Code 43 Knock Sensor Circuit Failure Issues

[Daveo91]

I decided to start a new post for this situation since it's very specific. But background here (http://gearhead-efi.com/Fuel-Injection/showthread.php?9959-First-start-7060-to-7427-Conversion-Crazy-rich-at-idle)on my '7060 to '7427 conversion if it's helpful. Still lots to resolve on the conversion, but I think the most important thing right now is to get the KS issue resolved.

So the weird thing about the code 43 I'm getting (in TunerPro and SES illuminates) is that it will set either immediately after starting the engine OR it will set after about 30 seconds of key on, engine NOT running. For both tests I'm sure the code is cleared by unplugging both ECM connectors for a minute and restarting TunerPro to clear the flag there. And both tests are consistent. Is it strange that the flag will set without the engine running? Stuff I've seen says that it does not do its diagnostic until after engine is running. Does that point to a possible ECM or Memcal issue? I will go through the FSM diagnostic procedure on the KS circuit, but wanted to check here to see if maybe that's not necessary based on the above observations.

[tayto]

do you have correct knock sensor? i believe on 7427 knock module is with the memcal, you can try reseating the memcal.

[Terminal_Crazy]

I decided to start a new post for this situation since it's very specific. But background here (http://gearhead-efi.com/Fuel-Injection/showthread.php?9959-First-start-7060-to-7427-Conversion-Crazy-rich-at-idle)on my '7060 to '7427 conversion if it's helpful. Still lots to resolve on the conversion, but I think the most important thing right now is to get the KS issue resolved.

So the weird thing about the code 43 I'm getting (in TunerPro and SES illuminates) is that it will set either immediately after starting the engine OR it will set after about 30 seconds of key on, engine NOT running. For both tests I'm sure the code is cleared by unplugging both ECM connectors for a minute and restarting TunerPro to clear the flag there. And both tests are consistent. Is it strange that the flag will set without the engine running? Stuff I've seen says that it does not do its diagnostic until after engine is running. Does that point to a possible ECM or Memcal issue? I will go through the FSM diagnostic procedure on the KS circuit, but wanted to check here to see if maybe that's not necessary based on the above observations.

Check the cables aren't rubbing the bodywork between sensor & PCM. If it's in a 4th gen F body, it runs up the side of the tunnel in silver foil next to the header. I don't think the diagnostic checks cable against ground.

Mitch

[Daveo91]

duplicate post

[Daveo91]

Check the cables aren't rubbing the bodywork between sensor & PCM. If it's in a 4th gen F body, it runs up the side of the tunnel in silver foil next to the header. I don't think the diagnostic checks cable against ground.

Mitch

Thanks, it's a '91 squarebody suburban, but I'll check for any wire issues. I have previously checked for continuity between the PCM and KS - good, but maybe there's a persistent problem with a ground wire somewhere.

[Daveo91]

do you have correct knock sensor? i believe on 7427 knock module is with the memcal, you can try reseating the memcal.

Pretty sure I have the correct KS: Delphi AS10016. As recommended by @daveW and it's what shows up for a '94 350 truck. I'll try and reseat the memcal - thanks great suggestion. I really think it's something along these lines since it flags as soon as engine starts or even after 30 seconds with key on. Also, KS appears to have been working for my first few runs after the conversion.

[tayto]

pretty sure 7060 and 7427 share the same knock sensor, but it's an easy check to do with DMM.

[Daveo91]

pretty sure 7060 and 7427 share the same knock sensor, but it's an easy check to do with DMM.

Yep, you're correct, they are. I tested resistance on both, old one was around 100K ohms which AFAIK is right within spec, but the new one was only testing around 2K. So I swapped them and am running the old now. But I'm still having the same problem with the flag setting within 30 seconds, engine running or not. I'm wondering if the low resistance could have killed the memcal or something in the PCM?

[Daveo91]

I might have solved this issue with the KS, but not exactly sure how.

First, I'm fortunate to live in the same city as Dave W (Vancouver, WA) He let me try a different 7427 and memcal. Exact same issue, no change at all so we were both mystified. In the course of this, Dave informed me that the KS with the much lower resistance - that I had taken out - was indeed the correct one. So tonight I reinstalled the correct (Delphi AS10016) KS, and, lo and behold, no code. I have not driven it yet, but left the key on for twice as long as it was taking to set the flag before, and then started engine and let it run for a bit. No code. Before it would have flagged within 20-30 seconds of either just key on or engine running. Won't know for sure till I drive it which won't be for a week or so now.

With the correct KS installed I back-pinned a paper clip in to the B15 terminal so I could check voltage on the KS circuit to engine ground: 2.43 (key on) and I think 2.48 or so engine running. Do those voltages seem correct?

[stew86MCSS396]

Not sure if this covers your year PCM...
https://www.2carpros.com/images/question_images/395315/original.jpg

[In-Tech]

That's a great fault tree, thanks for posting. Is that for the 7427 auto pcm or for the '95 manual ecm with a 2732a and not a memcal?

[stew86MCSS396]

Not 100% certain but do seem to recall the latter would accompany escm and the fault tree would make some sort of reference to it.

[In-Tech]

I can't remember if it had a knock module. I put a whipple on that '95 manual and quickly converted it to a 7427 'puter. Sheesh, I want to say that was in 95 or 96 so, a couple days ago :happy:

[Daveo91]

Not sure if this covers your year PCM...
https://www.2carpros.com/images/question_images/395315/original.jpg

Awesome - thank you for the fault tree. So sounds like the consensus is that this is for a '7427 auto truck?

[Daveo91]

So I'm not entirely sure what I did other than re-install the correct knock sensor, but now my code 43 problem has shifted to intermittent. It does not come on within 30 seconds of key on or engine running like previously, now it comes on after about 10 minutes of driving. Both times it was after a decel event, coming to idle. One of those two times, the SES light went out on the way back home, the other time it did not.

Even while the flag is set, the knock sensor circuit appear to be working and showing knock counts in TP. A lot of them in fact, 10s of thousands per drive. A single acceleration event can trigger thousands and consistent 10 degrees plus of retard. There are no audible knocking sounds, and no unusual mechanical sounds ,that I can hear at least. I do have a roller cam (stock L31 complete set up including cam, lifters, heads, spider, retainers, etc. And I used a new set of Comp Cams stamped steel, roller-tip rockers. Could the knock sensor I'm using be incompatible with roller cam? A Google search indicated that can be an issue. If so, thoughts on which KS to use?

I have not done this yet, but I will remove a couple of plugs to check condition. I would think tens of thousands of knock counts would be pretty obvious by looking at the plugs?

[stew86MCSS396]

Idk if this is what you’re describing. I’m thinking it wouldn’t throw a code if it is.
http://www.gearhead-efi.com/Fuel-Injection/showthread.php?3290-24*-Knock-retard-no-knock-counts

[tayto]

do you have another memcal you can try? also f found wrong heat range, or fouled/worn spark plugs can cause high knock counts. also if you have exhaust touching the chassis or a worn motor mount is a good place to check. other thing i should mention, what did you torque the sensor too? IIRC, 14 ft-lbs. tightening it to much can cause issues as well.

[NomakeWan]

As others have alluded to, changing the knock sensor isn't going to change the outcome. The only thing the ECM cares about regarding the sensors is the resistance. It uses that resistance to check if the sensors are physically in place, but nothing else. The sensors will continue to operate properly even if the resistance is out of spec and Code 43 gets thrown. This is very common; the resistors inside this type of knock sensor tend to fail well before the sensor element itself does.

But the actual thing that's taking the signal from the knock sensor and telling the ECM that it's knock? That's the knock module, which on your ECM is integrated into the memcal. To change how your ECM interprets knock, you would have to change that module to a different one. Or disable knock feedback entirely, forget about your knock sensors, and just make sure your tune is up to snuff. Or I guess run some sort of external knock feedback rather than relying on the ECM.

Best of luck.

[Daveo91]

Thanks everyone. I went out and poked around tonight. My PS exhaust manifold has a heat stove around it (the kind that sends warm air up to air intake for cold operation) that is loose and may be rattling on the manifold. Could that cause a knock signal? I don't use it any more so I think I will remove it to be sure that's not causing the problem.

Also pulled four spark plugs to take a look. 3 of the 4 looked pretty normal to me, but the one from #1 looked silver-ish, like maybe it's too hot. First pic is representative of the three plugs that looked similar, while the other two are from the one I'm thinking is showing hot. I'm thinking I'm going to replace them with one heat range cooler. Currently have Delco R44LTS, probably going to go to 43s.

16978

16979

16980

[tayto]

can stuff the heat shield with steel wool dish scrubbers.what compression is your engine? i had a sbc 10:1 compression with vortec heads and went from 44LTS to 42LTS and nearly got rid of my knock counts that were maxing out every datalog.

[Daveo91]

can stuff the heat shield with steel wool dish scrubbers.what compression is your engine? i had a sbc 10:1 compression with vortec heads and went from 44LTS to 42LTS and nearly got rid of my knock counts that were maxing out every datalog.

Thanks, great input. My 383 has about 9.5:1 - I agree I think the 44s are too hot for it. I also have a Vortec 355 with similar compression in my '76 Corvette and I've always run 43s in that. But it's got a q-jet and no knock sensor.

Should I be worried about going too cold on the plugs though? Low RPM truck motor...could foul plugs more easily?

[tayto]

yes if they foul then too cold

[tayto]

dave, any update on this? I was getting 1000s of knock counts just idling. Found the knock sensor connector was broken, fixed that and also replaced my knock sensor w/ new ac delco. torqued to 14 ft-lbs per FSM. counts have gone down considerably but still there. i can get it to active the knock retard by just reving it in park. thought it might have been a belt accessory, but with belt removes there is still a racket. since this is a high mileage rig, 211,000 miles, i am thinking the timing chain is slapping around. i will be replacing along with new cam this winter, owner needs truck back soon.

[Daveo91]

dave, any update on this? I was getting 1000s of knock counts just idling. Found the knock sensor connector was broken, fixed that and also replaced my knock sensor w/ new ac delco. torqued to 14 ft-lbs per FSM. counts have gone down considerably but still there. i can get it to active the knock retard by just reving it in park. thought it might have been a belt accessory, but with belt removes there is still a racket. since this is a high mileage rig, 211,000 miles, i am thinking the timing chain is slapping around. i will be replacing along with new cam this winter, owner needs truck back soon.

Sorry for the long delay in reply - I was out of town with no access through Sunday night. After changing plugs and fixing the things I think were causing noise, I haven't driven it much - no serious data logging yet. I really just drove it to get it warmed up to set base idle/min air. I did have the computer monitoring on at least one of those drives and I was watching knock counts a little. It looked like I was still getting some knock, but not near as much as before. And it seemed like it was not doing much knock retard any more. I hope to do some data logging this week or if not, this weekend. I'lll post results.

Will you just turn off knock retard / Code 43 error in the BIN for now?

[tayto]

no, i only get SES if i hold higher rpm @ idle in park for extended periods. i had to work at it too get code to set. should take my own advice and replace plugs, the current ones look ok but i didn't install them so you don't know if they've been dropped or overtightened and porcelain has a hair line crack. drivebility seems ok, i should disconnect knock sensor and datalog to see if i'm still getting counts....

[Daveo91]

no, i only get SES if i hold higher rpm @ idle in park for extended periods. i had to work at it too get code to set. should take my own advice and replace plugs, the current ones look ok but i didn't install them so you don't know if they've been dropped or overtightened and porcelain has a hair line crack. drivebility seems ok, i should disconnect knock sensor and datalog to see if i'm still getting counts....

I finally got a chance to do some data logging tonight. 27K knock counts in about a 10 minute drive. Watched the whole datalog so at least I know it didn't turn over in that time frame. So it's less than it was before I put new plugs in and tightened stuff, but still significant. I'm getting a lot of knock retard in the 3 deg or less range, but also getting some up to 10 deg in heavy accel or load. And it was showing 3 deg on the freeway at steady 60 mph, no hill. (but it was only a mile on the freeway)

How do I know if this is real knock or it's just hearing something mechanical? Truck has stock L31 valvetrain (roller lifters) and also Comp Cams stamped steel roller-tip lifters. Would either of those cause noise?

In your case, how would it register any knock at all if the KS is disconnected? (Not being a smart ass, I just want to know) Should I do that test?

One last question, I have PE turned off for narrow-band VE tuning. Should I turn it back on? Maybe a lean condition during accel is causing knock? I'm about to switch to WB tuning so I think it should be on anyway for that?

[tayto]

mine wasn't registering knock with sensor disconnected. If it was i would suspect ECM. I worked on a '727 swapped into an older car that was running funny. found if you tapped the ecm case it would detect knock. replaced ecm and problem went away.

[NomakeWan]

How do you know if it's real knock? Put high octane fuel in and pull all the timing. It won't make much power, but it also won't knock. If it says you're knocking anyway, it ain't real.

[Daveo91]

How do you know if it's real knock? Put high octane fuel in and pull all the timing. It won't make much power, but it also won't knock. If it says you're knocking anyway, it ain't real.

I'll try that - eventually. I just put $100 of 87 octane in, as I was pumping I realized I should have put in 93 to see what it does. But I am going to pull timing and see what that does. I found this thread (http://www.gearhead-efi.com/Fuel-Injection/showthread.php?1233-1998-Vortec-Timing-Table-(Black-Box-Computer)-Open-Closed-Throttle-ONLY) with the '98 vortec timing tables. They are all lower values than the ones in my BIN, up to 7 or 8 degrees. My engine is a 9.5:1 vortec 383 so I'm going to see what happens with those timing values.

[Daveo91]

Switched to the L31 timing tables (all cells at least 3 deg less than what I had before, many up around 7 or 8 less) and still getting high knock counts and lots of knock retard. Attached is a screen shot of knock retard running average values from about a 20 min log with freeway driving. It seems like it might be mostly fake knock? I'm also still getting an intermittent code SES light with Code 43 showing up in the log. Light comes on after 15-20 min of driving while coming to a stop. Usually turns off after a minute or two more.

One thing I've been thinking about: I switched knock sensors a couple of times due to confusion about which one is correct. A new in box sensor has the orange teflon tape for sealant, but after a re-install I didn't trust that so I used a dab of ultra black Permatex. Could that be part of my problem? What is the proper sealing method after re-installing a KS?

17144

[NomakeWan]

Generally adding more sealant makes a knock sensor less sensitive, not more. Also usually when looking at knock I look at a scatterplot; it's what Trimalyzer does really well, for example.

[tayto]

Generally adding more sealant makes a knock sensor less sensitive, not more. Also usually when looking at knock I look at a scatterplot; it's what Trimalyzer does really well, for example.

dave just curious but what datalog cable are you using?. i read an old post on thirdgen that may be something to consider.... https://www.thirdgen.org/forums/diy-prom/247135-idles-faster-when-aldl.html

Basically it's saying to use 10K resistor to connect then turn it off once connected. i know mine was always on 10k...

[NomakeWan]

dave just curious but what datalog cable are you using?. i read an old post on thirdgen that may be something to consider.... https://www.thirdgen.org/forums/diy-prom/247135-idles-faster-when-aldl.html

Basically it's saying to use 10K resistor to connect then turn it off once connected. i know mine was always on 10k...
Considering the 10k resistor thing only applies to the earlier ECMs that used Pin E for 160 Baud ALDL, I'm gonna go ahead and say that won't be it. The '7427 is a '93 PCM, so it'll be using Pin M 8192 ALDL without the resistor.

[Daveo91]

Generally adding more sealant makes a knock sensor less sensitive, not more. Also usually when looking at knock I look at a scatterplot; it's what Trimalyzer does really well, for example.

Regarding the sealant, I'm wondering if that could be the cause of my intermittent Code 43? Body of sensor not getting good ground to block at times? Regarding the scatterplot, I've never used Trimalyzer - is it free? I'm decent with Excel, possible to just plot data in Excel with TP output?


dave just curious but what datalog cable are you using?. i read an old post on thirdgen that may be something to consider.... https://www.thirdgen.org/forums/diy-prom/247135-idles-faster-when-aldl.html

Basically it's saying to use 10K resistor to connect then turn it off once connected. i know mine was always on 10k...

I'm using the Moates ALDL cable, the one with the regular USB-A on the computer end. Same cable I was using with the '7060 and I didn't mess with the 10K resistor when I switched. I also have a Moates AutoPROM with a ethernet-type plug on the computer end. I haven't used the AutoProm at all since I switched to the '7427. I had a lot of challenges with it when I used it with the 7060 so I gave up. I think I'll try again, just to see if it helps and it may help with my tuning by allowing me to try things on the fly.

[NomakeWan]

Regarding the sealant, I'm wondering if that could be the cause of my intermittent Code 43? Body of sensor not getting good ground to block at times? Regarding the scatterplot, I've never used Trimalyzer - is it free? I'm decent with Excel, possible to just plot data in Excel with TP output?
It's possible you aren't getting a good ground, sure. But as I believe I've said before in this very thread, the code isn't really important; the code doesn't mean anything about the signal from the sensor. It's looking for a voltage difference due to a resistor inside the knock sensor that isn't actually related to the function of the sensor itself. So if you're getting proper signal from the sensor, Code 43 is irrelevant. Again, on my '95 Corvette I just disabled the code rather than bothering to replace the sensors, because the sensors were still providing good output.

Trimalyzer is absolutely free. You just need to tell it how your CSV logs are formatted so it can plot the data for you. http://fbodytech.com/trimalyzer

It is absolutely possible to do a scatterplot with Excel as well (since that's all Trimalyzer is actually doing).

[Daveo91]

It's possible you aren't getting a good ground, sure. But as I believe I've said before in this very thread, the code isn't really important; the code doesn't mean anything about the signal from the sensor. It's looking for a voltage difference due to a resistor inside the knock sensor that isn't actually related to the function of the sensor itself. So if you're getting proper signal from the sensor, Code 43 is irrelevant. Again, on my '95 Corvette I just disabled the code rather than bothering to replace the sensors, because the sensors were still providing good output.

Trimalyzer is absolutely free. You just need to tell it how your CSV logs are formatted so it can plot the data for you. http://fbodytech.com/trimalyzer

It is absolutely possible to do a scatterplot with Excel as well (since that's all Trimalyzer is actually doing).

Thanks - Downloaded Trimalyzer. Really easy to use, had a scatter plot of my knock counts in a few minutes. (attached) but not sure what it's telling me. Most points are in the middle RPM range, not surprising. Looks like there are some anomalies in the low RPM/low load area?

OK, I'll start learning to not worry about the code 43 and will probably eventually disable it in the code. But I am still concerned about my very high knock counts. 17149

[NomakeWan]

Holy crap. Either you have a ton of knock, or your knock filter is shot, or...I dunno, you just magically have an engine with a sound profile that, over its entire range, just happens to fall within your filter parameters?

I've never seen a knock plot that busy before. That's insane. Even the one time I definitely pushed timing too far (and disabled knock retard), all that happened was I developed a blob in one specific spot. Not a ton of tiny dots all over the entire plot.

[Daveo91]

Generally adding more sealant makes a knock sensor less sensitive, not more. Also usually when looking at knock I look at a scatterplot; it's what Trimalyzer does really well, for example.


dave just curious but what datalog cable are you using?. i read an old post on thirdgen that may be something to consider.... https://www.thirdgen.org/forums/diy-prom/247135-idles-faster-when-aldl.html

Basically it's saying to use 10K resistor to connect then turn it off once connected. i know mine was always on 10k...


Holy crap. Either you have a ton of knock, or your knock filter is shot, or...I dunno, you just magically have an engine with a sound profile that, over its entire range, just happens to fall within your filter parameters?

I've never seen a knock plot that busy before. That's insane. Even the one time I definitely pushed timing too far (and disabled knock retard), all that happened was I developed a blob in one specific spot. Not a ton of tiny dots all over the entire plot.

Yeah, that's what I'm saying, an insane amount of knock. I really don't think it's truly knocking. It seems I'd hear it and the motor sounds pretty good throughout the range, with the possible exception of WOT at high load. I've asked a couple times in this thread - can it be the L31 roller cam and valve train? Did the vortec trucks use a different KS and/or filter parameters? I also have roller tip rockers but I don't think they'd be causing much noise?

I started playing with my autoprom last night. It's always been finicky with connections both to PC and ALDL. I had to fiddle with COM port settings to get it to connect to PC at all. It eventually connected, but then wouldn't connect again after that. I'm really hoping it's a faulty USB cable, it seems that the connection was sensitive to touching the cable. It uses an old school USB-A to USB-B; I'll have to hunt around in some boxes to see if I can find another one of those.

[NomakeWan]

Did the vortec trucks use a different KS and/or filter parameters?
Every GM vehicle had a unique knock filter, because GM understood that every operating environment was different, so to improve their ability to detect knock across a wide operating range, they tuned each one according to the platform in question. If your knock filter does not match your platform, then it's not doing its job correctly. Either find a knock filter that matches your setup (chassis, engine configuration, etc), make your own knock filter if such a filter does not exist, or as I think I also mentioned before, just disable knock feedback entirely since it's not working correctly anyway.

[Daveo91]

Every GM vehicle had a unique knock filter, because GM understood that every operating environment was different, so to improve their ability to detect knock across a wide operating range, they tuned each one according to the platform in question. If your knock filter does not match your platform, then it's not doing its job correctly. Either find a knock filter that matches your setup (chassis, engine configuration, etc), make your own knock filter if such a filter does not exist, or as I think I also mentioned before, just disable knock feedback entirely since it's not working correctly anyway.

Well finding an exact match is out - no such thing as a square body suburban with vortec, roller-cam 383 running a '7427 PCM. Closer match, I'm not sure. I do have a vortec black box PCM but I have no idea if there's any compatibility there. And making my own filter is probably out, too, since I wouldn't even know where to start. So unless I figure out some other magic solution I may have to just disable knock feedback but definitely not an ideal solution since I'd really like to get closer to perfection on ignition timing.

[NomakeWan]

Generally any tuning place worth their salt is going to have a knock box. It's an external knock processor with its own knock sensor. Mount it to the motor you're tuning, put on headphones, actually listen for knock and tune that way. You don't rely on the ECU to do it for you, you do it yourself.

It sounds like that's the route you need to go.

[tayto]

Considering the 10k resistor thing only applies to the earlier ECMs that used Pin E for 160 Baud ALDL, I'm gonna go ahead and say that won't be it. The '7427 is a '93 PCM, so it'll be using Pin M 8192 ALDL without the resistor.
Weird because the 7427 swap I did wouldn't connect to Turnerpro RT unless I had my cable set to 10K. i do not have the vehicle here to confirm is turning 10K off will stop my knock counts climbing.

I would also try another PCM/memcal just as a sanity check before getting too carried away here

[NomakeWan]

Weird because the 7427 swap I did wouldn't connect to Turnerpro RT unless I had my cable set to 10K. i do not have the vehicle here to confirm is turning 10K off will stop my knock counts climbing.

I would also try another PCM/memcal just as a sanity check before getting too carried away here
That would be very strange indeed. I confirmed from the Blazer forums that they are indeed using native 8192 baud comms out of Pin M (or Pin 9 on '95 vehicles), without any resistor.

[tayto]

I will try again the next time the truck is here, could have been something else and switching to 10k was just a coincidence

[tayto]

Truck was by today, would connect with switch in Open or 10K. Also having 10K position had no affect on the increasing knock counts. Most of the counts were on start up and switching between R and D ranges. Also, when counts went up I never saw the knock retard actually pull any timing FWIW. I still think the issues with this truck are worn valve train and timing set, will know early next year when I do cam swap.

PS; I really need to log this thing and build a better VE, BLMs were 113. I guess better rich and pulling fuel than lean........

[Daveo91]

Truck was by today, would connect with switch in Open or 10K. Also having 10K position had no affect on the increasing knock counts. Most of the counts were on start up and switching between R and D ranges. Also, when counts went up I never saw the knock retard actually pull any timing FWIW. I still think the issues with this truck are worn valve train and timing set, will know early next year when I do cam swap.

PS; I really need to log this thing and build a better VE, BLMs were 113. I guess better rich and pulling fuel than lean........

thanks for the follow up. I’m taking a break from worrying about knock, KS, KR etc. I have min temp for KR set to max right now. (I was definitely getting Knock retard - at all the MAP/RPM values.)

ive been playing with my autoprom. I had a bad USB cable so that was an easy fix. Got it emulating and works pretty good although there is an occasional stutter where everything zeros out and RPM cuts out etc. seems like a known issue after researching - might be solved by making an ALDL adapter that only uses the three connections needed for the data stream. I still have a lot of work to do on fuel tables and related so focusing on those next.

after autoprom emulation fixed and fuel tables in good shape I’ll go back to messing with spark and knock. Still not convinced it’s not a wiring issue or something similar. Screwdriver tapping on valve cover or exh manifold seems to have no effect

EDIT: I too played with 10k or open on the ALDL - no effect on knock count