Weird CEL 31 (knock) when cold

[iamtall77]

So the past week or so, my car has been throwing a check engine light for the knock sensor. This only happens when I leave the house on the way to work. Obviously it's not up to operating temperature, but why would I get knock at all? I'm not getting on the throttle or even getting into boost. This is light throttle at under 3k rpm. And my daily commute is in the afternoon in Vegas where outside temps are 90* and up.

I was checking my fuel trims to see if maybe I was running lean enough to cause knock and I found some weird results. My setup is mostly stock:
1990 FWD
14b
3" downpipe back (no cat)
2.5" UICP w/1G BOV
stock side mount
hacked MAS
SAFC-II

I'm running about 12% at 1k rpm in the SAFC and it goes down to about 6% at 2k rpm. My fuel trims are showing almost 115% on the low fuel trim at 1k rpm (and 02 fuel trim) and closer to 85% on the 2k rpm fuel trims. I have RPM Points set at 1k, 1.5k, 2k etc.

Why is there such a variation? I would think they would at least be closer. 30% is huge in my mind. I'll check for boost leaks as usual, but where else should I look?

[99mitsuspyder]

Have you logged it and made sure everything is where it should be?

[iamtall77]

What do you mean by where it should be?

About the only thing that is throwing up a flag right now is engine vacuum. I seem to be getting 20in/Hg at idle and more on decel. I thought 15-17" at idle was about right for stock cams. The car may have skipped a tooth or two on timing belt causing the extra vacuum and leading to a leaner condition.

However, the MAS should take care of that no matter what. It's reading the extra air going in. Unless of course I have a boost leak. I haven't had time to check it out, but will do so ASAP.

Any thoughts on what I should be looking for on the logger? I was just logging fuel trims to see where they were at while driving. They are all over the place from idle to 2500rpm like I stated before. I'll check my ignition timing to see if it is reading correctly, but I don't have a timing light right now to check my base timing. It was set correctly a couple months ago when I set the car up.

Let me know what all I should be looking for. In the mean time I'll be checking for boost leaks and looking at the static timing to make sure the belt didn't skip or anything. It does have all new timing components as of about 5k miles ago.

[iamtall77]

Well, after 2 months of this issue still being around, it hasn't gone away. However, the CEL is so consistent that I think I figured it out. The CEL comes on or near a given engine temperature (about 160*) and usually only on cruise at around 2k-2.5k rpm. The only variation on this is outside temperature. I have an air to oil cooler, and the colder it is outside, the longer it takes for the CEL to come on. Even at the same coolant temperature.

This leads me to believe it's simply knock caused by the lifters with cold/thick oil. Since it's such a specific point in the drive cycle and the knock sensor operates under a small noise signal, this makes sense to me. I know the DSM knock sensor is not that accurate and that's why it picks up lifter tick as well as true engine knock.

I haven't gotten the fuel trims figured out, but my palm is dead so I can't do anything anyway. Hope that if anyone has some phantom knock issues that are similar to mine, that it doesn't frustrate you. I don't have lifter tick that is audible and already have 3G lifters. I'm running 10W-40 Mobile 1 with some Marvels Mystery Oil. Anyone have any advice for myself or anyone in the future?

[Kibo]

How old is your knock sensor? Do you have the "blue wire mod" (or has your car ever had the "blue wire mod") to the SAFC? If so, disconnect it and verify that the wires aren't damaged, disconnected, or shorting.

The PDF version of the DSM manual that I have doesn't even list a code 31, so the following information is from the GVR-4 manual:

31 Knock sensor

Diagnostic content: Open-circuit in the knock sensor circuit.

Major cause:
(1) The knock sensor is troubled.
(2) Open-circuit, short-circuit or improper connector contact occurs in the knock sensor circuit.
(3) The engine control module is troubled.

Based on this information, the DTC is probably due to an electrical issue in the knock sensor circuit. I would suggest inspecting your knock sensor (i.e. is the 'goop' melted out the back?), check your knock sensor connector and harness, verify that your block is grounded, pull the cover on your ECU and look for evidence of damage (have the caps been replaced?). If you have access to one, try swapping in a known-good ECU.

For what it's worth: I also run 10W-40 Mobil1, have 3G lifters, no audible tick, and a 3G MAS that doesn't seem to be compensated correctly despite EPROM changes. I get partial-throttle knock that's either due to mechanical noise or the MAS compensation, but I never get a DTC for the knock sensor. I'm fairly certain you have some sort of electrical problem.

[iamtall77]

I'll check on the knock sensor itself. My conclusion came from the fact that it only throws the CEL (DTC) when the car is cold. As soon as my Palm is back up I'll log my drive to work and see if there is audible knock or if it's an electrical thing.

I do not have the blue wire mod. I eliminated that shortly after I got my SAFC years ago. I do have another ECU to swap in, but it's a 1.6L turbo unit and the fuel trims will not likely be close at all. I might do it anyway just to see. Last time I looked at the ECU (less than a year ago) it looked good on initial inspection. If I can't trace down the issue, I'll look into replacing the caps or ECU itself and see if that solves it. We'll see.

[Kibo]

When you eliminate the blue wire mod, did you make sure there isn't any exposed wire that may be shorting to something else? The knock sensor wiring is shielded, so it's possible there's something wrong with the wiring. The joy of old cars. :p