ITT Knock Vs. Driveline Noise on NB Knock Sensor
 

I've been experimenting a bit with timing recently with the help of a friend. We have the same setups, essentially, and he's been running a timing map for ages on that setup with no ill effects. He did, however, swap to a NA8 block at some point of time, so I pulled some timing across the board to make up for the compression differences. With that said, I present the following scenario.

The timing map I ran can be seen below. I don't think I would call it all that conservative... atleast relative to brain's basemap on the ms3x. This is running significantly more timing in places.
https://cimg1.ibsrv.net/gimg/www.mia...25394d5285.jpg

While reviewing a few logs during EBC tuning, I noticed that my knock % values were higher than they had been with brain, with the spikes being in the range in red above.

As an experiment, I played with timing. I did multiple runs with the same tune except for the area highlighted.

Run 1: As above
Run 2: -5 degrees across highlighted region
Run 3: +5 degrees across highlighted region.

From everything I've seen, knock isn't really a predictable value. The spikes are pretty intense and from what Brain said in an email to me... you'll know the spine when you see it.

Here are the spikes I saw in the runs mentioned. Logs are attached with the respective labels. I've cherry picked the best runs but did ~2-4 runs with each scenario. I only did 2 at +5 because I was scared of blowing ish up.

Run 1: ~47% (other log was ~43)
Run 2: ~27%
Run 3: ~58% (other log was ~53)

This seems fairly linear-ish. Is this just an amplification of driveline noise? Does anyone have any instances of knock that they've caught in a log with the stock NB sensor?
Would it be a reasonable assumption that true knock here would be represented by a more drastic jump in the knock % vs what is caught here? I.e. linear gains with driveline noise until there's a jump with true knock?

Ive not got any logs to hand, and I can't view yours but even with 5-10 degrees of timing pulled compared to your ignition table and 102RON fuel I see a very saw tooth looking knock percentage in a log past 5000RPM.

Interestingly I'm about to add extra timing past 6000rpm after looking at brains MS3 NB map

​​​​​​I notice his map was commanding another 3 degrees at 6500-7000rpm compared to mine, and an extra 10 degrees past 7000rpm (I run a 7500rpm Rev limit and shift at 7300 or thereabouts)

I too see a sawtooth like pattern, however the values I pointed out occur as spikes in that 'noise'.

Are my logs not opening for you? They were fine for me last night. If anyone else has issues I'll try to reupload.

I'm just not near a computer to look at your logs. I'm guessing it looks very similar to mine (DSP Knock value on my Reverant MS2)

Yours looks very very different. Can you post what your knock filter settings look like? What you see looks way more like 'true' knock than what I have going on. You're on the stock miata knock sensor, correct?

For the record, I'm running 93.

I'm attaching my tune here for reference.

There's a copy of my current tune.

I have a bunch more settings that I need to be at a computer to post a screenshot of which are part of the CAN expander on my MS2. Stock NB knock sensor. That's more or less what knock looks like when logged regardless of how much boost or timing I dial in

So, first of all... ignore the fueling. I'm still working on evening that out.

Out of curiosity, I logged those pulls in virtual dyno and this is what turned out.

This is the 'baseline'. It's the timing map posted above.

https://cimg7.ibsrv.net/gimg/www.mia...1e061e15be.jpg

The below is a run with +5 timing where a knock count of 58% was detected.


https://cimg8.ibsrv.net/gimg/www.mia...d263064519.jpg

And the last one is with timing pulled 5 degrees in the indicated cells.
https://cimg2.ibsrv.net/gimg/www.mia...a187d3ab81.jpg

So looking at this data, it looks like i might actually have been hitting knock since I increased torque in the upper rev range by pulling timing.

I'm looking at my knock sensor data: 36.8, 26, 39.6, 58. From cylinder 1->4.

I'm just trying to make sense of the knock values on the stock knock sensor. Is there any filtering that can be done on megasquirt to make more sense of it?

No wonder you're knocking. ~12 AFR at wot. You should be 11.5-11.8 there. Also Your timing map is extremely aggressive in my opinion.

Brain's basemap targets 11.8 at WOT so that's what I'm trying to target there.

To adjust VE it's just:
Current VE X (target AFR / Current AFR) X 100
Right?

Did you look at the logs for the different timing changes? The 'knock' values aren't singular spikes, they're the high values in the sea of 'noise'.
That said, at the highest timing level, there was the most background noise, so that's clearly not what I want! Like I said, I'm just trying to see if there's a way to use the stock knock sensor to actually check for noise.

The fact that the changes are mostly linear wouldn't really make sense for knock would it? But the fact that I'm seeing better torque with the lower value setting means that I'm likely past MBT.

I would start from 16 degrees at 2520 rpm tapering to 18 at all boost rows and go up from there. I would still recommend going det cans.
There is really not a lot of difference 11.8 to 11.5 re fuel, and you could add a degree or two so fuel consumption evens out and obviously your gain more power.

Out of curiosity, can you post what such a map would look like?

This is brain's basemap for reference. It's what I was using before I started looking into alternate maps.

https://cimg6.ibsrv.net/gimg/www.mia...969cebd3f5.jpg

QFT

I have a great timing map. Perfect map. Perfect.

I'm sure mr. lexy-lex

Working on it hoss.

Changing timing changes fueling a bit brah.