Jump to content
Ultimate Subaru Message Board

Engine Knock?.... Oh Crap! The End Of Murphy???


Recommended Posts

Murphy, the $500 87 Turbo Wagon, may be in serious trouble! Last month, my CEL came on. It was giving the code 22 knock sensor alert. Well, when I first got the car, I had that code come up. Replaced it with one from a junk yard, and all was well.

 

That was some 6 years ago.

 

I figured this time the fix would be just as easy....

 

Um.... No.

 

I got a new one from Rock Auto this last week, and after I installed it and cleared the codes, it CEL soon came back on! Crap.

 

I thought maybe I had gotten a bad sensor (it happens), but when I pulled the sensor and tested it with my multimeter, I found it was generating milivolts in response to taps from a wrench. Crap Crap.

 

Reinstalled it, and listened more carefully, and yeah, I think I'm hearing some knocking. I tested the timing without and with the green bypass terminals connected, and sure enough, the timing is advancing when in normal mode as an activated knock sensor would do.

 

Thing that really really sucks is that the tires and both cats are brand new and less than a year old. To make matters worse, I was going to sell the 94 Legacy wagon to raise money to start a new business. That's off the table now.

 

So, here are my options. Either I shell out another $600 + for a replacement engine (w new gaskets / seals), or just give up on Murph and wash my hands of it.

Edited by Sonicfrog
Link to comment
Share on other sites

Um, if the knock sensor is detecting knock it retards the ignition timing in normal mode. Advancing is normal vs the static timing with the connectors together.

 

I don't think it can set a code for detecting too much knock, just for an electrical open or short on the circuit between the the ECU and the sensor ground. It may be that the wire or connector to the sensor is failing.

 

A knock sensor code doesn't mean the engine is detonating or that rod bearings are going, so don't panic that the engine is about to blow up because it set a code. It's an electrical issue.

 

So you get to put your multimeter to good use rather than your wrenches.

Link to comment
Share on other sites

What does it show with the connector plugged into the sensor and the sensor installed? Same resistance reading as measuring just the sensor alone? If you have an assistant tap on the sensor with a wrench with it installed do you see the same voltage at the ECU connector?

 

Make sure the ECU pins are all clean too. It could be a failing ECU too.

 

It is highly unlikely the mechanicals of the engine are setting your code. The knock you heard could be light TOD from a hydro lash adjuster or piston slap or a hundred other harmless things. Don't convince yourself it's on the way out just because it set a code.

Edited by WoodsWagon
Link to comment
Share on other sites

What does it show with the connector plugged into the sensor and the sensor installed? Same resistance reading as measuring just the sensor alone? If you have an assistant tap on the sensor with a wrench with it installed do you see the same voltage at the ECU connector?

 

Make sure the ECU pins are all clean too. It could be a failing ECU too.

 

It is highly unlikely the mechanicals of the engine are setting your code. The knock you heard could be light TOD from a hydro lash adjuster or piston slap or a hundred other harmless things. Don't convince yourself it's on the way out just because it set a code.

 

I think you forgot this is an EA82T.

Link to comment
Share on other sites

Is the assumption that any ea82t is an engine on the way out?

 

If you listen carefully to any engine with a paranoid mind you will hear noises that you can convince yourself mean DOOM!

 

A knock sensor code does not equal a knocking engine. It equals an electrical failure. I've seen engines with pounding rod bearings and no knock sensor codes or ignition retard because the sensor is not tuned to the frequency of a pounding bearing. It's tuned to the higher frequency of detonation.

Link to comment
Share on other sites

Is the assumption that any ea82t is an engine on the way out?

 

On my end, no. There are a number of USMB'ers who almost loath this engine. I'm not one of them. Treat them well, and you will get good life out of them.

 

If you listen carefully to any engine with a paranoid mind you will hear noises that you can convince yourself mean DOOM!

 

Very true, and thanks for the reminder that i need to keep that in mind.

 

A knock sensor code does not equal a knocking engine. It equals an electrical failure. I've seen engines with pounding rod bearings and no knock sensor codes or ignition retard because the sensor is not tuned to the frequency of a pounding bearing. It's tuned to the higher frequency of detonation.

 

Some newer knock sensors are built to only respond to about 7kHz, the frequency typically produced from engine knock. The ones ones on the EA82T's are not. The frequency is filtered in the ECU.

Link to comment
Share on other sites

Since you think you know the noise, does it go away when you remove plug wires one at a time. It may be the beginings of knock or your paranoia. Don't worry, We shall all watch you do it, all 37,000 plus members..

 

MUHAHAHA

 

:brow:

I think I was indeed being paranoid. I took the plug wires off and honestly didn't notice much of a difference. Probably hearing a bit of TOD.

 

I'm going to take a look inside the ECU later this week to see if maybe there is a resistor or diode that is fried. I'm quite handy with a soldering iron, but still I'll be cautious. I don't want to fry my cars brain if I can help it.

Link to comment
Share on other sites

Checked the wire this morning. The ohmmeter shows a good closed circuit from the sensor connector to the ECU on terminal 38 (white wire on that end).

 

The wire also has a shield around it. The shield can short to the center conductor, so it's a good idea to also check from the wire to ground to see if it's shorted.

Link to comment
Share on other sites

The wire also has a shield around it. The shield can short to the center conductor, so it's a good idea to also check from the wire to ground to see if it's shorted.

 

There is a spot where the wire cover looks deformed. Will check it to ground when I get a chance. have to get ready for a gig right now. (Me + Guitar = Dogs Howling)

Link to comment
Share on other sites

No short detected.

 

Did a test for milivolts across the circuit path. When I start the engine, I'm getting about 21 mv at about 1500 RPM on a cold start. When I test to ground, the reading is actually a bit lower, and about 18 mv. When I plugged the knock sensor back into the ECU system the engine revved momentarily then went back to the 1500 RPM idle.

Link to comment
Share on other sites

So, yesterday I get in the car, drive it on a good 10 mile drive, and the CEL never came on. But after I turned off the car to do some work, and then came back and started it, the CEL was back. I'm assuming that would be normal for this to not show up when the car is cold?

 

And yet, a couple of weeks ago, it did just the opposite, CEL was on when cold, and then turned off as I traveled.

 

I'm probably going to run a wire direct from the ECU to the knock sensor just for the heck of it, in case there is an intermittent short that didn't reveal itself when I tested it the first time.

Link to comment
Share on other sites

Glad you didn't replace the engine or junk it?

 

Good to hear you got it sorted out. There's hope for Murphy yet.

 

Na! That was just me in panic mode. If I didn't love this car so much, it would have been gone a long time ago. And now with the Honda Accord rear spring mod, it's ready for a trip to the mountains and a camp out one last time before the season changes.

Link to comment
Share on other sites

Nice job! One note though, I believe the shield of the cable is only connected at the ECU end, and not at all at the sensor end. If it were connected at both ends it would create a ground loop.

 

One trick I've done with the sensor is to insert a resistor in series with it. Something like a 470 ohm, but I'd have to look it up if you're curious. I was getting some false knock from my worn out 160k mile engine and this helped a lot. But in any case I didn't get a CEL before or after this mod.

Link to comment
Share on other sites

I pretty sure it wouldn't matter if you connect the shielding at one end or at both. You are connecting the shield to ground, and as long as there are no voltage differentials from either the ECU or the engine, i.e. one suffers from a poor ground, little voltage would pass through. Even then, it wouldn't make a difference.

 

The shielding is acting as a Faraday cage to keep strong inductance signals from influencing the voltage source. Because the knock sensor is supplying very low milivolt signal, inductance could corrupt the signal. The original wire is run through the wiring harness, and that wire is bundled in with all the other wires leading to the engine compartment, many of which carries stronger voltages that are potential sources of inductive electrical interference. If the new wire is run on a different path, away from the wiring harness (on the firewall for instance), there are no sources of electrical interference that would contribute to induction.

 

Still, I used a shielded wire anyway. Had it laying around, and figured what the heck! :cool:

Link to comment
Share on other sites

You know enough to be dangerous. :-p You're using "inductance" wrong; what you mean is electromagnetic.

 

If you connect the shield to ground on both sides, you now have multiple paths for ground current to flow. A ground loop. You know, that 60 hz hum you have in your home stereo. Or that "alternator whine" from the radio in your car.

 

So now when your injectors fire, your IAC is energized, etc, a portion of these currents are flowing through your new ground connection. This new ground connection just so happens it's trying to shield a millivolt signal, and in fact it's now inducing electrical noise into this signal. I'm certain the stock coax cable is not attached to ground at both ends.

Link to comment
Share on other sites

If you connect the shield to ground on both sides, you now have multiple paths for ground current to flow. A ground loop. You know, that 60 hz hum you have in your home stereo. Or that "alternator whine" from the radio in your car.

 

So now when your injectors fire, your IAC is energized, etc, a portion of these currents are flowing through your new ground connection. This new ground connection just so happens it's trying to shield a millivolt signal, and in fact it's now inducing electrical noise into this signal. I'm certain the stock coax cable is not attached to ground at both ends.

 

And yet the wire from the radio antenna is grounded at both ends. As long as the engine is well grounded (which mine is, I've made sure of it) the pulse from the injectors are going to travel the shortest distance through the nearest ground to the battery. That said, because grounds can fail over time, and you can get a fraction of the signal across any ground, I did only ground one end just in case.

 

You're using "inductance" wrong; what you mean is electromagnetic

 

Nope. Inductance is the process where an osculating magnetic field transfers an electric signal from one conductor to another nearby non-connected conductor. Step-up transformers are a wonderful example of devices using induction. You can have electromagnetism, but, if there is no conductor near enough to be effected by the electromagnetic flow, there is no inductance.

 

But now we're quibbling.... And doing this! :brow:

 

Someone+is+wrong+on+internet.png

Edited by Sonicfrog
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...