Knock Sensor Diagnosis

[Hokiefyd]

I have a fairly well cared-for 2009 Ridgeline with right at 100,000 miles. I know the original owner (my brother's father in law) and the truck's service history. The engine has a very repeatable spark knock/ping, and it occurs under certain conditions that I can easily repeat. I believe very strongly that this is spark knock and not a heat shield problem that some earlier trucks had that presented similar symptoms. (I don't think my VIN is included in the heat shield TSB anyway.)

I get a light-but-consistent knock when engine loads are above 80% (OBD-II engine load). At loads that high, the PCM removes EGR from the engine because combustion can become unstable at such a high load with inert gas. At this point, I believe it relies entirely on its preset spark mapping and the knock sensor to determine spark advance. I can watch engine load, spark advance, and EGR commanded/EGR error on my phone with Torque app and, again, I can very repeatedly call up knock by putting enough load on the engine that EGR % goes to 0% and I then begin to hear knock. So I don't believe I have an EGR problem -- the computer is actually calling for no EGR. It'd be one thing if it were calling for EGR and not getting it, but it's commanded the EGR valve closed under these conditions. So I don't think the EGR system plays into this at all (in terms of a bad EGR valve, EGR tubes being blocked, etc).

I do not have any current or stored codes in the computer. My question is, can a knock sensor get lazy and not report knock, and also not set a code at the same time? I've run a few tanks of premium fuel through it, and it doesn't seem to make a lick of difference. It will knock at higher engine loads, and immediately back off if I lift off the throttle (and EGR comes back in). I can apply more throttle, and as long as load stays low enough for EGR to persist, the knock stays away. It's almost like a "mode" or "phase" switch. I don't have a reliable way to determine if it's doing into fuel enrichment correctly. The highest fuel flow rate I've seen is about 14 gal/hr. Are my injectors actually delivering that? I don't know. After periods of heavier throttle, I do smell that "stinky catalytic converter" smell, so it seems that it is enrichening the mixture over normal.

There are few threads on the Ridgeline forum about spark knock, so it makes me think it's not "normal" and that I perhaps have a lazy knock sensor. I don't see much online about testing a knock sensor. Seems they're either "good" or they're "not". The knock sensor is buried deep inside the Vee of the engine, but it's easily accessible with the intake manifold off, so I'm thinking about changing it when I do the valve adjustment in another few months.

Any ideas are appreciated.

 

[CT8]

Try a few runs of injector cleaner . Try using a long punch and put it on the block and with the engine running tap the punch lightly Knock and watch to see if the timing moves whith the engine at operating temps and a 3000 rpms or so..

 

[Ethan1]

I think you're on the right track with the knock sensor, but how often do you go to WOT dude?

If my car had a problem above 80% engine load, I would never know it

 

[slacktide_bitog]

Do you have the P0325 code? A bad knock sensor will trigger P0325, but it will not turn on the CEL by itself.

Sometimes, you don't need to remove the intake to replace the knock sensor, if you can reach your hand in there and it isn't blocked by other stuff. But I have no experience with V6 Hondas.

 

[Grebbler]

Put that Torque app to some good use:

See if you can pull up some knock sensor and misfire data.

Whap the motor with a tire iron to simulate a knock while watching the timing graph for changes.

Watch the fuel trims under that 80% load and see if its running lean.

You might also want to find the knock sensor harness for a connector that may be loose or to check the sensor with a meter for proper OHM reading, open, or short.

 

[Hokiefyd]

Originally Posted By: Ethan1

If my car had a problem above 80% engine load, I would never know it

I bet your engine is at 100% load more often than you think.

100% load doesn't require 100% throttle. Your engine can easily be at well over 75% load with just 25% throttle under some conditions.

 

[Hokiefyd]

Originally Posted By: Grebbler
Put that Torque app to some good use:

See if you can pull up some knock sensor and misfire data.

Whap the motor with a tire iron to simulate a knock while watching the timing graph for changes.

Watch the fuel trims under that 80% load and see if its running lean.

You might also want to find the knock sensor harness for a connector that may be loose or to check the sensor with a meter for proper OHM reading, open, or short.



Thanks. Timing does not change with hammer hits to the engine, but I haven't tried it at high rpms (just at idle). ST and LT trim is excellent on this engine. LT trim stays near 0% all the time... ST trim will jump around some. I don't think I've tracked ST trim specifically during periods of knock. That's something I'll check.

Knock is unfortunately not a standard OBD PID, so it's not something I can monitor with a generic app like Torque.

 

[Brad_C]

Originally Posted By: Grebbler

Whap the motor with a tire iron to simulate a knock while watching the timing graph for changes.


It has been many years since that was an effective test for a knock sensor, unless you are capable of syncing that whap within a couple of degrees of the firing time for a cylinder. Most modern (and I mean since about 1985) ignition systems know when to expect a knock and only sense for it at or around that time.

 

[Hokiefyd]

I think I have at least a vacuum leak, here. LTFT is always very close to 0%. But STFT will go increasingly positive at hot idle in drive (such as at a stoplight). It will go as high as 20-22% sometimes. If I put it in neutral and give it some gas, STFT pretty quickly corrects back close to 0%. This engine has a MAF sensor, and I understand that in MAF sensor systems, a high STFT at idle that decreases with engine speed indicates a vacuum leak. Do I understand that correctly?

 

[Hokiefyd]

I went out and created a vacuum leak, and I'm not sure now that I really have one. At idle in Park in my garage, STFT will remain close to 0%, but may stray as high as about 7%. I sneaked the purge valve vacuum line off the intake manifold and STFT immediately jumped to 40-50% and kept climbing until I plugged it with my finger. Then it returned to close to 0%. If I let it suck ANY air passed my finger at all, it immediately responded with 40% or greater of trim. So I don't think there's a gross leak, and I don't think there's a consistent leak, because STFT often is at 0% at idle.

It is odd how it tends to creep up the longer I'm at hot idle in drive, though. Maybe that's just the purge valve working, opening at times to allow evap gasses to the system. Maybe that would cause the STFT to sometimes climb a bit.

I just need to find a Honda powertrain engineer and pick his or her brain!

 

[Grebbler]

That sounds about right.

A faulty IAC (Idle Air Control/Circuit) can cause that but usually the IAC will accumulate carbon and crud making it richer rather then sticking open and making it lean.

While a vacuum leak don't explain the symptoms of a bad knock sensor - is does shed light on why premium fuel didn't seem to help.

Next step is find the leak by taking an unlit propane torch and waving it around the usual suspects and listen for idle changes or watch the timing and trim values in Torque.

You can use anything from water to carb/brake cleaner for this step but I prefer propane for the following reasons:

Water is just messy and my GM motors have knock sensors that go bad when wet.
Spray cleaners and solvents can dissolve crud and get it sucked into a leak, plugging it temporarily, making the leak harder to find.

 

[Grebbler]

Lol. You snuck in another post while I hunted and pecked through my reply.

I'm not familiar with anything other than GM nor do I have any foreign Alldata libraries I can look at to offer any pin point suggestions.

 

[Hokiefyd]

Thanks for your tips! I do have an Alldata account and have both of my vehicles on it, but I wouldn't call the information on there particularly helpful. The IAC valve is a good idea -- I hadn't thought of that. I agree with you that you'd think it'd cause a rich condition with age, but perhaps this one is different.

The connector of my knock sensor is actually pretty easy to access. And it's just a single pin connection. My understanding is this knock sensor, with its simple piezo element inside, simply makes the connection during knock. It's just a passive device that closes to ground during a vibration in a certain resonance range. In other words, if I can monitor that with a test light or tone to ground, I should be able to see or hear that knock sensor pulsing when I hear ping, right?

My motive, here, is if I can prove to myself that the knock sensor is working, I can both a) save myself the 30 bucks, and b) accept the fact that the computer recognizes what knock is there and apparently allows that as a normal condition.

 

[SHOZ]

Originally Posted By: Hokiefyd
Originally Posted By: Ethan1

If my car had a problem above 80% engine load, I would never know it

I bet your engine is at 100% load more often than you think.

100% load doesn't require 100% throttle. Your engine can easily be at well over 75% load with just 25% throttle under some conditions.
You need to log absolute load. The basic load setting only goes to 100%.

 

[circuitsmith]

Originally Posted By: Hokiefyd
My understanding is this knock sensor, with its simple piezo element inside, simply makes the connection during knock. It's just a passive device that closes to ground during a vibration in a certain resonance range. In other words, if I can monitor that with a test light or tone to ground, I should be able to see or hear that knock sensor pulsing when I hear ping, right?

Piezo sensors put out a small voltage (too small for a test light) in response to being squeezed or vibrated.

They always have high DC resistance and read high or infinite resistance on an ohmmeter.
They are basically a microphone.
If you have a multimeter that can read in the millivolts AC range or an oscilloscope you can see pulses when tapping near the sensor.

 

[SHOZ]

And I think the knock sensor is more like a microphone and the ECU is looking for specific frequencies.

 

[Hokiefyd]

Yes, I see that I don't be able to monitor the knock sensor myself. It looks like an o-scope is about the only tool to use here (or the OEM diagnostic software).

I'm now half convinced that this is just how this engine operates. It runs out really strong, and is super smooth everywhere in the operating band. I think it's just tuned closer to the edge than our MDX's engine is (which never pings, regardless of octane). I will use some MMO in the fuel over the next few tanks and may use a dedicated upper cylinder cleaner in the intake manifold one day if I get the itch to do so. STFT at WOT doesn't point to gummed-up injectors, but I figure MMO in the fuel can't hurt. Because the knock sensor is so inexpensive, I may just replace it when I do the valve adjustment when it's easy to access.

 

[SHOZ]

Can you connect a timing light to the engine and then tap the block with a hammer?

 

[Hokiefyd]

This engine has coil on plug, and the timing light I have is old school and uses the clip for an ignition wire. In reading, I've learned that the computer knows which cylinder is knocking based on when it receives the signal from the knock sensor, so it can and may adjust only that one cylinder's timing, making a timing light or even an aggregate timing advance measurement such as available through Torque only partially helpful.

 

[Hokiefyd]

Just an update...I found what I think is a good PID to monitor in Torque: Commanded Equivalence Ratio. The CER is essentially a multiplier to stochiometric that the computer is targeting. For example, under nearly all conditions, the CER is 1.0. That is, the target A/F ratio is stochiometric (14.7:1) x CER (1.0) = 14.7:1.

During most deceleration events, when the computer is cutting fuel, the timing advance goes straight to -10 degrees BTDC (which is 10 degrees ATDC) and the CER goes to 2.0 (which means its targeted A/F ratio is 29.4:1). I don't know if 2.0 is simply as high as it will read, or if it really is targeting "stochiometric x 2.0". If it's truly cutting ALL fuel during deceleration, you'd think the CER would be "infinity". But anyway...

The CER, under very heavy throttle (close to WOT), will change from 1.0 to 0.8, which means its target A/F ratio is 11.76. I think this indicates power enrichment. And, I didn't have a lot of opportunity to test this yesterday, but I think my light pinging goes away under power enrichment. The intake roar is already loud enough then, so it's hard to tell, but I do think it goes away, or subsides significantly.

My take-away from this is the system is tuned a little on the aggressive side in terms of spark advance, and it accepts some light pinging under certain conditions knowing that, at real heavy throttle, it'll go away with power enrichment. I will have to see if our MDX's computer supports this parameter. It may not (it doesn't support the EGR parameters). If it does, I'm keen to try to see if it either doesn't give quite as much spark advance, or perhaps it applies power enrichment a little sooner than the Ridgeline does. Maybe the Ridgeline's computer is just a touch more stingy with fuel.