imported_Cablerat

93 SI with JDM H22A

Pretty much anytime while driving my ecu throws a knock sensor code.. say 10 mins, 20 mins, 45 seconds or whatever of driving...

This Mean my engine is denonating and the knock sensor drops my timing to protect it.

I've read up on it and it seems that basically the fuel is igniting early thereby creating an odd vibration for the knock sensor to pickup. Odd vibration being the opposition of pistons firing/etc..

Apparently higher octane fuel can reduce the chances of knock because they don't ignite as easily but I don't think octane is my problem... I've tried 89, 92 and 94..

I've learnt that advance timing could cause this to occur aswell. I'm going to get a timing gun soon and check out my timing.

It's not the knock sensor itself though cuz i've replaced it with 2 others and they all did the same thing..

It's a P13 JDM ECU.

Error Code 23..

I've reset the system many times and it comes back.

It's only 1 god damn wire too. It detects vibrations and sends voltage to the ecu. I need to figure out how to test the sensor to make sure it's good..

Also it's not at any particular rpm.. like yesterday right as i hit vtec it would come on. Then today on the highway at 3300 it came on.. then 20 mins ago I shifted at 5500 thru each gear and NO CODE..

Anyone have any experience with this sensor before? what did you face and how did you fix it?

Any help would be greatly appreciated..

Thanks

imported_Cablerat

I have this posted on TPC as well. They said that if knock code is appearing its' not cuz the engine is knocking and sensor is kicking it..

They said either the sensor, wiring or ecu is defective.. But performance dies bigtime with the knock sensor code kicks in.. could be cuz the engine realizes no more knock so no more protection so kill the timing/fuel/etc to save the engine just incase?

bbarbulo

iTrader: (2)

The knock senosor (KS from now on) is a piezzo crystal that gives off voltage when it detects knock... It may actually not be detonation in the engine that's setting it off... it could be some other vibration or noise. How is your timing.... about 15 degrees or so? Back it off, change your plugs to the correct heat range, and try again. While driving have your ECU on the floor, have a buddy hold a multimeter to the ECU pins (from the back so it's still plugged in) where the knock sensor input is (refer to wiring diagram for pinout and wire color), and have him hold it till the code sets. See if the voltage spiked before it set the code. A piezzo crystal sensor is tough to break I think... it's not like it wears itself out... and if you tried other ones, and same thing happened, then it's likely not the sensor. What it could be, is if you have a break in the wire insulation, and it's picking up like an ignition signal or something.... If there is a spike in the voltage, I'd run a brand new wire from the sensor to the ECU and try again. If it persists, then your engine is knocking, which could also be (other than octane, hot plugs, and timing) carbon deposits in the combustion chamber... or it may also be a bad fuel injector that leans out or something... But you have to make sure everything is set properly before you accuse the knock sensor of being faulty...

imported_Cablerat

Lol I'm posting to myself..

I think I'm gonna start by checking my wiring continuity.. make sure i got a solid connection to ecu..

Then test sensor..

If I don't find the prob there then its most likely ECU problem. In that case i'll go throw this back in the importers face!

bbarbulo

iTrader: (2)

Are you hearing ANY knock at all?? --> read my post above

imported_Cablerat

Ok but the main thing I need to clear up is IF the code 23(KS) is thrown then it's not cuz the engine is knocking but because the knock sensor returned an abscure voltage and the ecu recognized it..

IS THAT TRUE?

SatanSRV on TPC says the knock accomodates timing without throwing a code.. he said he also read it in the Helms..

Hmmmm..

imported_Cablerat

Can't really tell if there's any knock, atleast it doesn't seem like it.. I do however have a leak in the header gasket to down pipe and it's hard to tell forsure.

I replaced plugs and wires.. they were absolutely shot.

Intake is pretty dirty too.

bbarbulo

iTrader: (2)

This is also true... it will accomodate knock TO A CERTAIN EXTENT! Remember, this is a ONE WIRE SENSOR, meaning it has NO reference voltage like most other sensors. If it were dead, there would be NO voltage, and the engine would never know if it was knocking or not. Kinda like the early one wire O2 sensors... so now... you know... the voltage being put out by the piezzo crystal is TOO HIGH, either because it's detecting severe knock, or because it picks up the wrong signal. Now, a basic question that should be obvious.... are you sure you hooked up the correct pin to the correct ECU input... cuz maybe you have like a MAP signal going to the knock sensor input on the ECU LOL... I'm sure you checked it, but I say run a wire straight from the sensor to the CORRECT ECU pin. Be certain that it's the correct pin, you don't wanna fry the ECU. Also, the engine must be in closed loop to accept signals from all the sensors, otherwise it runs on ECTS, MAP, TPS, and that's it!

imported_Cablerat

Lol.

The wiring was a pretty big mess, but I checked it over quite a few times. I have wiring diagrams all over the floor in my garage.

The connections were all soldered and shrink wrapped! But of course first I'm gonna do is test for voltage spike. I mean considering it comes on under completely different circumstances. Sometimes at like 3000 rpm and then nothing at 6000 rpm?

I tuned my TPS too recently was sittin at .41 voltage closed, should be .50!

K Thanks Barbie!

bbarbulo

iTrader: (2)

Aiight... lemme know what you find, surely it's something simple. This is why I LOVE OBD2. Plug in the laptop, read all sensor inputs in real time... real good stuff.

donesixer

I thought the P13 was OBD1...
Those things are pretty sensitive. They can point toward rod knock, piston slap, grounding issues...but try a knock sensor that you know is good first.
Personally, I would ditch the P13 for a chipped P28 and get rid of the stoopit thing. The stock sensor has no real purpose in a performance application...

imported_gatherer

ok electronics dude time...


1) a Piezzo crystal will give off a voltage when it is vibrated...

2) a Piezzo crystal will vibrate when electricity is applied to it (creating a pulse within that electricity, thats how your computer works comes down to the crystal)

3) there are 2 wires to the crystal (knock sensor) you just don't see one (since it is ground) and that wire is the casing which connects to the block and is grounded

4) when knocks occur int he engine the lnock sensor should be vibrated enough to create a voltage in reference to the ground ...

5) what your looking for with bbarbulo's test is an increase in voltage in reference to ground (black wire to ECU)

6) like donesixer stated that sensor is super sensitive to vibrations .. do you have a loud stereo, thats happened to be on every time a code is thrown (joke)

7) quickiest way to solve this is as bbarbulo stated run a fresh fully insulated wire for the knock sensor

8) have fun

blk93si

its not jdm if you don't have a check engine light

my friend just finished his swap h22 in a sedan - he had the same problem code come up -

wrong wiring -

you should be also getting a EGR - trouble code too

check the egr

imported_nibularr

i have those exact problems
what was the wiring problem??

imported_Cablerat

Muzi,

My EGR and everything is fine.. just knock sensor.

BBarbulo.. Here's the scoupe..

I tested continuity, it was fine. I was shaking the damn wire all over the place and nothing..

So then I disconnected the sensor completely from the ecu harness(removed the pin once and cut the wire out 2 inches second time).. Drove around for 20 mins and engine code came on..

Next weird thing.. and i tested this like 20 times!

Everytime I drive uphill, regardless of GEAR or RPM or SPEED.. if the upward motion is like longer than 5 seconds it will throw the code.. I mean hills ranging for like 20 - 60 degree.. I went up like 20 different hills and 95% of the time the code was thrown.. I drove around completely flat roads and nothing..

What i didn't have time to do is run a fresh insulated wire. I'll probably try that tonight or tomorrow. Working late too much!

Knock plugs in a D3 on prelude clip. Yellow-White!


Gatherer,

I don't have a radio in the car right now..

Stock 1.5" civic exhaust.

I'm starting to think it may be an ECU problem.

bbarbulo

iTrader: (2)

I wouldn't blame the ECU right away, esp considering the thing with the angles... if you can, sure... get another ECU from the importer. That would be ideal. How much clearance do you have... cuz what I was thinking, maybe going uphill, something around there contacts the chassis or whatever, and it sends a vibration through the block causing the sensor to spike. What mounts are you using?

imported_Slvr-Bullet

Just throwing this out from left field but if you think it's the ECU, can you not just unplug the KS and see if it'll still throw the code??

donesixer

Do you have any vibration when accelerating?
Intake on strut tower, exhaust, driveshafts?
Does it happen going downhill?
How much gas is in the tank? Fuel pump good? Maybe starving it a bit.

bbarbulo

iTrader: (2)

Slvr-Bullet, he tried that... code was on. I just re-read this post... ignore my last post, it makes no sense based on this above quote.... I would still double check that you have the CORRECT pinout for THAT ecu.... I'd go to Honda with the ECU and have them reproduce a diagram based on the ECU code...

imported_Cablerat

The car vibrates quite a bit inside the car as you go thru LOW rpms, I'm thinking it's cuz the mounts are tourqued too tight right now.. The code kicks in under every rpm/vibration/anything.. There's no defiinite pattern except for the uphill crap.

But all those things should not matter though cuz i disconnected the KS completely from the ECU and still throws the code.

Yeah i think I'm gonna need a honda wiring diagram.. it may be f'd..

KS is RED-BLU on sensor side.

BBarb, I think my best bet is to ensure correct ECU Pin out, and then swap ecu. I don't think he'd gimme **** for ecu swap considering if it messed It's his fault.

There's 2 single wire sensors on the back of the block.. 1 is a knock and the other i'm not sure about sittin here.. gonna have to verifty later.