I’m not concerned. Just word to future people. Don’t open a log without having EEHack open first. It will fuck EEHack all up.

[roachjuice]

When I get a moment to play with a winblows machine I'll dig through the registry and possibly have a suggestion to fix this involving deleting the file association subkeys for .eedata. I'm relatively sure it's related to the DDE transaction that happens when explorer feeds a flie list to the associated executable.

Of course, the obvious solution to this one is not to associate .eedata files to anything. DDE is the brainchild of insane coders from the 80s.
Yea I’m sure it’s related to the DDE. I know what that is. I just don’t wanna mess with it lol. Minor glitch for an awesome program.

[steveo]

its crazy this is happening. eehack is a really basic program with really generic (but bloated because i use qt) library set.

i made the file association with this generic installer-maker-thingie program i was using.

i lost the install script and forget which program i used....... so ill be using something else for the next version

hopefully that means problem solved....?

we shall see.

i refuse to give a crap about or research or look into how windows file associations actually work in modern versions of windows.

[spfautsch]

i refuse to give a crap about or research or look into how windows file associations actually work in modern versions of windows.

Now you're starting to sound almost as curmudgeonly as I do. When I went to work for my current employer (21 years ago) I was the Microsoft kool-aid drinker, and my boss was the dyed-in-the-wool linux bigot. Ironically, our roles have completely reversed. His current punishment for me is deploying Exchange server after running sendmail and then postfix for the past 25 years. I would rather drink gasoline.

Personally, I would simply remove the file associations but that's because I never bothered creating one in my current setup and there is no linux installer package. But that's me - the first time I went to open an old log in eehack I brought the datalog window up and habitually pressed ctrl+o only to find there are no accelerator keys.

[steveo]

i think what might be happening on first glance is that when opening that file via windows, it messed with the 'start in' path and no longer gets the definition file, then freaks out and crashes. it's crashing when loading the definition not the actual log.