I'll keep that link handy for the hose candy. Right now I'm waiting for the machine shop to fix my exhaust manifold...

 

Learned enough and going on to the second version. Here is a tease.

 

That's a turbo !

Still going to be rear mount with no IC ?

 

The rear mount idea went out the window a long time ago, but still running with no IC and loving it. More to come.

 

Are you running WMI, if so what nozzle?

What is the length of the piping and BOV spring?

 

Honkin' turbo. Compared to mine anyway.

 

Running a 55lb ev-14 fuel injector preturbo. It flows about 800ml/min water at 80% duty and 150psi. The injector is metered with mass flow at around 25%, but I have never established the perfect number. I just know that 25% works without issue and a little more or a little less doesn't seem to matter much.

The turbo outlet to throttle body is about 6" but it has some bends in it to make it work.

The bov I have is a stock one off a Saab, I think a 99. I'm setting up to play with a BOV controller.

My wastegate has a 4# spring and a 4 way solenoid. I've had it up to 8psi, but never got it very far tuned up there. Still improving hardware.

And speaking of, here is a spoiler for what's coming:

 

What was this a PCB of man? Still kicking at it I hope. Keep smashing....

 

Looks like an all inclusive version of the breadboard project. Multiple analog inputs. Wastegate signals, ion sensing, CAN bus, etc.. The middle is where the teensy would plug in. The right side of the board looks dedicated to the ignition control. Not sure what chip would go there though.

 

I'm getting back to it after being severely derailed, by work and family. The build is getting a major upgrade to the electronics before I get back on the road, but that takes dedicated time. Should have a better update in a couple days.

 

So,
basically, you would like an analog circuit that measures amplitude of 3kHz, then trig a digital output signal if amplitude is above a certain level? Or do you want an analog output proportional with 3kHz amplitude?

I might have overseen something, but this should be possible with a sharp band pass filter and some additional electronics. And we could do it with balanced signal wiring, maybe, that will cancel most common mode noise. Proper shielding will be beneficial though. I'm usable at analog electronics, so feel free to ask for advise.

 

All you need is an equalizer with the right bands and a VU meter. I used it early in the build. Or you create a 0-5v signal proportional to amplitude and then feed it to a aftermarket ECU with a spark retard input.

It's not that difficult, it works pretty well. And I still believe the stock knock detection works, if we had PIDs for the actual knock values off the ecu, including knock retard, we could tune the car easily. But that's just what I believe from what I've seen out of the response of the ECU. If someone has better info on what the ECU is doing, I'd love to hear it.

 

I had the impression that stock knock sensor was too sensitive. I think Kane described it as "useless". I think mine is very sensitive at least, so I disabled retard with Cobb(sat it to 0), and now it pull more predictable. It still run, so at least mine looks to be oversensitive for some reason.

Did you look at the signal and verified to be 3kHz? Or was it from searching? Just curious.

I'd guess that a proportional feedback would be gold for a tuner, who want to be at the edge of performance. From what I understand, tuners tend to be very at the safe side, since if they can hear the knock, it get catastrophic very easily. At least that's the attitude of my friend who's doing racing. He is well regarded locally for his builds and tuning on fully peripheral ported motor.

 

My data shows ~2600hz, Mazda literature says ~3200hz. It doesn't really matter though on the spectrograph its more of a hump than a peak, so it will be obvious anywhere in that range.

Still working on getting ion sensing based knock. But from my own testing I can say that it is not actually necessary.

I don't think the stock knock detection is necessarily too sensitive, but rather we keep getting HP gains after mild detonation starts. That may be why so many people end up tuning to engine death. The only way to be sure is to have a baseline knock reading with race gas or water injection and mild timing, that way you know what no detonation sounds like. You could also run an extra knock sensor through a equalizer and record the output for later analysis.