SR20 ECU only sending 3.3v to Knock Sensor?

[cbh148]

So my s13 SR20DET recently threw a code 34 (knock sensor) and ran like crap. Checked the wiring to the knock sensor and it seems fine -- no breaks, continuity is perfect.

Checked the voltage that the ECU is sending to the knock sensor at pin #27, and it's only 3.3 volts when I believe it's supposed to be 5 volts. What gives?

With my old knock sensor plugged in, and backprobing the harness at the ECU with the key on, that 3.3 volts is unchanged. However, with this knock sensor that I got off a friend's KA plugged in, it's 2.0 volts.

It throws the code 34 immediately with the old knock sensor. But with the used KA knock sensor, it doesn't get the code at all, but it feels slower (even though it wasn't amazing before). A lot slower. Like, can't connect the same turns drifting that it could before.

So would the 2.0 volts it's sending be close enough to the 2.5 volt signal (what it's supposed to be if there's no knock/problems) that it's NOT throwing the code 34, but it IS pulling timing and making me lose power?

Hell, for all I know, the old sensor was acting up all this time and causing power to be robbed, and it wasn't until it got fully messed up that it caused the code 34 to be thrown.

Still not sure about the 3.3 volts coming straight off the ECU though. Anybody got any insight with this? I don't mind bypassing the knock sensor with a resistor if I have to, but if it's not getting 5 volts in the first place, I wonder if I would need a different resistor to make the return signal end up at 2.5 volts -- so not the usual "1 megaohm 1 watt" resistor that everybody else seems to use.

[cbh148]

Something else I thought of -- would it be possible to get the 5 volts power from another component that's supposed to have 5 volts? Like the TPS I think (assuming mine is actually getting 5 volts like it should)?

I'd rather figure out why it's only sending 3.3 volts to the knock sensor and fix that, but after seeing a number of other people having the same exact 3.3 volt issue as me with no explanations, I'm kinda preparing myself to improvise. Any insight would be greatly appreciated!

[Luvs2slide]

Did you go through the FSM troubleshooting stuff?

[cbh148]

Yes I did, and the FSM is wrong about testing the knock sensor for resistance. EricTheCarGuy proved that in his video working on a Subaru (which uses the same knock sensor which requires 5v reference signal unplugged and 2.5v signal plugged in with no knocking). He showed how the FSM incorrectly instructs to test resistance when actually they need to be tested for voltage, because even his brand new knock sensor was "failing" the FSM's test for resistance. The manufacturer of the knock sensor told him the proper way to test them.

And by checking my ECU's reference signal to the knock sensor (as in, when the knock sensor is unplugged) I am following what the FSM says should be there, just of course finding 3.3v instead of 5.0v. And still no idea why.

Martin at RS-Enthalpy says it's pointless to even measure the knock sensor voltage because its something that a "normal" multimeter can't read. I just don't know enough about it. :(

[cbh148]

I'm seeing where this guy seemed to have a problem with his SR similar to mine, although he didn't explicitly say that his ECU was only sending 3.3volts to his knock sensor.

Surely this isn't my problem (alternator) if my ECU is sending that low voltage to the knock sensor even with the engine off, right?

http://i44.tinypic.com/10r4t1e.jpg

[cbh148]

Finally figured this out on my own, after reading countless threads about the strange 3.3 volt reference signal. Turns out that it matters whether you're using a legit multimeter or a crappy throwaway multimeter. So this wasn't my issue -- turns out my ignition timing was only 5 degrees BTDC. Silly me lol. Got it straightened out though.

http://img829.imageshack.us/img829/9460/6gc8.jpg

[supersmashboi]

hey bro did you use a timing light to check your timing? and i used a fluke meter to check my reference signal and still got 3.3v is the ecu having an issue?

[cbh148]

hey bro did you use a timing light to check your timing? and i used a fluke meter to check my reference signal and still got 3.3v is the ecu having an issue?

I did use a timing light, and I thought I was doing it correctly.
-- I unplugged the TPS when I went to set the ignition timing. Also used a spark plug wire between the #1 coilpack and the spark plug so that my friend's old timing light could get the inductive signal off that plug wire.

I tried revving the SR up to about 3500 or so 3 or 4 times to put the computer in "timing mode", but I don't know exactly what "timing mode" is or how to make for certain whether I was in/out of it.

At this point, I went to actually adjusting the CAS to set the timing. It seemed to have a little ~4° jump to it that occured every second or two, right at the exact same moment as this little hiccup sound that my SR makes when it idles. Maybe it's a little misfire? Not sure, but it is on all 4 cylinders when driving (doesn't sound like a subaru haha). (Also, my buddy's hardbody that we put an SR into doesn't have that little hiccup timing light jump at all. The crank pulley appears perfectly still when I shine the same timing light on it.)
Anyways, I put the 15° mark somewhere within the ~4° range of "jumpyness" thinking this would be ok. My SR felt a lot better after setting the ignition timing because it was retarded to begin with, however, it still doesn't feel anything like I think it should. Several friends with stock SR's feel twice as fast as mine.

My SR doesn't make power like it should, and I don't know why. I even swapped back to a stock t25g because I was running a 2871r (.64 hotside, 52 trim, aka HKS GTRS) with stock tune/injectors/MAF and was scared that this would damage my SR, but it doesn't feel ANY better at all now. Compression numbers are shown (warmed up, throttle wide open, all spark plugs removed, fuel injectors unplugged too) and they look good to me. Also, here's a pic of my CAS adjustment after I set it. It only has this one bolt holding it in because the other bolt was already snapped off into the head when I got this engine.

http://img833.imageshack.us/img833/5899/u4ft.jpg
http://img4.imageshack.us/img4/9109/bvzc.jpg

[KiLLeR2001]

Man, I would think only one bolt holding the CAS down would be enough to cause some movement.

[ultimateirving]

CBH148

I know the hiccup you speak of, my SR used to do it too. Its almost like it idles a little high, then kinda skips a beat idle drops then it goes back..
Main things i have done to correct this is:

first and foremost The system has to be completely leak free, Get a pressure tester and pressurize the system over the desired boost level. This means making sure all the vacuum lines and t-fittings you have are not seaping.

Next is IACV functionality, until i changed my iacv it was really bad and my idle would actually be all over the place, dying on decel, etcetera. Verify the proper voltage is getting to it and that it is not gunked up or frozen.

Coilpacks also are a culprit in these old motors. They often times can fire a spark that will at least let the car run, but its not always up to par, the gap of the plugs plays a part of this too. I run about .028 gap with 19psi..

Knock sensor is something that can cause issues, but normally its under load.
I have bypassed my knock sensor with the little resistor and havent looked back. Eventually when i get my new tune, i plan to disable it entirely so i dont have to worry about my bypass. As long as your tune is solid and you keep an eye on the wideband... the knock sensor can be removed.
Besides by the time it starts to detect knock it may be too late.

Try these

[cbh148]

I would think that, geometrically speaking, if the CAS is in its place, then that tight fit only allows it to twist. Then by adding a single bolt, it's not able to have any motion at all. Doesn't feel like it's got any wiggle to it. I just assumed the 2 bolts were more for backup or reassurance, but not actually crucial to have both bolts. Maybe I'm wrong here?

CBH148

I know the hiccup you speak of, my SR used to do it too. Its almost like it idles a little high, then kinda skips a beat idle drops then it goes back..
Main things i have done to correct this is:

first and foremost The system has to be completely leak free, Get a pressure tester and pressurize the system over the desired boost level. This means making sure all the vacuum lines and t-fittings you have are not seaping.

Next is IACV functionality, until i changed my iacv it was really bad and my idle would actually be all over the place, dying on decel, etcetera. Verify the proper voltage is getting to it and that it is not gunked up or frozen.

Coilpacks also are a culprit in these old motors. They often times can fire a spark that will at least let the car run, but its not always up to par, the gap of the plugs plays a part of this too. I run about .028 gap with 19psi..

Knock sensor is something that can cause issues, but normally its under load.
I have bypassed my knock sensor with the little resistor and havent looked back. Eventually when i get my new tune, i plan to disable it entirely so i dont have to worry about my bypass. As long as your tune is solid and you keep an eye on the wideband... the knock sensor can be removed.
Besides by the time it starts to detect knock it may be too late.

Try these
Thanks for the insight!

My SR's idle seems to be pretty damn good, as far as never dying or changing much. Seems to consistently sit right at 800rpm's. Doesn't require any throttle to start up, starts up quick, and never dies on its own.

The hiccup that happens every half-second or whole second is hard to describe, but it's almost like each individual hiccup is a tiny little studder. You just hear it in the exhaust note of the engine. It's almost like, in that exact moment that it studders, the rotational speed of the crank goes from 800rpm to 500rpm then right back to 800rpm -- so fast that the tach couldn't even register it. I'm guessing that's a misfire?

Either way, every single time the hiccup happens, the timing marks (when viewed with a timing light) will jump about 4° or more to the right, then come right back. It's an individual flash of the timing light (maybe once every 3 flashes?) that shows the different timing reading some 4° advanced.

Brand new NGK BKR7E plugs, gapped tight to .8mm (.032 inches).

I'm assuming my IACV is straight since the actual idle speed seems to be great and under control (again, no random dying or random high-ass 1400rpm idling like my CA18 would do).

Definitely planning on deleting my knock sensor whenever I get the chance. Seems like a lot of people hate the idea and/or are scared to do it, but then everyone that's deleted theirs and/or understands how unnecessary they are (and apparently how ineffective they are) says it's a great way to repair/prevent random drops in horsepower since the ECU will pull timing if the knock sensor is acting up, and with no code (unless it's totally fucked), then you'll never get a code since the codes on these old ECU's is more for fixing and diagnosing the circuits rather than for fixing and diagnosing actual engine problems that the sensors claim they're picking up on.

You used a 1 megaohm 1 watt resistor? And just put the resistor between pin #27 on the ECU and a ground? (such as one of the black ground wires right there in the harness by the big blue ECU plug).

[cbh148]

Also, figured I'd share this video I made. It's a bunch of clips of my boost gauge when making a 2nd gear pull and several 3rd gear pulls (all done at 2500rpm cruising, then floored it).

Seems like my buddy's stock SR (like mine) pulls twice as hard, and shoots up into boost 4 times faster on his gauge. He sent me a video of the same pull 2500rpm 3rd gear pull and it's a lot quicker.

Perhaps if my ignition timing is somehow retarded then that makes boost build slower (as well as just making less power in general)? And if so, dunno if the ignition timing is retarded from me somehow setting the CAS wrong with all the hiccuping and whatnot, or cause my replacement knock sensor from a KA is still very old and has some little cracks in it (but it doesn't cause the code like my old SR knock sensor does immediately).

mESiJwO9WfI

Youtube smoothed out the shakiness of the video and made the text captions spazz out like crazy lol.

[S14DB]

Stock ECU?
Have you hooked it up to consult? Then you could see what your knock sensor and what not are doing under load.

[cbh148]

I haven't hooked it up to consult. Never actually known how to go about doing that because I've never looked into it or what all it can tell me.