We haven’t really evaluated the different knock sensors.

A good generic one is the Bosch donut one that you find on aftermarket knock kits and Subarus.

Knock sensors won’t save your engine, especially a rotary. The most useful thing they can do is use them for listening during tuning.

Otherwise to wire it up, it’s just follow the diagram. If it’s a 1-wire sensor like the FD RX7, then the 1 wire connects to the knock input on the ECU.
If it’s a 2 wire, then it’s just like on the 440 / 420 diagram, which is that the signal wire goes to the ECU input (pin 2J).

Then for the setup, they need to:

1. retard the timing so it doesn’t knock
2. set all the knock background noise levels to zero (under analogue calibrations)
3. log some power runs to find the knock levels
4. put those knock levels (a bit higher) into the background knock levels

I hope this helps!

 

Thanks for the clarification. Makes life a little simpler!

 

Is there a way to wire in 2 knock sensors Elliot? Say if you ran 2 RX-8 sensors, one in each housing. This would be for a 440 universal application.

Thanks

 

I think you can only do one on a the 420/440 IIRC. I will have to ask Andy about that.

I wouldn't worry too much about it since you're on E85

 

Using the Bosch broadband sensors, like the Subaru, is a bad idea with the 420/440. The broadband sensors have no built-in signal filtering. Any noise produces an output. The 420/440 also does not have the ability to filter one frequency from the next. So any noise can be seen as knock. Ideally, you would want to have the ability to band pass filter the input from the sensor to only look for the specific knock frequency of the engine you're working with. Since the 420/440 will not do this, you'll want to use a sensor that has built-in filtering. The stock Mazda sensors have this filtering. Since engines of different configurations have different knock frequencies, using a sensor with built-in filtering from a different type of engine is not recommended.

In short, use the Mazda sensor or don't bother.

 

Thanks, I was wondering about that as well. Since our engines do not have valvetrain, there will definitely be something specific to our rotating mass and such.

 

I saw on one of our wiring diagrams (1JZ) where two knock sensors signals are connected to the single knock input of the e420d/e440d. I'll also ask Andy about that.

 

It really has nothing to do with the rotating mass, valve train, or lack thereof. Every engine configuration "rings" at a specific frequency. For a piston engine, this is determined by the diameter of the cylinder bore. The issue the ECU needs to be able to overcome is that other things create noise in the same frequency range. So, even if you band pass the frequency and cut out everything except the specific knock frequency, you can still have false knock from a similar noise created somewhere else. This is where knock windowing becomes important. Where the ECU looks only for knock during the window of crank rotation where knock is likely to occur. Noise outside of that window is ignored because it can't be actual detonation.

Without proper filtering and windowing, listening for noise is of very limited value IMO.

 

i use two of the new style Bosch knock sensors from the RX8 which i assume has approx the same frequency as the BREW. 3500 cps. my ECU does offer window setting.

i consider knock to be one of the keys to proper tuning and use the Link Knock headphones.

howard

 

^^^ What Mr Ludwig said, regarding the frequency selection.

Just to clarify, the 420 / 440 do have a bandpass filter but it's quite broad.

The 420 / 440 also have an algorithm to detect standing signals and subtract them, so the 420 / 440 looks for peaks above the average. So if you have a constant level (just the background sound of the rotating engine), the knock reading will show zero, but peaks above that will trigger the detector. Obviously this works a LOT better if your sensor is more frequency selective.

But yes, if you can get the standard sensor for your engine (I'd expect the 12A would ring at a different frequency from the 13B, because it has a different "bore") that would be the best.