Knocking sound that goes away under load/WOT?

2012 Mazdaspeed3. Started as a faint ticking sound that evolved into a knock that also registers as KR on my accessport. It has been almost impossible to pinpoint because it does not do this at idle or under load.

If I slowly take off and gently accelerate through the rpm band it starts around 2k and increases in cadence with rpm until about 4500 where it stops. It will also keep the same cadence if I hold the engine at the same rpm.

Does it in or out of gear as well.

The weird part is that if I go wide open throttle and load the engine there is zero knock either audible or registered by the sensor at any point

Also there is no knocking sound or KR registered if I for example do a hard pull and then let off and engine brake down through the rpm band or even shift into neutral and coast until it rolls to a stop. Literally nothing.

It only occurs under low to mid acceleration input and requires acceleration input to occur.

I thought spun rod bearing as everything else I can think of has been checked but also thought that a spun bearing would just do it all the time. Am I wrong in that assumption?

Any guidance at all would be appreciated. Ive spent a dozen hours with a stethoscope trying to pinpoint and cant localize enough to say where its coming from. Kind of hard to listen for it when it almost requires the car be moving.

What is this KR?. And can I presume the “access port” is the OBDII connector?
I’m going with the noise being “rod knock” doesn’t need to be a spun bearing for that to happen. A worn bearing will do.

You could be hearing piston knock. At light loads, looseness between the piston and the bore makes a higher pitched knock than a rod. Usually gets better under load. It is also louder when cold and quieter when hot.

A loose piston pin will also knock but won’t be as affected by temperature. If tbe looseness is between the pin and piston it will get worse when hot. If is between the pin and connecting rod, it will knock all the time. Both tend to quiet under load and knock at idle.

1 Like

The accessport is OBD2 connected. Basically for programming, changing maps, flashing the ECU etc. The KR is just the knock sensor sensing knock and retarding timing (knock retard/KR).

So, if someone says worn rod bearings what will you do? Think about it? Change bearings? Sell car?

1 Like

How heavy a knock? Or a tap? High pitch sound? Or low pitch sound?

At this point are you only getting a KR? Or still an audible noise? Or both? Because what you describe might be spark knock and could be expected under those conditions.

What year is the car and how many miles on it? What engine is in the car (assume 2.3L turbo)? I assume you don’t have a check engine light, right?

I see this on Wikipedia…

L3s produced prior to 2010 are notorious for losing tension in the chain. The loose timing chain causes the timing of the engine and the VVT in particular to be improperly timed, often resulting in cylinders colliding with the valves.

This could very well be your problem. The variable valve timing may be keeping the valves away from the pistons at idle and WOT but allowing light contact to the piston tops at the lower revs. You ight want to have the cam chains checked… A boroscope into the cylinders could show contact marks of the valves on the piston tops.

1 Like

What’s the knock period or frequency? If it’s every revolution then it’s likely something on the bottom end. If it’s every other revolution then it could be hitting valves as someone said or valve knock.

Just swap them. It doesnt even require engine removal on the 2.3 MZR. Im just almost done feeling like im hunting unicorns to save the motor. I would say im a thing or two away from just giving up and changing platforms.

It is (in my opinion, havent measured it) not rapid enough to be every revolution.

Its more of a tap tap tap. Like a socket being tapped on a hollow portion of the motor like the valve cover. Lower pitch than high pitch.

Im getting the audible knock/tap sound and KR registering at the same time. The knock sensor on the 2.3 mzr is notoriously sensitive and will trigger from any or all engine noise external or internal so theres no way to know if its even genuine knock. Although I would say it is.

Being hot or cold doesnt have any affect. Someone mentioned spark knock but it’s constant across 3 sets of plugs all gapped to spec and verified to be torqued to spec. Verified clean fuel(tested Ethanol/water content) etc etc.

I cant think of any changes with spark that would be causing it outside of a bad coil pack but there arent and codes/no CEL and it runs fine outside of the problem area i mentioned.

So can that be checked by positioning the crank to TDP and then checking that the camshaft is in the correct location, or has little play?

Yes, checking the timing marks and chain tension would tell you if that is the problem. Might be a bit more invoved that removing the spark plugs and taking a look at the piston tops.

Pulling the valve cover, however, gets you a look at the valvetrain, too.

A rod knock is lower pitch. Much lower but would occur all the time.

I notice you never posted the year or miles on the engine…both bits of info might be helpful and should always be posted in the question so for the second time I ask for this info.

The Mazda Speed 3 is equipped with a 2.3 liter turbocharged engine and 6 speed manual transmission.

2012 mazdaspeed3 with 71k so about 10k past the suggested service interval for the VVT.

Also, 50k or so of that is big turbo at 22 and 27 pounds of boost on 93 and E35 respectively and tuned at the edge of injector capacity (there are no aftermarket injectors) so im more shocked that it was problem free for so long than I am that i finally have one.

Service your VVT, check your timing chain and boroscope your pistons before you blow this thing up!


I think they mean by spark knock, simple engine pinging. You might try filling the tank with high octane fuel, if that helps, good chance problem is engine pinging. If so, other possible causes

  • crank or cam position sensor problem
  • engine deposits need cleaning
  • faulty ignition system
  • faulty drivetrain computer
  • throttle position sensor

One problem with this theory , usually engine pinging is worse under load and wot. Have you experimented with this by driving up a steep hill in maybe a gear higher than optimal, and listen for what happens? It’s possible the computer’s wot algorithm compensates for pinging but not for going up a steep hill.

I had a pinging problem on my corolla one time, most noticed when going up hill as above, caused by spark plug gap got too wide.

1 Like

I tried full 93 and a drain and refill with an E40 mix(97ish octane equivalent give or take)and it did the same thing in the exact same way.

The MZR does go pig rich under WOT to pre-compensate for knock and it cant be tuned out completely even if you command it out in thr fueling tables on the map/tune you flashed. I havent tried what you said.

The problem started right after a spark plug change (as in immediately after it) so I went back to the old plugs. Still did it. New set of new plugs. Still did it. Checked gap every time. Spot on. Still did it.

Unless it was just a coincidence which is entirely possible the only other thing that was touched during the plug change would be the coil packs and the coil harness. A loose or broken wire in the coil boot or a pigtail connection in the harness could have possibly been damaged but I simply assumed that if that was the case it would be a more all the time thing instead of “here in this one spot only”.

1 Like

This started immediately after a spark plug change? And reverting to the prior plugs as a test, it still did it? … hmmm … Could be the coil idea you mention, but first step, suggest to double check all the grounds got properly reconnected. Especially grounds for the cylinder head. On my Corolla there’s an important ECM ground in that area too.