BlownCorona Posted January 29, 2019 Share Posted January 29, 2019 On 23/01/2019 at 20:34, ajg193 said: Chinese sensor is still working perfectly. As verified by your Chinese sensor 5 Quote Link to comment Share on other sites More sharing options...
shrike Posted January 29, 2019 Share Posted January 29, 2019 Ordered some more parts for my Speedunio (have sold the car it was going into however) also got myself an 8x8 3.5 Teensy setup for sub 300AUD so now have two budget ecus for future projects Going to Japan on Thursday and once I get back will be putting money into my project car thats still in NZ. Will look around for a cheap hack to play with as well haha 2 Quote Link to comment Share on other sites More sharing options...
ajg193 Posted January 29, 2019 Share Posted January 29, 2019 11 hours ago, BlownCorona said: As verified by your Chinese sensor Pretty much, I wouldn't use it on any high performance vehicle. But it seems to read the same as the Bosch sensor. The tune map hasn't changed. Spark plugs agree with the O2 reading. I also have spare bosch sensors (brand new and slightly used) to put back in the car the moment the chinese one stops working (I give it about 2 months..) Quote Link to comment Share on other sites More sharing options...
Roman Posted January 30, 2019 Share Posted January 30, 2019 Not using Speeduino but I've got a spare Teensy 3.6 here that I've mounted the audio processing board on. Will fit my spare canbus backpack on it. Will be able to use the audio processing for some knock detection, and hopefully some ion sensing with a burst reading on the ADC on a single cylinder just past an ignition event. Hoping there's an rpm/load range where there's something interesting from the ion sensing, at the resolution and frequency I'm able to measure it. one thing on the Link G4+ I've thought is a bit so-so is its knock detection. I think its silly to detect knock, pull timing, and then just slowly creep the value back up until it knocks again then do same thing over and over. Like it never builds up a "smart" table it just does the same dumb thing over and over. IT might knock on the same load/rpm 1500 times and it will never change its behaviour. Even if you never turn the engine off (it doesnt store the table) the table it generates even if you ran the engine for say 10 hours doesnt actually show you anything useful as it might be half way through its pull knock / advance timing again in each cell. So I want to make a long term trim table for knock like OEM stuff does. Every time it registers short term knock it will take say 0.5 degrees off the long term table. Then the next time it gets to that same region it is already 0.5 degrees less total timing instead of just banging its head against the wall over and over. 3 Quote Link to comment Share on other sites More sharing options...
NickJ Posted January 30, 2019 Share Posted January 30, 2019 How would a long term table react to a 'bad' tank of gas? would manual reset be needed? I've pondered the same thing but never had to implement it 1 Quote Link to comment Share on other sites More sharing options...
Roman Posted January 30, 2019 Share Posted January 30, 2019 Yeah well another annoyance is things like cutting over ripple strips at trackday can trigger knock sensor. Audio detection of knock is a bit ghetto really which is why I'm really hoping I can see ion sensing data with some decent resolution. My understanding is that some modern BMWs have a knock detection system that uses just ion sensing. Some preliminary looky-loos with a scope have shown what looks to be indication of knock on the coilpack return signals. But useless if you can only see it on a scope and not in a device which can process it! So just need to make some more progress on hooking it all up. Quote Link to comment Share on other sites More sharing options...
Roman Posted January 30, 2019 Share Posted January 30, 2019 It's also a bit of a moot point as I can run 50 deg timing well past MBT and not get any knock during cruise conditions. God bless lean burn haha Quote Link to comment Share on other sites More sharing options...
kpr Posted January 30, 2019 Share Posted January 30, 2019 Have never bothered with ecu knock detection for mostly the same reasons. Would be nice to have something that works properly, rather than ignition timing on the conservative side and putting your engines life in the hands of your petrol company of choice. but in saying that, if an engine is tuned that close to the edge, maybe pump gas isn't for you. or shouldn't be lending your car to some monkey who fills it up with 91 @Roman What does the rpm signal look like on your link? and on timing light? did toyota step up there game on the beams triggers? Also wouldn't think you would have any knock issues on your engine? Quote Link to comment Share on other sites More sharing options...
Yowzer Posted January 30, 2019 Share Posted January 30, 2019 11 hours ago, ajg193 said: Spark plugs agree with the O2 reading. Amazing that you can read 0.1 lamda resolution on the fly from a spark plug 3 Quote Link to comment Share on other sites More sharing options...
ajg193 Posted January 30, 2019 Share Posted January 30, 2019 19 minutes ago, Yowzer said: Amazing that you can read 0.1 lamda resolution on the fly from a spark plug You know it. One day next month I will throw the bosch sensor in and compare the values using datalogs. Quote Link to comment Share on other sites More sharing options...
Roman Posted January 30, 2019 Share Posted January 30, 2019 3 hours ago, kpr said: Have never bothered with ecu knock detection for mostly the same reasons. Would be nice to have something that works properly, rather than ignition timing on the conservative side and putting your engines life in the hands of your petrol company of choice. but in saying that, if an engine is tuned that close to the edge, maybe pump gas isn't for you. or shouldn't be lending your car to some monkey who fills it up with 91 @Roman What does the rpm signal look like on your link? and on timing light? did toyota step up there game on the beams triggers? Also wouldn't think you would have any knock issues on your engine? When I went to the dyno I found absolutely nothing adding a whole bunch of timing all through the rpm range, and not a single blip of knock. When I had staged injection running with a blocked injector it just wheezed but nothing blew up. This motor has put up with heaps of punishment haha. Youre right there are no knock issues on my engine with 98 so I might need to run on some 91 to torture test it. Triggers and RPM are nice and stable! 1 Quote Link to comment Share on other sites More sharing options...
h4nd Posted February 11, 2019 Share Posted February 11, 2019 On 30/01/2019 at 16:02, Roman said: ....indication of knock on the coilpack return signals. But useless if you can only see it on a scope and not in a device which can process it! So just need to make some more progress on hooking it all up. Plumb 'scope into dash, and feed thence to ECU? Of anyone, you could 3d-print a surround! Quote Link to comment Share on other sites More sharing options...
Roman Posted February 11, 2019 Share Posted February 11, 2019 Hahahaha looking at some sweet sweet waveforms while driving. The dream 3 Quote Link to comment Share on other sites More sharing options...
bigfoot Posted February 11, 2019 Share Posted February 11, 2019 On 09/12/2018 at 21:35, bigfoot said: I have a link atom 2 and the wiring is doing my head in. Is somebody in auckland able to help me finish it/finish it for me. It is the last thing to do to have my car running. @Roman? @Stu Car is running and working on making it drivable. Huge thanks to @Roman answering all my stupid questions. Ran the first log tonight, hopefully the changes I made work 6 1 Quote Link to comment Share on other sites More sharing options...
Roman Posted February 11, 2019 Share Posted February 11, 2019 Great stuff! Now the fun part begins Quote Link to comment Share on other sites More sharing options...
Roman Posted March 11, 2019 Share Posted March 11, 2019 Sooooooooo my favourite mystery topic. Part throttle Variable cam tuning. I did some tests tonight to try find out what cam angle might be "best" at a given throttle angle. By best I mean max airflow as read by the MAF. I used E-throttle to limit throttle angle to 30% max and did some pulls through the rev range, logging the MAF and MAP values as I increment through the intake cam angles in 5 degree steps. As expected, both the MAF and MAP values change as you advance the cam. But what was interesting is that there is loosely a correlation between achieving the strongest vacuum, and highest MAF reading. At 4000rpm with 0 degrees cam advance we are losing 11kpa and have 0.406 grams per cyl. At 45 degrees advance we are losing 16kpa and have 0.488 grams per cylinder of airflow. So 45% more vaccum and 20% more airflow. At 5000rpm with 0 degrees cam advance we are losing 20kpa and have 0.416 grams per cylinder. With 35 degrees cam advance we are losing 24kpa and have 0.456 grams per cylinder. 20% more vacuum and 9% more air. So whats interesting about this is you dont always have a MAF available for cam angle tuning but it loosely correlates to MAP. As a general rule from my testing it seems if you can get 10% more vacuum you'll get 5% more airflow. Similar to ignition timing though it looks like it ramps up quickly to best values then plateaus over a cam timing spread. I'll post something better later on once I've had a chance at figuring out how to format this mess of data. But seems like there's a fairly clear correlation which makes sense, if you are improving VE via cam timing then you are increasing the air demand... which the throttle restricts hence extra vacuum. I'll do some more tests at lower throttle with higher vacuum too and see if I'm getting similar results. Something else thats really interesting is that the best cam angle settings even at this low throttle angle are much higher than I would have expected. Based on absolutely no evidence I've just been slowly ramping the timing down towards zero throttle. It looks like past maybe 10-15% throttle it's just going to be the cam in a position very close to WOT for max airflow. It will be interesting to see if we see the same trends with much stronger vacuum. Will do some more tests through the week! 4 Quote Link to comment Share on other sites More sharing options...
kpr Posted March 11, 2019 Share Posted March 11, 2019 guessing it depends what you are shooting for? economy or power or just science? will assume science. but if economy. would it not be better to go the other way to a point, less advance and higher throttle opening to get the same airflow. which would reduce pumping loss? 1 1 Quote Link to comment Share on other sites More sharing options...
ajg193 Posted March 11, 2019 Share Posted March 11, 2019 Are the camshaft actuators designed to be used so much? They seem like the sort of thing that would wear out quickly if used 10x as often as OEM. /Chinese oxy sensor is still working Quote Link to comment Share on other sites More sharing options...
Yowzer Posted March 11, 2019 Share Posted March 11, 2019 They move constantly. If you've ever watched the live data on a VVTi car, they dance around all over the show. 2 Quote Link to comment Share on other sites More sharing options...
Roman Posted March 12, 2019 Share Posted March 12, 2019 16 hours ago, kpr said: guessing it depends what you are shooting for? economy or power or just science? will assume science. but if economy. would it not be better to go the other way to a point, less advance and higher throttle opening to get the same airflow. which would reduce pumping loss? Yes so this is why part throttle stuff is so complex/interesting. As you say, what is the point of part throttle... You're obviously not trying to make max power or max airflow. So yes for me economy is the primary consideration, but also trying to get a good understanding of why things are beneficial while I've still got a good suite of sensors attached. As I want to come up with a good methodology for finding best cam angle when only MAP or TPS available as a reliable load axis. (ie quads) At the moment I can observe the effects and see why changes are beneficial but I honestly have no idea exactly why they are. Seeing the map signal change how it does in relation to MAF was an interesting thing. Sooooo I think I've found what I'm after. It only happens in the higher rpm of my test because 30% throttle doesnt pull enough vaccum to make it happen at lower rpm. But generally you see the MAF signal increase as the manifold pressure goes down because of increased VE. Buuutttt when intake manifold pressure gets low enough and there is enough overlap you start getting exhaust gas pulled back into the cylinder. When this happens map signal goes up and maf signal goes down. You can see here it only starts to happen right at 45+ degrees cam advance, the trend reverses: (Sorry havent labelled axes cos I'm a cretin but that's cam advance on horizontal and arbitrary rescaled numbers on the vertical) I think the 45 deg mark is the critical point where you start getting EGR effect happening. So thats whats relevant for economy, finding that point at each load/rpm combo. So I would expect if I run this test again at lower rpm with the same vacuum level I will start to see those converging lines start to happen at a much lower cam advance angle. As at half the rpm you have twice as long for a given amount of overlap to start pulling air back in from the exhaust. These tests even at 5500rpm are still at 75kpa give or take, so not much vacuum generated! Will try again later tonight and try get some ~2500ish rpm results at maybe 40-50kpa hopefully. I think when I swap to a dual VVTI engine and can adjust the exhaust side too. You will be able to make convergence happen sooner without the VE increase that comes from advancing intake cam. So can set the intake cam angle for worst VE / best pumping losses and exhaust cam for best EGR effect. I hope it works out as clearly as it seems it might! 5 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.