When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
After some logging and looking at the data I am a bit puzzled how to interpret the log labels in the Modular ecu. The car is a 13B REW w EFR83. 1 stock knock sensor on the rear housing. In below screenshot what directly catches the eye is that sometimes the knock_cyl_1 stays completely at zero and sometimes even both do. (the background map is not calibrated and just has 10 mV in it). I am unsure how this comes but this physically impossible
Also what wonders me a bit is in the below window that there seem to be 2 windows missing when plotting over 720°. This looks to be drawn for a 4 stoke piston engine with 720° firing interval an not for a rotary with 360°. If this is how it looks it means that only 50% of the cycles are considered.
the next thing is how the ecu calculates the knock_cyl_x from the label Knock_ampl (raw) or (filt). there is quite a difference and not any visible correlation between them.
I sent last week an email to adaptronic explaining how the logic could be changed to achieve a working an more robust knock detection. but I think they aren't really interested in spending any more effort on the modular range and rather invest time in 'new' stuff.
Originally Posted by email to adaptronic
For the knock logic I could give you guys some hint that normally serial ecus do not look at the peak excitation of the sensor but integrate the output of the filter during the angle window. Then the detection threshold is a ratio of the current cycle integrator end value divided by the the background noise. The background noise is a rolling average (over a calibrate able number of cycles) of the integrator end value of cycles. Off course if a cycle is detected as a knocking cycle it is excluded from the rolling average. This noise and detection threshold is doubled for each rotor/cylinder.
The integrator value is a more robust way to use the sensors output instead of looking at just the peak output. For example in a piston engine a really loud peak can occur when the piston 'slaps' or trust side. This happens usually in the knock window and can make the signal to noise ratio of the detection very poor when just looking to peak excitation and not to a longer integral. Knock never shows up as a single peak on the pressure trace. Its always a oscillating something which makes it very suitable to be integrated.
The advantage is that this is a much more robust way of determining knock. As background noise is very different from engine to engine and from operating temp (clearances etc etc), load, gearbox noise, etc. Therefore if the ecu can calculate it background noise in real time by averaging the last (lets say 8-24 cycles) instead of reading it from a map you will normally always have the real and correct noise. If you would build the logic like this and make a detection threshold of 2.5-3 I assume it will work very good already without doing any calibration. Normally you also have a limitation on the noise so that in case the knock sneaks in slow and quiet (ratio just < detection threshold) it would prevent the rolling average noise from becoming too high so that a real knock would not get detected anymore.
almost 200 views an not anyone can shine some light on this sort of magic going on in the adaptronic knock logic?
The only thing I can sort of find peace with is that the rotor on which the sensor is bolted, is the loudest. here in my case the rear rotor (cyl 2). but still it is impossible that the the front rotor would make no sensor excitation at all, and give 0 as output.
What could be is that the knock control is more advanced then explained in by adaptronic and it indeed has some rolling average of the background noise and the value one sees on the cylinder individual channels are sensor - or / by background noise. While this is better then no automatic background noise it is still clear to anyone that it is not very good to have 1 threshold and one background map for multiple cylinder/rotors as physics are that the rotor closest to the sensor will naturally have the biggest excitation of the sensor. (this is also clearly visible when looking at the scope)
@Tuned By Shawn do you have any input here or is this another unknown function of the no-manual device?
Dude makes a quality well thought out post with good data, and nobody knows the answer it seems or it doesn't exist?
Wish I could help you out man, but I have no idea how the background logic works for knock control on one of these things.
In regards to the fact you emailed them and it was clear they aren't going to spend any more effort or time on the modular units due to 'new'stuff is beyond sad. Both the modular hardware and software is flawed and tons of people have purchased this product only to learn that they were going to have to own it for years for it to live up to what they thought they bought, but now it sounds like that isn't even on the horizon and its just getting scrapped and going to leave them hanging.
Here are 3rd gear pulls, -5 0 and +4 deg offset on base timing map. Wastegate duty cycle is zero so just spring pressure, interesting to see also the different magnitude of boostcreep after 6000 ish rpm vs timing. Water injection off in these pulls. Anyone has a clue why even with +4 deg the knock on front rotor goes to zero after 4000-4500 rpm and why the values measured before look rather similar between the 0° and +4 deg pull?
I think next is swapping the sensor to the front housing and see if the trend changes or not
Here is a 3rd gear run with senor in front housing. totally blows me away the difference in measured amplitude.. Its clear now that with this if at all you can only protect one rotor. the rear rotor is now in the entire pull 0
If there would be at least different noise setting for front or rear or different thresholds.
I'm sure it's not of much help, but if your placement changed and your settings are all the same, then it's possible it was picking up motor mount and gear box frequencies and the software probably thought it was a front rotor issue giving false noise since the rear rotor is more prone to vibrations.
I was playing a bit more with the thing. Using the headset gives to me no audible combustion noise. There is a lot of background noise but no clear effect of load or timing on it.
Then I found these settings:
the top one adjust the headphone volume, but the bottom one seems to affect the reading of the knock voltages. Stranglely when putting it to -9 dB the values go completly crazy and hit 32V easily. if the value is put to +6 dB for example values for the knock voltage increase but there does still not seem to be a decent trend in terms of signal magnitude vs ignition timing
Has anyone learned any more about this? I have a stock knock sensor and I've been using a supposedly conservative tune that I've made using tons of information I read on this forum some, of the AUS forums, and a combination of the Adaptronic base Ignition table and Turblown's base ignition table I found on here. I haven't been using the knock functions because I don't really understand it and I can't find much useful information on it. That seems like it's been a mistake because it seems I've lost some compression across both rotors (which, using a regular piston compression tester, all 6 faces are fairly close around 60psi as of today, but they were around 90 psi about 500 miles ago after I finished the break in on a freshly built half bridgeport 13b-rew). I made a short run up to the mountains this past weekend and the car seemed like it was running great. I didn't push it a ton since it was raining half the time, and I was watching my computer and logging the entire time. Nothing looked out of the ordinary, everything looked within targets, and she pulled hard like a freight train. But now after checking the compression I think I may have damaged my engine. Looking through the logs, the knock readings just seem so random. I've been seeing knock on cyl 1 or 2 randomly spike up to around .10 to .18 or so even if the AFR's are below my targets (which is set to 11AFR at 15psi and up to around 14-15 degrees of timing at higher rpms). It happens fairly often during fuel cut as well... I've seen the knock values spike up to .3 or higher even after a few seconds of fuel cut with 0% throttle.
If anyone who has more experience would care to help me, I can send some logs and the tune I'm using. I have a BW 8374 running 15psi max, full fuel system, half bridgeport, Sakebomb ignition kit, 3" exhaust, and I've currently been running 4x Bur9eq plugs, but I was about to upgrade to some race plugs so I can push the engine harder when I decided to do the compression test while I had the plugs out.
Last edited by Slideways_ap3x; 08-24-21 at 11:27 PM.
What seals did you use? I had the same about 3 times with the RA super seals (3mm). I tried early timing late timing premix up to 2.5%. In the end I feel probaly the issue was too short run in. Even after the 2nd set of seals went banana the rotor housing still had the mattish appearance from new. Now i have the Ianetti I-rotary steel seals in there and they also were starting to show the effect after a few 4th gear pulls.
now I said the owner drive the car easy for a few k miles and it seems now it is fine. Before you could drop compression from 7.5 to 3.5 bar in 20 sec of WOT even if running rich which conservative timing and egt’s below 950*C. 1 bar boost w efr83 and ron98 pump fuel with quite some water inj
I had it built by Pettit Racing and it says they used 2mm 2 piece Mazda apex seals. The owner said to basically keep it under 6500 rpm and out of boost for a few 100 miles. After that I could rev it higher with light boost for another few 100 miles and then I should be good. So I basically just took it easy for a 1000 miles, I think after 500 miles I started boosting up to 5-7 psi, because I've seen lots of people recommend longer break miles. After the 1000 miles I did my 2nd oil change and removed by boost protections and started doing quick rips in 1st, 2nd, and 3rd gear. I did have over boost issues at first because my internal wastegate boost line had ripped from clearance issues with the power steering line, which I didn't know until I over boosted to 17 psi a couple of times where I set my boost protection as I was feeling out the turbo. I tried to be very careful because I know lots of people have boost creep issues with my setup so when I did overboost I was only around 4k rpm trying to see where it would stop, but I knew something was up and pulled the turbo out when I checked and saw the hose ripped. Now it stops at 15 psi so thankfully no boost creep issues.
Yep. The issue is def seal temp. Now 2 main things heat seal.
1: combustion heat. Too early or too late timing will both cause higher then needed heat input in seal/rotor
2: friction from sliding over the housing surface. This mainly depends on rpm and seal to housing friction coefficient which depends on housing/seal surface and lubrication.
i think once the seal starts to bend/wear in the middle over the leak u loose pressure and combustion becomes slow, rises egt and its like a domino effect. Also the hot gas passing the seal it more then likely gonna heat it even more
How does the seal bow or bend when it is constrained in the rotor slot with just a bit sticking up?
I believe it is the rotor housing near the spark plugs expands and sticks up and wears that part of the apex seal away as it passes over the spark plug "mountain".
Even mild detonation that isnt strong enough to break engine parts puts an extreme amount of heat into the combustion chamber burning off lubricating oils and causing localized heating of engine parts.
The RA Super seals I had worn off like 1.5mm in a coule tanks of fuel w few wot pulls. I think once the seal gets hot enough it just smears itself out. The superseals are way softer then mazda 2mm seals.
The wear on the seal is banana shaped over the entire width. Much wider then the bump around the leading plug
whats strange also on for example a s5 T2 when you run a bit too much boost on stock tuned ecu you hear the knock very well. My dad drove his car quite a bit with the hose of the wastegate off with a bnr stage 3 and stock 3 piece seals with 150k km took al the audible knock just fine.
On this car with same exhaust so basically same engine ‘noise’ I didnt hear a single knock event still the seals were gone. Therefore still I think the issue was too late spark which gives too much egt and this w soft heavy seals with not enough break in. It was running like 10.5 afr so
That's crazy... could it be related to SPM (spark plug mountain)? Maybe that could be my issue? I'm basically running the Adaptronic base map AFR table but .2 AFR richer all around and I removed around 1-2 degrees of timing everywhere. I mean it's 14 degrees of timing at 8k rpm at 15psi. I can't imagine that's too advanced or too retarded based on the timing maps I've seen other people running, plus the little bit of extra fuel I'm running aside from my MAP running in the 10's at 15psi past 7krpm because I hadn't finished leaning it out. I'm also premixing 1oz to 2 gallons plus using the OMP with the stock Adaptronic OMP table.
Even High Performance Academy made 393hp on stock ports with an 8474 on the Adaptronic base map with no problem. My engine builder said my pictures of my spark plugs show signs of detonation on the center electrode of the trailing plugs (they look a little burned back and rounded on the electrode but I swapped the leading plugs after most of my break in miles and they still look pretty normal after 600+ miles), but still I'm wondering if that's it. Maybe it's too much carbon build up wearing down the seals, or maybe my starter is running slower and I need to normalize the piston compression test readings. I just wish I knew how to read the knock values in Adaptronic. What level is just noise and what is serious knock? And how do I get the highest readings by far in the middle of fuel cut?
I think if you want to rely on the adaptronic knock detection to tune your engine all hope is lost.
It seems to me as you have nearly identical setup as the one I was tuning, with almost identical spark timing and identical apex go banana issue
This engine was just streetported and running quite a bit of water injection on 98 RON pump fuel. oiltemp never exceeded 100°C and water temp 80°. Had a big vmount setup and 80°C thermostat. rotor housing had the coolant jacket modifications done around the sparkplugs.
12-17-19, 03:55 PM
