And the recent event I had with severe KR at WOT kinda proved that out. The owner saw 5.8 KR as the engine rev'ed past the "sweet spot" and none after. The engine was >/= 6k rpm when the BIG detonation event happened and I'm "almost" sure at least one rod is bent to the point of knocking.

There's more than 1 reason the stock calibration closes the throttle past 5500 RPM. One of them is that the knock sensor goes deaf shortly after.

Well despite what click says I do get it and I will be paying special attention to timing at low rpm.

FYI 06speed6 diesels don't generally premix unless it is just for reducing charge temp. Diesels inject during the power stroke, and thus have a very flat cylinder pressure curve.

So with this, does this mean that Flashes/Tunes that allow the stock throttle plates remain wide open past 5500 RPM (think CPE Throttle Flash and Piasini Flash) put those cars more at risk because the knock sensor has a harder time distinguishing knock?

Or is there another element to this I am not seeing based on what you said?

Just curious to know if those flashes are actually doing more harm.

Not really "doing harm", just making more power and as such, definitely increases the odds for a failure somewhere along the line. As all who mod for more power must know, that NOTHING comes without any risk.

The knock sensor does not work past 5700 RPM. So if you have high KR up to that point, the knock will be worse past that point since KR goes to 0.

Yep, the way I understand it, the motor is making so much noise, the sensor is really ineffective. Therefore when we see KR at WOT high rpms, LIFT for safety, OR stay with it for the WIN!

I could rev to 6800RPM in my SRT-4 with the BS chain removed and full solid mounts all around and my knock sensor would still pick up knock all the way to redline. Full stock motor at about 420WHP, I never had a single problem.

So either the DISI is an insanely noisy engine with bad vibration problems, or our knock sensor just plain sucks. Which is it?

Always FTW!!

I am figuring that its noisy. I mean who knows how it sounds on the inside, but fuck even on the outside shit is all diesel crazy/tick tick/clack clack/whoosh sounding. LOL

It could be that the frequency of knock on this motor (it all depends on the characteristics of the block etc), and the injector firing (or a harmonic thereof) is misconstrued as knock, thus the ECU ignores it.

A knock sensor is nothing more than a microphone, i find it hard to believe that it just plain sucks.

*Sigh* Direct Injection fires the injectors at ~100 VDC under a fuel pressure of ~1800+psi directly into the CC at max compression, so yeah, it makes a bit of noise too close to the frequency associated with "knock". It has nothing to do with "sucking" or overall engine vibration.

Sucking FTW! :P

What does the injection cycle have to do with the frequency associated with knock?

The injection "cycle" requires that the injector must be just a bit more "robust" than the puny little port injectors operating at maybe 60 psi. If you still don't understand the why/how an injector firing 60 psi into a vacuum/low-pressure intake runner vs an injector firing 2k psi into a cylinder under max compression pressure, then I can't explain it to you in a post. Maybe try a web search and read for a while then come back to discuss.:wave:

I just wanted to say thanks to everyone for their input, and trying to help the community better understand these engines !!

On a side note does anyone else think that mazda engineers are reading this thread, and saying 日本語教師がしっかり日本語教えます ただ今、1ヵ月間の無料体験授業受付中 While laughing - These idiots dont have a fucking clue !!

I understand that a high pressure injection of gasoline into a high compression chamber is more volatile than injecting low pressure gas into a low pressure chamber, and therefore has a higher chance of detonating without spark, but I fail to see how that has anything to do with interfering with the knock sensor's ability to pick up knock at high RPM's.

Also, if its as easy as "reading the results of a web search", please provide links to these glaringly obvious technical documents that you have access to. Otherwise, drop the patronizing routine.

I can tell you're missing my point entirely. Here's an analogy that might put it into perspective.

Port injection is like hitting a tiny fiinishing nail with a 1 oz hammer into a 1/8" thick trim panel. Direct Injection is like driving a 10 penny nail with a 2lb claw hammer through a 4x4. It takes a bit more force and makes a bit more noise.

Sorry, I know it's tough having to read and research a topic. Hey, if you have a USB port on your forehead, I'll just download and email you my brain memory files and you won't need to expend ANY effort.

Start here and follow all the links......

Search: direct injection gasoline engine basics - MetaCrawler

Ok, so I'm a bit tense today.....:burnout:

I understand your point but I don't believe its valid at all. You're trying to tell me that direct injection by itself makes more noise than an engine with no balance shaft chain and full solid engine mounts?

By your own logic, you're telling me that any DI engine in the world cannot detect knock above 5500RPM because DI causes too much interference into the frequency of the knock sensor. What about Porsches that inject at up to 2400PSI? Hell, what about all of the Cobalt SS/TC's that are modded to 400WHP+ and push over 2100PSI? Are you telling me that their knock sensors go "deaf" at 5500RPM? Also by your logic, the more PSI you're pushing, the lower RPM your knock sensor becomes worthless?

Sorry, but the logic you're trying to use is completely flawed.

I don't think it's flawed at all. Rather than the DI injection event itself being the "noise", it could instead be harmonics of those events that correlate to the knock frequency of our motors. So say at 5700rpm to 6300rpm or something like that, the injectors are opening and closing at a frequency where harmonics would readily register as knock if the sensor was actually active.

Yep, in the frequency of concern.




Okie Dokie. Suit yourself on whatever you want to believe.

Let me just expand a bit more....


A motor will typically "knock" at a very narrow frequency, and there is a lot of filtering on the ECU's side of the "knock sensor" that filters out all the other engine noise. So, while motor mounts and balance shaft deletes make the ride much more "noisy" to you as a driver, it's all filtered out by the ecu... unless of course, it put sufficient energy into the "knock frequency" that the ecu is looking at (which can definitely happen, but is generally quite rare).

But (and this is pure speculation), if the natural frequency of knock on our blocks falls into a multiple of the injector events (say at 5700rpm for instance), then they would put sufficient energy into that frequency, and register as knock. Thus from a design perspective, would be sufficient reason to simply ignore the sensor at that rpm(s).

With the explanation he gave, the logic does not compute. If he had thrown in that extra information it would have made sense, but just saying "DI makes lots of noise" does absolutely nothing to bring good factual information into consideration.

But now begs the question: how many of you actually know what the frequency of the injectors or injection process is? I have a feeling none of you do, and that you're all merely trying to make a guess, whether educated or pure speculation, as to why the knock sensors goes "dead" after 5500RPM.

I full understand what you're saying, and I'm not disputing the idea. What I am disputing is the factual information behind it, of which there is none from what I can see.

Welp, if it means anything to you... once i get my car running, investigation of the knock sensor and knock on this motor is top of my list.

And i have a few tricks up my sleeve. Until then, we can only speculate :)


You can buy (or build your own) amplifier with head phones, and actually listen to your knock sensor. Just an FYI.

It is speculation that they turn off the knock sensor due to noise although many OEMs employ this strategy. They could also turn it off in part due to the way the stock calibration drops power/boost past that point. I am fairly certain there is a parameter in the ECU that can be modified indicating until what RPM the sensor is listened to.

Until that parameter is discovered and exploited the next best thing is to devise your own amplifier as DJ stated or keep an eye on knock up to 5500RPM and tune conservatively thereafter.

And when you do decide to dial in timing beyond 5700rpm, do it on a dyno and keep an eye on your torque.

Good luck, vibration analysis is tough. Isolating why the motor would experience enough noise to KO the knock sensor would by quite the undertaking. It's probably why Mazda didn't do it in the first place.

I am sure 5400 is mega safe but, since I flat-shift at the track, I need to leave headroom for that flare. I *maybe* lose a touch not revving it out a little further but, I think I gain more back by keeping it on boost all the way down the track instead of lifting to shift. Stock turbos nosing over hard after 5500 anyway so, why rev it to 6800?

Here's some tech-line quotes concerning knock sensors. I'm still looking for additional detail to confirm, but I'm pretty sure our DISI utilizes the resonance type piezeoelectric sensor....

......There are two different styles of piezoelectric sensors -- mass and resonance. The mass sensor has its "ears" up to all vibrations. It produces a voltage at all vibration frequencies, creating its highest voltage at around 7,000Hz. This sensor forces the PCM to filter out the voltages above and below those generated at the 7,000Hz level.

The resonance sensor is much smarter. It sends a significant voltage to the PCM only when it detects vibration in the 7,000Hz range. At all other times it is quiet as a mouse.....

Lot of good information in here

I can't say one way or another whether the knock sensor "cuts out" after a certain RPM, but what I can say is the definitive way to find out is to run some extra wires off of the sensor to a logger that can take aux inputs.

Log that voltage against actual Knock Retard and we'll know for sure whether the knock sensor is registering anything past 5500RPM.

Sounds like you already have a good test plan. All we need now is for you to execute it and report back! lol

The sensor is a transducer - a microphone. It's the ECU that decides to no longer listen to it. The sensor looks like a standard bosch unit:

http://www.thor-racing.co.uk/images/..._sensor_sm.jpg

The knock sensor converts the frequency it picks up from a certain range into a voltage. The ECU then reads that voltage, determines and spits out level of knock. Like I said before, the only way to determine whether the ECU is actually ignoring that voltage input is to datalog both parameters at the same time and compare them.

If the knock sensor registers voltage above 5500RPM and knock retard is null, then you are correct in assuming the ECU ignores the sensor. If neither registers anything, then the whole "DI frequency interference" is a bunch of poorly conceived BS.

There is no other way to verify that. Everything else is still just speculation.

If you guys are patient enough, i'll do a full blown investigation. I already have all the right tools at my disposal. I just need a car.

^ I need to get me a portable scope :)

that there is a nissan knock sensor looks just like the one from my se-r. see the nissan emblem around the bolt hole:)(actually i think that is the one in my se-r)

You're correct in your test method and I understand your skepticism. It may even be warranted as far as whether the sensor stops outputting or the ECU stops listening, but a simple(?) fact(AFAIK) remains that the ECU doesn't output any KR after ~5500 rpms in ANY log or real-time AP or DH display. Based on this bit of empirical information, the deduction is made that EITHER the ECU stops listening, or the sensor output is attenuated.

I just recently experienced the situation where there was no KR shown on the AP >5500 and had an obviously severe detonation event near redline.

The other alternative is that the sensor just doesn't register knock in high RPM situations. Who is to say knock at a higher RPM doesn't generate a frequency outside of the "listening" range of the OEM sensor?

Mazda designed or selected all of the OEM parts to function properly in a stock environment. It would make sense that Mazda designed the engine to taper power from 5500RPM to redline since the turbo is simply blowing hot air and increasing the chance of knock in that range. It sounds stupid, but Mazda may not have cared about knock above 5500RPM for that specific reason - since power is tapered down on the stock tune, anyways - and decided that this specific knock sensor is "good enough".

Whatever the situation, nothing is proven until tested. I still don't believe the DI system interferes with the knock sensor, I'm more apt to believe one of the above scenarios before believing the DI conjecture.

knock frequency depends on structural characteristics of the block, not rpm.