My check engine light came on a few months ago and I had the codes read at Autozone. I had a "knock sensor low voltage" code P0327. My truck ('05 GMC, 5.3L) has two knock sensors but only one of them was throwing a code. I went ahead and replaced both of them, plus the harness/pigtail that goes under the intake manifold. Disconnected the battery to reset the ECU, check engine light went away, but came back within 50 miles of driving. Had the codes read again, and now I'm getting the same low voltage code BOTH knock sensors.

Removing and reinstalling the intake manifold (actually all the crap on top of it) was quite a job, took several hours. I'd like to avoid doing that again if at all possible. I've been trying to figure out how these things work to know how to effectively troubleshoot them. Most of what I've read on car forums and seen on YouTube is coming from gearheads who have a basic understanding of what's going on, but still seem to be stabbing around in the dark because they don't have the proper tools. Most of them are using the $5 (or free) Harbor Freight DMM set to "AC Volts" to measure the output of the Piezo sensor which is in the kHz range. Some have said they replaced the sensor several times despite having "good" readings on the meter, and still can't get the codes to go away. I have read several times that only the factory knock sensors from the dealership should be used because the generic ones at autozone are manufactured to looser electrical tolerances which are appropriate for a wider range of vehicles. I used the ones from AutoZone; didn't know any better at the time.

The sensors have a 100kΩ resistor between the sensor wire and ground. Some gearheads say the ECU sends 5V to it, some say it sends 2.5V to it, some say the ECU doesn't send any voltage to it. My theory is that the resistor is part of a voltage divider with another 100kΩ resistor in the ECU; the ECU applies 5V and if the sensor is present, it should have 2.5V bias voltage on the input. Then the AC waveform from the piezo crystal in the sensor will be riding on top of this DC voltage. Is that correct?

If that's correct then maybe the bias voltage is just off a little bit. Maybe an external resistor either in series or parallel with the sensor could bring it back into spec and make the light go away. To be clear, I'm not proposing to "fake" the signal or bypass the sensor, as some have done by replacing the sensor with a resistor. I propose to correct the bias voltage with the sensor still in the circuit so that the ECU will see the proper AC waveform and won't get hung up on bias voltage.

I haven't even put a DMM on the truck yet; maybe I'm getting ahead of myself. Just looking for some confirmation/correction of my theory and some pointers for this weekend when I break out the o-scope and start fiddling with it.

 

I'm not ignoring you, I just don't know any better than what you know.
Get the shop manual, follow the instructions. That beats the hell out of guessing how it's wired!
Right now you don't know if the MCU is sending a voltage, the wiring harness has a problem, the ground is dirty, etc. etc.
But you already knew that.

 

I just don't know any better than what you know.

I don't know if anybody really does. I spent over an hour last night reading through a thread several pages long where a couple of seemingly intelligent guys having the same problem were going back and forth detailing their troubleshooting steps and progress. The their conversation spanned several months and neither of them ever resolved it. From there it was necroposted every few months for years with "I'm having this exact same issue. Did you ever resolve it?" There are several more threads like it across the internet.

I think I'm going to shoot some video as I go. If I stumble upon the cause and a solution, I'll post up the video and link to it on those forums.

 

Just fumbling around with some ideas...
A knock sensor must be detecting impulse energy which is stronger than the energy transmitted by normal combustion. The word, "piezo" came up in the conversation. That sounds reasonable. The knock impulses must be synchronized with the ignition timing because the combustion event is what causes the knock to happen. If I was writing software to detect knocking, I might look at a specific time during the revolution of the engine. I would want some AGC (automatic gain control) to tell the difference between normal combustion events and knock events.

So, basically, this is mostly software. If you can get a voltage spike out of a piezo sensor, the rest is distinguishing what a knock is.

 

What I gathered from the gearhead description of how it works is, it looks for "pre-detonation" ... That is, when your gasoline engine starts to behave like a diesel engine and combustion occurs before the spark plug has fired. In that scenario it advances or retards the timing to make it stop.

I am curious to see what the output of the knock sensor looks like. I will put a scope on it this weekend and find out. One of the videos I watched indicated that these knock sensors are "tuned" for certain frequencies and don't just output any 'ol wideband vibration signal, so simply smacking the engine block with a hammer might not cause the sensor to change its output. I guess that implies a filter inside it. All I saw in the video was a resistor. I wonder if I could excite one with a subwoofer tuned to around 200Hz (simulating 3,000RPM on a V8)

 

I wonder if I could excite one with a subwoofer tuned to around 200Hz (simulating 3,000RPM on a V8)

I don't think so. The knock event is a short event.
True, 200 CPS is the firing rate at 3000 RPM, but a knock is a tiny portion of a piston stroke. I think you should be looking at 2000 Hz and above.

 

I don't think so. The knock event is a short event.
True, 200 CPS is the firing rate at 3000 RPM, but a knock is a tiny portion of a piston stroke. I think you should be looking at 2000 Hz and above.

True. So I googled "frequency of engine knock" and found what I think will teach me more than I ever wanted to know about engine knock.
http://www.ti.com/lit/an/spra039/spra039.pdf

...and this is why I post here. Sometimes I just don't know the right questions to ask. Thanks for the help!

 

Thanks for the help!

How about that! I stumbled on to a reasonable estimate.

All that, "computational power" and we are still getting worse MPG than we did 30 years ago??? What's wrong with this picture?

 

How about that! I stumbled on to a reasonable estimate.

yep, kudos! and here's another good reference
http://www.diycardoctor.com/knock_sensor.htm

All that, "computational power" and we are still getting worse MPG than we did 30 years ago??? What's wrong with this picture?

My truck was getting 20+MPG before this knock sensor debacle. been getting about 16mpg since it started.
'05 was a good year I think. It has all the computerized efficiency+power stuff, and is reasonably aerodynamic. It came before the tall boxy truck fad. They keep making trucks bigger and less aerodynamic, and people keep buying them. My sister just bought a brand new Tundra and can't get more than 15mpg in it. Well no shit, it's so high off the ground (stock) that I can't even reach into the bed.
And my older trucks didn't get more than 15mpg. I had an old carbureted ford that didn't get above 9mpg.
I think the mechanical efficiency has improved greatly, but instead of holding onto those efficiency savings, they spent them on aesthetics.
I'm going to keep this '05 model until they decide to start making trucks practical again. Heard Tesla plans to produce a pickup truck. We'll have to see what that's all about; I bet an electric truck would/could have some hellacious towing torque from a stop.

 

It came before the tall boxy truck fad.

If you look its not just the trucks, cars too. Seems to me I read somewhere this is due to a new law in the EU. That hoods need to be a certain height to keep from breaking lower legs in pedestrian accidents. And car makers did it here to keep from having to design two different car platforms. One for the US and one for the EU.

 

A knock sensor is designed to pick up vibrations caused by engine knock and output a signal to the ECM. This can be caused by anything in the engine that can produce the proper frequency. A valve train that is mistimed, carbon fouling causing pre-ignition or detonation, loose components such as a flywheel or torque converter, loose rockers etc. Any noise in the engine that is not normal and can produce the right frequency, will trigger a signal. Knock sensors on the whole, are not components that get regularly replaced and have a very low failure rate. Often times misdiagnosed because of other sources of noise. You can tap on an engine and watch the signal with an oscilloscope.
As for aftermarket sensors, I would stay away from them as they are not manufactured to the same specs. Anything electrical or electronic on a vehicle, my rule has always been OEM all the way. I ma sure I will catch some flack about it but that has been my experience. Less headaches and sometimes OEM is cheaper than aftermarket. OEM vs Aftermarket left for another arguement.
Anyways, I would scope the signal and see if it is constant or random. Random suggests something mechanical going wrong in your engine. Good luck to you.Check the integrity of your wiring and connections to the ECM as well.

 

My check engine light came on a few months ago and I had the codes read at Autozone.

I can't help much with your knock sensor, but I recommend you invest in your own code reader. They're well under $20, work great, and allow you to clear the DTCs from your computer, to see if the code returns.

I can get my van to report it's trouble codes to the dash, and there's even a procedure for clearing them that I've recently learned (I think it's holding the gas pedal to the floor for 10 seconds with the engine not running, but ignition on). But the cheap little reader thing is more convenient and it can do cool things like collect data as you drive. Lots of free apps for reading and plotting the data. Some of the paid apps look very cool but so far I haven't been able to justify anything more than the free ones.

 

Here in Australia the Ford 6 cylinder engine has a Piezo knock sensor that must be fited with the correct Torque. If too much the ignition timing is retarded. If not enough too mutch advance & knocking (have seen one throw a con rod).

 

Have not yet put the scope on it (scopemeter was dead, charging it now) but I did put the DMM on it and the numbers are not as expected.

With the sensor harness disconnected, 100kΩ to ground as expected on both sensors (and on the two old ones as well).
With the sensor harness connected, 4.5kΩ to ground. So apparently no voltage divider as I predicted.
With the engine running, -90mV on the sensor wires, whether connected or not. So apparently no DC bias voltage as I predicted, or as described in my earlier link ...

^^^WRONG^^^ (for my truck anyhow, unless I've got two bad ECU inputs.)

 

Here in Australia the Ford 6 cylinder engine has a Piezo knock sensor that must be fited with the correct Torque. If too much the ignition timing is retarded. If not enough too mutch advance & knocking (have seen one throw a con rod).

I did not torque them to spec. I do not own a torque wrench.
Thankfully the truck is not actually knocking.
I feel safe driving it around, but I'm getting bad gas mileage and my annual emissions test is upcoming and guaranteed to fail.

 

My check engine light came on a few months ago and I had the codes read at Autozone. I had a "knock sensor low voltage" code P0327. My truck ('05 GMC, 5.3L) has two knock sensors but only one of them was throwing a code. I went ahead and replaced both of them, plus the harness/pigtail that goes under the intake manifold. Disconnected the battery to reset the ECU, check engine light went away, but came back within 50 miles of driving. Had the codes read again, and now I'm getting the same low voltage code BOTH knock sensors.

Removing and reinstalling the intake manifold (actually all the crap on top of it) was quite a job, took several hours. I'd like to avoid doing that again if at all possible. I've been trying to figure out how these things work to know how to effectively troubleshoot them. Most of what I've read on car forums and seen on YouTube is coming from gearheads who have a basic understanding of what's going on, but still seem to be stabbing around in the dark because they don't have the proper tools. Most of them are using the $5 (or free) Harbor Freight DMM set to "AC Volts" to measure the output of the Piezo sensor which is in the kHz range. Some have said they replaced the sensor several times despite having "good" readings on the meter, and still can't get the codes to go away. I have read several times that only the factory knock sensors from the dealership should be used because the generic ones at autozone are manufactured to looser electrical tolerances which are appropriate for a wider range of vehicles. I used the ones from AutoZone; didn't know any better at the time.

The sensors have a 100kΩ resistor between the sensor wire and ground. Some gearheads say the ECU sends 5V to it, some say it sends 2.5V to it, some say the ECU doesn't send any voltage to it. My theory is that the resistor is part of a voltage divider with another 100kΩ resistor in the ECU; the ECU applies 5V and if the sensor is present, it should have 2.5V bias voltage on the input. Then the AC waveform from the piezo crystal in the sensor will be riding on top of this DC voltage. Is that correct?

If that's correct then maybe the bias voltage is just off a little bit. Maybe an external resistor either in series or parallel with the sensor could bring it back into spec and make the light go away. To be clear, I'm not proposing to "fake" the signal or bypass the sensor, as some have done by replacing the sensor with a resistor. I propose to correct the bias voltage with the sensor still in the circuit so that the ECU will see the proper AC waveform and won't get hung up on bias voltage.

I haven't even put a DMM on the truck yet; maybe I'm getting ahead of myself. Just looking for some confirmation/correction of my theory and some pointers for this weekend when I break out the o-scope and start fiddling with it.

What kind of fuel are you using? Could you really be getting a knock?

 

What kind of fuel are you using? Could you really be getting a knock?

87 octane (regular). I think there would be a different code if I was really getting a knock. There are several code a knock sensor can throw. This one is about low voltage.

 

I do not own a torque wrench.

The auto parts stores near me have very generous tool loaner programs.

 

Ok I got a scope reading, aaand... I didn't really think this through. I really didn't know what I expected to see on the scope, or what it would mean.

I guess in order for this to mean anything, I would need to compare the output of the sensors I have installed, with the output of a known good sensor, installed in the same engine at the same time. I do have a known good sensor; the one I replaced which was still good. But in order to test it against my installed sensors, I have to take the engine apart and put it back together, then take it back apart. I could clamp one of my installed sensors in a vise alongside the known good sensor and bang the vise with a hammer, but again I would have to take the engine apart. I guess I've come to the conclusion that I'll be taking the engine apart again. Since I'm doing that, I might as well just go to the dealership and get the OEM sensors and install them. All this research and testing just to end up spitballing same as anyone else, replacing parts just to see if that fixes the problem same as anyone else, feels like I'm not being very clever.

Anyway, about this scope reading, I don't know what the spikes mean; I suspect it's the sound of 8 cylinders firing properly (no knock - I hear no audible knock) at idle. One cylinder is much louder than the others for some reason, and there's one other cylinder that is also louder but not quite as loud as the first. I suspect it is the sound of ignition because the time period between the large spikes (call it cylinder #1) corresponds to half of the ~600RPM idle (four stroke engine, each cylinder only fires every other rev). Could it be the sound of an actual knock? I don't know; I don't hear it. Could it be something else? I don't know, probably.

But in any case the sensors are picking up something, which means they aren't dead. If there is in fact not supposed to be any bias voltage from the ECU, and I'm getting a "low voltage" code, then I guess these spikes just aren't as tall as they're supposed to be. After I put the OEM sensors in maybe I will see spikes twice as high, I don't know.

 

I would sugest that the Ignition timing is locked in retard, hence poor fuel economy. This would be why there is no pinging or reasonable knock sensor output. I would be checking for continuity in the wiring loom from sensors & ECU.