ok… bit of an update - head is back and is ok - and engine is being fitted as we speak, I am hoping to pick it up Friday to run it in on a trip to Wales next week - that will be on the old ECU, then I will go back and get the EMU put back on and a mapping session.
We have discovered a potential reason for all of my issues - there were rubber washers either side of the knock sensor - desensitising it - so if I was getting any knock, the ecu was not hearing it. I have racked my brain as to when this may have been done… the only explanation is when I had my katana fitted (Andy please check yours !!) as I did have some pinking issues and a MIL light, that were ‘mapped’ away - maybe the washers were added then…
The knock would maybe explain piston ring failure in cylinder 1 causing the compression issue and scoring… and also head gasket issues that allowed coolant into the engine - maybe I will never know - but it seems the most likely explanation, as the throttle body and charge cooling system are all ok and not causing coolant escape.
So fingers crossed I will pick it up Friday and can get back into enjoying the car again…
Pleased you have a potential cure. If the knock sensor was isolation mounted they Ecu would drive the ignition harder and harder into knock, even it it is knocking. What octane fuel did you use? Higher octane gives less knock.
That’s interesting, as I’ve been googling a lot around knock sensors on the 2zz recently which has given me a dangerous level of knowledge… so take all this with a pinch of salt.
When I was going through my EMU logs trying to find clues for this ‘VTEC clatter’ mystery I came across the knock sensor readings and I saw a lot of noise going up with RPMs on high load pulls. I knew what knock was, but not what it was supposed to look like in a log - and it did concern me at first but I queried it with RRR back and forth a few times and came to learn that the EMU can be configured to set an engine noise ‘baseline’ at varying levels of load and RPM so you can retain knock monitoring without triggering knock control methods like adding fuel or pulling timing.
Learning this did send me down a bit of a rabbit hole though and I came across a load of threads from Phil @ BOE Fabrication in the US who was at varying stages of history leading the charge on remapping the OEM EFI ECU and a limitation that he continued to call out was the fact the SC’d 2ZZ is inherently very noisy on the very knock frequencies that you’d expect from an engine of this size/bore. This lead to lots of false positives on the OEM maps which meant heavily used cars would often suffer from retarded timing for no real reason. I also found discussions around an old Lotus TSB referencing the 2-11 (I think) crowd who were using known-good race fuel with a recommendation to remove the knock sensor, wrap it in foam and tie it to the bulkhead somewhere… seriously!
Part of the issue seems to be that the knock sensor is essentially sandwiched between the SC unit and the inlet manifold, and it’s just really noisy. Through my research I got a hold of a handful of maps (or screenshots from maps!) showing how knock control was configured by various mappers on various ECUs and everyone seems to have a different approach to it - and tbh none I would consider perfect (including my own from RRR).
I won’t name names because I really don’t know what I’m talking about, and I’m in no place to ‘call out’ the pros on this but the two approaches seem to be:
Accurately check for knock, but be lenient on the knock control settings so that the performance loss isn’t significant.
Check for totally the wrong frequencies or highly desensitise the detection (via software) so that the noise is not picked up.
I’m not sure which is better or worse, but in my novice opinion the right approach would be to correctly benchmark the ‘normal’ engine noise and strictly check for anything that exceeds those parameters. I’ve been working with RRR to do exactly that on my map, but it’s probably still not perfect… A relocated or fancier knock sensor is probably the only proper option.
All that is to say, the theory about the Katana install isolating the knock sensor does make sense, if the EFI ECU really is limited in its ability to tune out background noise (at least in its ‘encrypted’ state from Lotus) then you’re left with physically desensitising it as per the 2-11 TSB. It sounds like BOE overcame these limitations in the past, but whether this was after the development of the Katana kits or not I really couldn’t say.
Armed with this info I can also look back on my car with retrospect on the OEM ECU and on many occasions I would describe the car (probably written in my thread several times) as blowing hot and cold, sometimes it felt properly fast and zingy and other times it just felt ‘kinda fast’. Perhaps this was the original ECU falsely detecting knock and pulling timing a bit.
All interesting points, thanks for taking the time to post it fonzey.
I hope ythe cause of the issue has now been resolved anyway.
Going back a few years I turbocharged a mx5 and set up all the aftermarket (link) ecu myself, getting most of the map sorted through road testing and auto tune before it went on a dyno for the final bits. The knock sensor was something I also spent some time reading about and the notes above ring a lot of bells (baseline noise etc being caused by vibrating brackets and all sorts). When I was tuning that car there were options also on how it would store retarded timing after knock events e.g permanent or only until ignition off.
At the end of the day it was something which I was always apprehensive about so I had the car tuned safely on 95ron with the mapper using det can headphones on the dyno to make sure it was a safe map, then for added protection I always ran it on 98 or 99 ron to give another safety margin. When it was on the dyno I wasn’t interested in chasing the numbers so made sure when it started getting marginal we pulled it back a little further than necessary too.
Perhaps that was over the top but it was my first foray into the world of mapping and I’d already almost blown the engine up by plumbing in the waste gate wrong at first, sending about 20psi of boost instead of 8 into the engine. It was weirdly hitting a fuel cut at 4k rpm on the first couple of test drives which I couldn’t figure out until I saw the logs. Boy was it fast to there though!
I’m not sure what RRR have done in the past, I’ve struggled to get hold of a map from another 2zz SC car. I’ve seen a H111 one but I’m not sure what I’ve been sent was the finished map (it’s still got the base map settings which seem to ignore knock completely), so would be very interested to see how yours ends up being setup, especially because (you’d think) knock control is on the forefront of their minds now!
As for my map, it was setup on the side of “strict detection, lenient control” - so it was detecting lots of “knock events” but not really doing much with that information.
What I’ve since worked on is setting up an ‘engine noise baseline’ so that the knock events are no longer detected, this means that I can setup some more aggressive knock control for if something exceeds this line.
Brown line is actual sensor readings and blue line is where I set the baseline. There are some other factors at play too such as what frequency should the sensor be listening for (this is worked out based on an equation of 1800/(Pi*bore(mm)) If I remember right, which is around 7000khz for the 2zz. Some other maps I’ve seen check for other frequency ranges, I guess that filters out the SC noise a bit - but would it still pick up knock? I really don’t know. You can then also (on the EMU) change the gain (amplification of the signal) and the time constant integrator to bring the background noise into the ‘midrange’ of the 0-5V sensor. I’ve sort of done that on mine which is why my noise levels peak around 2.5-3V.
What I’m not sure about through lack of experience is what proper knock would actually look like. With the supervision of somebody who actually knows what they’re doing I pulled a load of timing out of my map temporarily and went out to log some ‘pulls’ and the noise profile was pretty much identical which gives me confidence that my engine is not knocking, but still if I ever get a bad batch of fuel I want to make sure that the ECU will do its job and pull things back a bit. I’m still working on that through dialogue with RRR and some other experts in the field, but needless to say I’d love to see how H111 approach yours to see how near/far my settings are.
You’ve basically stolen a future update to my thread, but it seems this information is more useful here for now
There isnt really too much difference in the katana kit in raw terms. Its the same supercharger at lotus use, but with a greddy intake manifold and some shockingly bad injectors.
I am glad I went aftermarket ECU now and OEM lotus kit!
Yeah that’s about as much as I know, but “peaks and troughs” I guess can be ambiguous without context when it comes to scaling on a graph. My earlier logs before RRR and myself started finetuning the knock detection was FAR more spikey, looked very scary - we’ve smoothed that out purely by altering the knock sensor parameters a bit, nothing has changed with the engine itself - so my only remaining concern is that things have been smoothed out too much and a dodgy batch of fuel could be masked by a poorly calibrated sensor, but I need to rely on the expertise of others instead of second guessing them after a few nights on Google.
Yeah I meant more in terms of what ECU/mapping capabilities were used at that time. I believe most Katana upgrades came about along with T4 reflashes, so it’s reasonable to expect they had similar limitations as BOE did back then with regards to knock detection. Perhaps rubber washers was their approach to tuning out the white noise.
One thing for sure, Katana conversions don’t seem to have aged well - I think switching to an aftermarket ECU was absolutely the right move.
You will see anomolies in the peaks and troughs. Imagine a Windows CPU trace on perfmon at idle. Random spikes. Now 3 users RDP connect to it and all start doing some powershell. That kind of trace it what I would expect.
I am not sure that anyone actually did anything with the t4 ECU, I think it was BS TBH. I think they stuck in larger injectors and hoped …
If it’s any use, when I used to produce controllers for production cars we used to start (from key on), with a conservative knock setting and then trim towards knock and back off when it arose. After that there was a very slow dithering into and away from knock to get the best performance with environmental changes (temp, pressure, humidity etc)I don’t think we saved settings over an ignition cycle. Refuelling could cause a big change. I think fuel level (or sudden increase) was used as well, at least we talked about doing so.
That was a long time ago and there may be better strategies
But that’s the whole problem here, using the CPU analogy you’re not looking for RDP sessions on an idle server, you’re trying to spot RDP sessions on a busy SQL server that’s got an anti-virus on demand scan running rampant It’s not just the fact the 2zz with a SC is noisy, it’s the fact that the ‘normal’ noise is on a very similar frequency to the ‘bad’ noise - so picking it out on a graph is very difficult. Standalone ECU’s like the SCS/EMU/etc allow you to smooth out that sound graph to remove the ‘normal’ peaks and troughs but who’s to say that smoothing by the ECU isn’t also smoothing out bad noise when it arrives? I can’t get an answer on that which I’m happy with to be honest, and without obtaining a donor Corolla, connecting my ECU and intentionally blowing it up - I’m not sure what it would take to gain that confidence
All this is to say, I understand why the rubber isolation washers were present on the knock sensor, it’s just a physical way to achieve what a modern standalone can do within software… I’m not saying it’s the right thing to do, but there does seem to be a legitimate issue with successfully identifying knock on a SC’d 2zz. The Lotus approach is probably the most sensible one, which is to just pull timing and have done with it, even if it’s a false alarm.
I’m definitely at the point of “too much information is dangerous” now, it’s not like 2zz’s are blowing up left right and centre in Exiges, so it’s clearly not a widespread issue.
The EMU has a similar approach, in many cases you would hopefully avoid knock ever coming up because the ECU already has ignition and fueling scales depending on IATs, pressure and coolant temp stuff like that, I guess if you added a flex fuel sensor you could be even more proactive about it. IF/when knock then occurs, it pulls timing and adds fuel based on the ratio of how much noise was being generated (over the preconfigured baseline) and then it slowly adds that timing/removes the fuel over the subsequent ignition cycles, which makes any ‘pain’ from a false detection relatively short term.
According to posts I found from Phil @ BOE from a few years back, the Lotus maps take much longer to scale back - even pulling the battery doesn’t reset the trims, you needed a Lotus Scan tool to do it. Later versions of the BOE tunes apparently deal with this better as they’ve hacked their way into more and more of the EFI capability.
I agree with you that the person who fitted the washers knew the effect - but I think they were put there more as a MIL eliminator, and to fix an issue that actually has damaged the engine longer term… although not provable of course
Main thing is my car was up and running today, so we are on the path back
Awesome news, sorry for sidetracking the thread - it’s just well timed that you uncovered this as I was balls-deep into reading about this sort of stuff.
I guess when all is said and done, disabling/mitigating/desensitising a knock sensor won’t cause any harm on its own - and most owners might never come across a problem, but if this was done specifically to get around a mapping issue first time around then that’s really worrying… Hopefully you’re on the right path now, and hope it doesn’t work out too expensive!
Wow,came to this late and in the process of deciding on a katana s/c kit(used) now you have me very worried.I am not after mega power 250ish would be fine ,but now I am wondering if its all worth it with the possibility of big bills that can occur.I have the emu on my n/a which may come in handy
