Ad: VTT N54 Built Motor Knock Sensors

[Chris@VargasTurboTech]

Hey guys,

We've had these Built-Motor N54 knock sensors on our shop cars for a couple years now but mostly held them close to our vest. Enough interest has popped up that we'd like to offer them to the built motor guys searching for a method of reliable knock detection without using the current work arounds such as trim pots or desensitizing knock sensors.

A little background, once we started playing with built motors, particularly closed deck, we ran into the same false knock issues that everyone else with a built motor has had to work through. We hired a knock specialist to come in and size up what was going on, understand why the stock sensors weren't working properly anymore, and most importantly to learn what the appropriate steps would be to get us where we needed to be.

We start with factory BMW knock sensors then modify them to suit their new role protecting a built N54 without having to hassle with false knock. $799 per set, plug and play with the N54. Eliminates 95% of false knock on built motors with no changes to the factory DME.

These are the same custom config sensors @Terry@BMS is currently successfully running on his VTT/Rebello motor powered 135. I think it's safe to say Terry is pleased with how well they work but I'll let him comment if he's so inclined.

We'll add these to the website in the next day or so, none currently built up, we'll update this post (or thread) with a photo of them once we do.

Cheers,

Chris

 

[V8bait]

These work great. Hopefully others will find this useful for their builds without having to silence the sensors or trim pot them.

 

[Jake@MHD]

Woah, didn't you guys initially claim (multiple times) that your built motors didn't have false knock issues?

A knock sensor is a glorified microphone, it listens and transmits what it hears as a voltage. The DME is what then processes and uses that signal to determine knock vs. noise. For $799 did you add a digital signal processor?

Once the Spool Street knock project gets funded, everyone will be able to fine tune their stock knock sensor response exactly as needed, perfectly for their own motor, just like you can with a standalone ecu. All the same level of configuration is present: frequency per cyl, amplitude per cyl, window begin / end per cyl, noise level per cyl, etc.

 

[Chris@VargasTurboTech]

Woah, didn't you guys initially claim (multiple times) that your built motors didn't have false knock issues?

A knock sensor is a glorified microphone, it listens and transmits what it hears as a voltage. The DME is what then processes and uses that signal to determine knock vs. noise. For $799 did you add a digital signal processor?

Once the Spool Street knock project gets funded, everyone will be able to fine tune their stock knock sensor response exactly as needed, perfectly for their own motor, just like you can with a standalone ecu. All the same level of configuration is present: frequency per cyl, amplitude per cyl, window begin / end per cyl, noise level per cyl, etc.

Jake, you are 100% correct that a knock sensor is a glorified microphone. It's all about finding the correct microphone, which is looking for the correct sound signature the engine is broadcasting. There are some very important subtleties to it that your brief description doesn't capture. This isn't revolutionary stuff, you just have to ask the right questions to the right people -there are those who have made a career out of the science of detecting engine knock. If you don't like our solution, I strongly suggest you go find your own consultant and do the same work we did a few years ago -bet you come to the same conclusions.

As for claiming no false knock issues with the VTT/Rebello motors -that's also correct, since we solved them the right way with the proper knock sensors a few years ago. We haven't been using stock N54 knock sensors all this time -no radio shack pots on our car, and turning down sensitivity until they stop picking up knock (which I hope I've made it clear is not a good idea at all).

Tony never wanted to share before because he likes a competitive advantage -you know how competition is, you never show all your cards. Times change, platforms progress on and it's no longer required to keep it quiet so he's willing to share. As to the price, it is high. Each set of custom sensors requires two new sets of sensors be used to create one and neither set is inexpensive.

Cheers,
Chris

 

[Jake@MHD]

The stock sensors will also work correctly for any variety of built engine once tuned properly. The necessary tables, logging parameters, etc just aren't available at the moment.

I would argue the "right" way to fix the problem would be to properly process the incoming sound wave signal based on your individual engine's harmonics, rather than replacing the microphone. The stock DME has a dedicated DSP on chip solely for processing the signals from the knock sensors, why not take advantage?

 

[Chris@VargasTurboTech]

The stock sensors will also work correctly for any variety of built engine once tuned properly. The necessary tables, logging parameters, etc just aren't available at the moment.

I would argue the "right" way to fix the problem would be to properly process the incoming sound wave signal based on your individual engine's harmonics, rather than replacing the microphone. The stock DME has a dedicated DSP on chip solely for processing the signals from the knock sensors, why not take advantage?

Honestly you're not wrong in that you could make it work, but you're really just confirming that you don't understand how knock works, never mind how knock sensors are configured, or the differences in the "microphones" as you refer to them. BMW (and every other manufacturer in the world) go through a specific and well understood process to design the knock sensors. The "inputs" to that design process are very well understood and point you in a very specific direction. By your logic every car would use the same knock sensor and just tweak the ecu to suit. It doesn't work that way. Increasing your understanding on the subject matter will have you agreeing with me.

I'm not saying there is no value in learning about more tables in the ECU, I'm saying that's a VERY different and separate process than picking the right sensor. Until you have the right sensor bolted onto your engine, you're using the wrong tool for the job no matter what you've done to the ECU.

Kind Regards,
Chris

 

[Bnks334]

We start with factory BMW knock sensors then modify them to suit their new role protecting a built N54 without having to hassle with false knock.

We haven't been using stock N54 knock sensors all this time -no radio shack pots on our car, and turning down sensitivity until they stop picking up knock (which I hope I've made it clear is not a good idea at all).

So to clarify, you guys are just modifying the stock sensors?

 

[fmorelli]

A knock sensor is a glorified microphone, it listens and transmits what it hears as a voltage. The DME is what then processes and uses that signal to determine knock vs. noise. For $799 did you add a digital signal processor?

I used to attend an annual auto tech conference ... long time ago so my info is likely outdated ... but I recall the engineer saying that the primary noise frequency for knocke was in a very noisy engine frequency area, making it difficult to discern knock (hence false knock). He said they actually looked at the knock's second order harmonic signature, which was in a much quieter frequency. This dialog made me think about that talk. Don't know if it is done that way still, or a smarter way these days ...

Filippo

 

[Chris@VargasTurboTech]

So to clarify, you guys are just modifying the stock sensors?

I’m not sure what you’re asking -you quoted me saying we start with stock n54 sensors then modify them to suit a built motor, then asked me if we’re modifying the sensors.

Can you clarify what you need clarified?

Chris

 

[Bnks334]

I’m not sure what you’re asking -you quoted me saying we start with stock n54 sensors then modify them to suit a built motor, then asked me if we’re modifying the sensors.

Can you clarify what you need clarified?

Chris

The second part of the quote is where you are saying that you don't use N54 stock knock sensors on your built engines and then you infer they are cheap radio shack quality pots.

It makes it sound like you're offering an aftermarket knock sensors whereas it seems that you're confirming you're just modifying the stock sensors. Again, to clarify, are you just modifying the stock sensors or not?

 

[Chris@VargasTurboTech]

The second part of the quote is where you are saying that you don't use N54 stock knock sensors on your built engines and then you infer they are cheap radio shack quality pots.

It makes it sound like you're offering an aftermarket knock sensors whereas it seems that you're confirming you're just modifying the stock sensors. Again, to clarify, are you just modifying the stock sensors or not?

Oh, I see. Perhaps you’re unaware, many guys with built motors are using stock knock sensors with trim pots on them to turn down the sensitivity and try to stop registering what (they hope is) false knock. I’m not inferring or implying anything. I’m stating we aren’t using stock n54 knock sensors and using trim pots to desensitize them.

The sensor unit itself that we use is not the stock N54 unit. It is completely different.

 

[Bnks334]

The sensor unit itself that we use is not the stock N54 unit. It is completely different.

We start with factory BMW knock sensors then modify them to suit their new role protecting a built N54 without having to hassle with false knock.

This contradiction is why I am asking... so they aren't modified BMW knock sensors? they are completely different aftermarket sensors?

 

[Chris@VargasTurboTech]

Not a contradiction. We start with OEM N54 knock sensors, which are an assembly which include the sensor and the wiring harness as one assembly. We remove the (inadequate) N54 sensor itself and replace it with the correct sensor for a built N54. This is why we wrote that we use two new sets of sensors to make these. It's both a completely different sensor and a modification of the OEM assy.

Chris

 

[V8bait]

I think everybody can agree doing it all in the dme is ideal but there's a few problems with that 1) current defined tables are inadequate and 2) it ignores the sensor.

How much the sensor matters is debatable, but generally they will be tuned to the cylinder bore size and most sensitive for specific frequencies (5kHz 7kHz etc), generally there's a little room for bore size discrepancy as long as noise is similar. The computer then filters this signal, older GM cars had an actual chip that does this newer cars have a dsp dedicated to it. In GM this filter is important to block out mechanical noise related to head design (DOHC vs pushrod), block design (open vs closed) sensor location etc. The computer then interprets the final signal compared to firing events to make interventions. The only tables we currently have that I've seen are basically the computer intervention side with cutoff values for yes knock, no knock, and how much timing to pull. We don't have access to the filters/processing applied afaik.

There may be even more to it now but without more tables and better understanding of the bmw system in play it's hard to say. Most stand alone systems really water down the options you have but OEM's are pretty complicated.

 

[Terry@BMS]

Having fought built motor false knock for years, including doing some of the orignal N54 knock sensor "desensitization" testing, both on the software side, and hardware side (using isolators, resistors, pots, and the sort), I can attest to there being nothing more frustrating than false knock robbing you of 100whp. I have this Vargas solution on our Vargas build and it's working well for us. Maybe down the road there will be better or cheaper options. But this solution is working for our build.

 

[noorj]

Honestly you're not wrong in that you could make it work, but you're really just confirming that you don't understand how knock works, never mind how knock sensors are configured, or the differences in the "microphones" as you refer to them. BMW (and every other manufacturer in the world) go through a specific and well understood process to design the knock sensors. The "inputs" to that design process are very well understood and point you in a very specific direction. By your logic every car would use the same knock sensor and just tweak the ecu to suit. It doesn't work that way. Increasing your understanding on the subject matter will have you agreeing with me.

I'm not saying there is no value in learning about more tables in the ECU, I'm saying that's a VERY different and separate process than picking the right sensor. Until you have the right sensor bolted onto your engine, you're using the wrong tool for the job no matter what you've done to the ECU.

Kind Regards,
Chris

Any flat response knock sensor would work and give the same signal output in terms of acceleration g's, regardless of how it is designed or configured if it is within it's relatively wide frequency range (0-20kHz). The entire issue with "false knock" is the signal processing characteristics are no longer correct because the physical hardware characteristics have changed, so the frequency of the acceleration you're looking to isolate as knock changes.

 

[Rob09msport]

I get what everyone is saying about what's best and so on but we live in a world that's compromised of today and today we dont have access to tables necessary or any other means of safely fixing false knock so these sensors are great for someone that wants to run their built motor now. Yes when the software gets further along these may become obsolete but until then it's great. Varga may not be software geniuses but they dont rig. He made the shotgun to fix fueling using the skills he has and it works safely. Yes the best way would be designing a better pump but a smart man knows his limits. It's pretty smart to take OEM items with tons of development behind them and adapt to our needs.

 

[Chris@VargasTurboTech]

Any flat response knock sensor would work and give the same signal output in terms of acceleration g's, regardless of how it is designed or configured if it is within it's relatively wide frequency range (0-20kHz). The entire issue with "false knock" is the signal processing characteristics are no longer correct because the physical hardware characteristics have changed, so the frequency of the acceleration you're looking to isolate as knock changes.

There is basically no such thing as a flat response transducer. They pick the 'ducer, do the work on an engine dyno, set the DSP to make it work more efficiently, and go from there.

 

[Chris@VargasTurboTech]

BTW, we're not against learning more about the tables and being able to tweak them -hard to imagine how that could ever be a bad idea! I mean... it's not like you couldn't use the correct knock sensors and adjust your tune to really dial them in. What we're against is just mindlessly desensitizing sensors with no validation that knock is/is not present, and further validation that real knock is still detected. If Jake/etc is willing to do the work, I for one support it. IMO it doesn't remove the importance of using the right sensor, but who knows, maybe he can do enough trickery with DSP and tables to make the stock ones work good enough for most applications. Either way, forward progress is forward progress and generally speaking a good thing!

Chris

 

[noorj]

There is basically no such thing as a flat response transducer. They pick the 'ducer, do the work on an engine dyno, set the DSP to make it work more efficiently, and go from there.

Yes there is. A non-resonant type knock sensor still has a "range" (0-20kHz is typical in modern sensors) but they are way outside of the first second and even some third order vibration frequencies . It is not a transducer, it is an accelerometer, way different. It's also not setting the DSP to "work efficiently" it's calibrating it to detect knock based on actual measured knock of in-cylinder pressure transducers - ie to work at all.

I'm speaking from experience, I've done multiple knock sensor characterization tests for multiple manufacturers on engine testbed with in-cylinder indicating data.