GeneralDisorder Posted January 11, 2008 Share Posted January 11, 2008 91 EJ22 Turbo. Engine is from a 94 as is the sensor. Got the code 22. Immediately upon starting the engine. Inspection of the wireing revealed severe heat damage to the harness portion of the sensor wire. The last 6" or so had three or four places where the wireing had cracked to the point where the (grounded) sheilding wire was touching the inner wireing due to both layers of insulation cracking. The original engine was severely overheated, and as such the sensor wire was exposed to extreme heat from the turbo. I repaired the wire. Folded back the sheilding carefully, and soldered a new section to the inner sensor wire. I show no continuity with ground on the new sensor wire, nor on the sensor iteslf. The code 22 now only shows up after running the engine till it reaches operating temp. This seems to indicate to me that before the repair the ECU was sensing a short to ground in the knock sensor circuit due to the sensor wire being in contact with the sheilding conductor. After the repair something else is happening. Either the knock sensor has malfunctioned, or the lack of a sheilding wire is causing a different problem. The sensor is the pre-TSB sensor as it has a grey connector. It is not cracked and shows no continuity with ground. My next step is to replace the sensor, but I'm also wondering about the sheilding wire. How critical is that last 4" or 5" of sheilding? It would seem most critical where the wire runs near other wires in the harness. This last section the wire is free from the harness and on it's way to the sensor connector. Also there is no sheild after the connector for about 4" till the wire enters the sensor itself. What say ye? Replace the sensor first or repair the sheilding (somehow) and see what happens? This engine apparently ran fine when pulled a few years ago and seems to do so now as well. Could the short to ground have caused any damage to the sensor durring the few test runs that I've put it through since getting it operational? Your opinions on the subject, and most especially on the specific subject of the importance of sheilding wires for these sensors are greatly appreciated. GD Quote Link to comment Share on other sites More sharing options...
keltik Posted January 11, 2008 Share Posted January 11, 2008 Due to the nature of the knock sensor, its my opinion the lack of shielding could cause a malfunction. After all this sensors feedback is a weak one which could be affectecd by magnetic inductance from the nearby plug wires/coil. However i think its far more likely the sensor unit itself is damaged, if it was involved in the overheat then this is a very likely scenario. Either way - would be a good idea to try repairing that shield. Quote Link to comment Share on other sites More sharing options...
Hondasucks Posted January 11, 2008 Share Posted January 11, 2008 The master tech that I work with (has worked on Subarus for 33 years) told me to get one of the newer style knock sensors, that has the connector directly on the sensor itself, along with enough of the harness to tie that into the stock harness. He did it to his 92 and hasn't had any problems with it. Quote Link to comment Share on other sites More sharing options...
Cougar Posted January 11, 2008 Share Posted January 11, 2008 I tend to think the problem is with the sensor itself due to heat buildup on it. If there was a problem with noise getting into the lead then the code would set more quickly. The short lenth of exposed non-shielded wire shouldn't be a problem in my opinion. Quote Link to comment Share on other sites More sharing options...
srs_49 Posted January 11, 2008 Share Posted January 11, 2008 I would not be so quick to discount the lack of the shield on part of the cabling. How much noise/interference is coupled onto a wire (or set of wires) is highly dependent on the frequency of the noise source. In general, the higher the frequency, the smaller the opening in the shield that is needed to allow coupling. Given the relatively low signal level out of the sensor, it would not take a lot of coupled noise to cause a problem. The fact that these wires were shielded originally means that the engineers at Subaru determined that such shielding was necessary. Due to the cost difference between shielded and unshielded wires (and thair associated connectors), they probably would not have specified a shielded cable if an unshielded one would do. I'm not saying the problem isn't the knock sensor itself. Just be aware that the cabling may be part of the problem too. Quote Link to comment Share on other sites More sharing options...
Cougar Posted January 11, 2008 Share Posted January 11, 2008 I agree with what you say there SRS 49 but the fact that the sensor generates the code when the engine is warmed up makes me think the sensor is at fault. Whatever noise is being picked up is going to be there at all times. One thing that may prove the sensor is at fault is to heat it up using a hair dryer before the engine is started for the day and see if the code turns on more quickly. Quote Link to comment Share on other sites More sharing options...
OB99W Posted January 11, 2008 Share Posted January 11, 2008 Does the ECU monitor the knock sensor when the engine is cold (during "open-loop" running)? Quote Link to comment Share on other sites More sharing options...
Cougar Posted January 12, 2008 Share Posted January 12, 2008 That's a very good question OB99W. I have no clue if it does, or doesn't. I would assume though if the sensor was disconnected it would set a code immediately. Quote Link to comment Share on other sites More sharing options...
srs_49 Posted January 12, 2008 Share Posted January 12, 2008 Does the ECU monitor the knock sensor when the engine is cold (during "open-loop" running)? Don't know for sure, but my guess would be that it does. As I understand it, one (the only???) of the functions of the knock sensor is to prevent engine damage cause by pre-ignition/detonation that could be brought on by running a lower octane gas than is recommended or by somehow having the timing advanced too much. Unless there is some other mechanism for retarding the spark timing when the engine is cold, these conditions could be present when the engine is cold (if a lower octane gas was used, for instance). Quote Link to comment Share on other sites More sharing options...
GeneralDisorder Posted January 13, 2008 Author Share Posted January 13, 2008 All excelent discussion, and thank you all. My very decent local dealer parts department came through with an in-stock post-TSB sensor (they changed them at some point) for 25% below list (love those guys!). Installed tonight, torqued to 15 ft/lbs, and ran the engine for 45 minutes. Drove it a bit, and let it idle through 5 or 6 cycles of the radiator fans. No CEL. So it was afterall just a coincidence that the sensor had failed as well as the vehicle portion of the harness. I perhaps wasn't as clear in my initial post as I might have been - the sensor was original to the replacement engine - it was not exposed to the heat damage that ruined the original engine and sensor wire on the harness. The plant was swapped wholesale from a 94 turbo wagon which was "running excelent" when the engine was pulled at only 109k miles. The engine was destined originally to be a swap into something that didn't have a turbo but sat for several years till the gentleman decided he no longer wished to complete the task. That's when the engine changed hands to the PO of the new-to-me 91 SS and was installed by said PO. He was unable and eventually became unwilling to get it running so that's when I stepped in and rescued the whole mess from the crusher. I wonder what could have caused the knock sensor to fail durring the couple years that the engine was in storage? GD Quote Link to comment Share on other sites More sharing options...
Cougar Posted January 13, 2008 Share Posted January 13, 2008 Thanks for the update GD. Glad you got it fixed. Quote Link to comment Share on other sites More sharing options...
SoobGoob Posted January 13, 2008 Share Posted January 13, 2008 hey im not unwilling im just lazy, poor and need a wagon Quote Link to comment Share on other sites More sharing options...
SoobGoob Posted January 13, 2008 Share Posted January 13, 2008 plus it was only the timing, heater core, and knock sensor. just to prove i really am lazy Quote Link to comment Share on other sites More sharing options...
OB99W Posted January 13, 2008 Share Posted January 13, 2008 Does the ECU monitor the knock sensor when the engine is cold (during "open-loop" running)? Don't know for sure, but my guess would be that it does. As I understand it, one (the only???) of the functions of the knock sensor is to prevent engine damage cause by pre-ignition/detonation that could be brought on by running a lower octane gas than is recommended or by somehow having the timing advanced too much. Unless there is some other mechanism for retarding the spark timing when the engine is cold, these conditions could be present when the engine is cold (if a lower octane gas was used, for instance). I can't say it's universally true, but I know that many Subaru (and other manufacturers') ECUs default to limited ignition advance for cold start/run. Considering that detonation isn't typical on a cold engine with little spark advance, that wouldn't seem to present a problem. Based on GD's experience with the code only setting after engine warmup, and that being resolved when the knock sensor was replaced, it seems to verify that the sensor signal is ignored until warmup. However... That's a very good question OB99W. I have no clue if it does, or doesn't. I would assume though if the sensor was disconnected it would set a code immediately.I don't know if a missing knock sensor would immediately set a code, Cougar. Anyone know for sure, or want to disconnect one and report back? Quote Link to comment Share on other sites More sharing options...
GeneralDisorder Posted January 13, 2008 Author Share Posted January 13, 2008 I can't say it's universally true, but I know that many Subaru (and other manufacturers') ECUs default to limited ignition advance for cold start/run. Considering that detonation isn't typical on a cold engine with little spark advance, that wouldn't seem to present a problem. Based on GD's experience with the code only setting after engine warmup, and that being resolved when the knock sensor was replaced, it seems to verify that the sensor signal is ignored until warmup. However... The very last time I started it before I changed the sensor I got a CEL immediately. Car was completely cold - been sitting for several days. I was a bit thrown by that, so I connected the read mem connectors and it was the code 22. So I don't know what was up with that sensor.... it seemed like the code would only show up after warmup on the previous test runs, but then on that last try I got the code right away. I didn't do anything different, nor did I disconnect the battery. But I didn't disconnect it when I changed the sensor either and the code didn't come back when I restarted it. I think the sensor was just intermittant and the ECU threw the code when it saw the sensor wasn't reporting. I don't know if a missing knock sensor would immediately set a code, Cougar. Anyone know for sure, or want to disconnect one and report back? Pretty sure it will. I'll try to test it in a bit... GD Quote Link to comment Share on other sites More sharing options...
GeneralDisorder Posted January 13, 2008 Author Share Posted January 13, 2008 plus it was only the timing, heater core, and knock sensor. just to prove i really am lazy Well - as good a guy as you are, and as much as you have learned from our adventures into Subaru mechanics so far, I think you just bit off a bit more than you could chew this time around. And the expense involved was insane. With what you spent, and what I've spent so far the cost to put this thing back on the road has got to be over $5,000. The older turbo's just need more lovin, and as many folks have found out the hard way - they aren't a good choice for someone that only has one car, a limited income, and a need for reliable transportation. GD Quote Link to comment Share on other sites More sharing options...
Gloyale Posted January 13, 2008 Share Posted January 13, 2008 The older turbo's just need more lovin, and as many folks have found out the hard way - they aren't a good choice for someone that only has one car, a limited income, and a need for reliable transportation. GD Can we sticky that as possibly the best piece of advice on this board ever IMO limited income is the critical part. With enough $ you can overcome the other 2 obstactles. Quote Link to comment Share on other sites More sharing options...
GeneralDisorder Posted January 13, 2008 Author Share Posted January 13, 2008 Can we sticky that as possibly the best piece of advice on this board ever IMO limited income is the critical part. With enough $ you can overcome the other 2 obstactles. I agree with that statement to a point. The point at which you have enough money to buy a newer car..... It's a very, very fine line. At what point do you say - "gee - I could just buy something with a warantee that's faster, newer, and gets better mileage". That's assuming you only have one car of course. Me - well I'm a glutton for punishment, and I have 6 Subaru's. There's absolutely no way I'm not making it to work and my commute is only 10 miles. The expense required to properly care for these 15+ year old turbo's is pretty outragous really. The mileage the thing gets is horrible, and the parts are rare and expensive. I wouldn't have taken it on if not for the amount of money and effort already put into it by SoobGoob - it would have been a pity to see it all for nought. He put in a new low mileage engine (resealed with new HG's, etc), fuel pump, turbo-back stainless exhaust, O2 sensor, and lots of other little bits. Probably spent nearly what the car is worth. Given the amount of time and money already put toward it, it seemed like he should at least get something out of it. So I'm trading him my Legacy wagon and I'm going to DD the turbo. My commute is short so mileage isn't such a concern for me, and he really needs a wagon for skiing as he pointed out above. I'm toying with the idea of an EJ20H conversion and maybe going with my EA81 Hatch as a daily.... Choices, choices! GD Quote Link to comment Share on other sites More sharing options...
OB99W Posted January 14, 2008 Share Posted January 14, 2008 The very last time I started it before I changed the sensor I got a CEL immediately. Car was completely cold - been sitting for several days. I was a bit thrown by that, so I connected the read mem connectors and it was the code 22.[...] Since the definition for code 22 is "Knock Sensor or Circuit", it seems that code is set after a continuity check by the ECU, and wouldn't be affected by engine temp. Based on that, it wouldn't surprise me if a disconnected sensor immediately sets the same code. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.