219 replies to this topic

[acres]

Had a few problems that have been ignored/worked around while sorting out the whole leaky valves situation but now the valves are working well and the primary is boosting as it should it was time to address these issues. First issue was that it would not run well with rich AFR's and boost so as a work around I just had the afr's set at low 13's while under boost. This gave the engine very little power but that was of no concern as the issue was getting the valves to function correctly and allow the primary to boost. With the valves sorted it was time to add some fuel and increase the power but over 4000rpm and approaching 1 bar boost, increasing fuel was causing horrid misfires on all plugs, pulling the plugs showed they were extremely fouled from carbon. The plugs that were fitted were just basic Bosch plugs so as a test these were removed and a set of used iridium plugs were fitted to see if there was an improvement. There was not:

 

 

The shiny specks on the end of the electrode look like bits of metal and at first I thought it may be bits of piston but surely I would of heard detonation if it was that bad? Turns out it was just clear specks on the electrode so I don't think it was anything nasty, it was possibly just starting to burn off by the time I got the car back to the garage to pull the plugs.

I moved ignition timing up and down by quite a degree to double check that it wasn't a timing or knock issue causing the poor spark but the result was the same regardless of timing so next was to try and crank up the dwell. I had this setup at 3ms but I don't have a scope to test the coils to see if that was accurate, I blindly set it to this based on an article I found online for these coils that was released by 'Link'. So as a test dwell was increased to 5ms, again I can't check with ascope whether this is too much but the coils weren't getting hot so I took it for a drive and the misfiring definitely improved. It was still breaking down but AFR was richer and was now in the mid-high 12's. Pulling the plugs again they were still caked in soot and being used (well used actually) they were probably at the end of their life so I threw in a new set of basic NGK's (BR6E) and the problem was then solved. No more misfiring and was happy to run AFR's of 10's and 11's. I keep forgetting to turn the dwell back down to double check whether the dwell was the culprit or whether it was simply bad plugs as I am curious since the bosch plugs were not very old at all. I guess they could have been killed while the tune was a bit rough though.

Next problem was one that I also have known about for a while but just didn't bother with it as it was out of the way at the time. It is a sync issue at 6500rpm. With the valves working, no more misfiring, a smoother changeover and the engine making a bit more power it was time to push the engine to the red line and ensure that there is perfect sync so that I can finally get this on the dyno and try and get some power out of it. 

I think it was talked about on previous pages but just to refresh, basically the biggest issue with this ecu is getting a clean signal from the VR sensors for the cam and crank. The sensors on these apparently throw out a very large voltage at higher rpm and a low voltage at lower rpm/cranking speeds making it further difficult to set the correct trigger levels. There is also a lot of noise in the signal that gives a lot of false triggers. To try and cut it short, the standard circuit's could not tolerate the noise from the sensors and it was impossible to even get the engine to start. Through trial and error the noise from the crank sensor was cleaned up using an inline resistor and the noise from the cam sensor was cleaned up using an inline resistor in combination with a shunt resistor to ground. The trigger levels for both were set as high as they could be before blocking the signal's entirely at crank speeds. With the noise removed the ecu could happily sync with the engine position however another problem turned up as the rpm begin to increase and it was again losing sync. By looking at the trigger logs I eventually worked out that one of the teeth on the cam wheel was being filtered out at higher rpm (if I recall it was ~3500rpm at that point of lost sync). By looking at the circuit schematics and with the help of google and some lucky guesswork I figured it was a capacitor that was causing the grief and again through trial and error the suspect capacitor was replaced with smaller and smaller values until full rpm could be reached. The max rpm before sync loss would increase as the cap size was reduced. At the time, with a sloppy tune and open wastegates to restrict boost, it was able to reach just shy of 8000rpm with no sync loss by reducing the value of the capacitor from 220nf to 22nf. 

This is a screenshot of one of the trigger logs that show the missing tooth:

 

 

Now that engine power has stepped up a bit another sync error has popped up at 6500rpm. The tooth logger shows it as noise this time though rather than a missing tooth. To try and clear the noise I added another 10nf cap in parallel with the 22nf as the next largest cap I had on hand was a 56nf which I know from previous experiments was too large and would filter out the third tooth on the cam wheel as well. The combined 32nf of cap was too large though and also filtered out the third tooth.

This is a screenshot of the noise:

 

 

I haven't fixed this issue yet but I find it interesting in the way the ecu responds to each sync loss. With the noise at 6500rpm it acts more like a nasty misfire and if I keep the peddle mashed it will re-sync and try it's hardest to push through the error. With the missing tooth it's a horrible violent complete shutdown that needs a restart to re-sync. When I added the 10nf capacitor and took it for a run I shit bricks when the loss hit, it was just below 6000rpm, secondary was online, afr's were rich and boost was high and it just shutdown with this god awful massive explosion. The car rolled to a stop with the engine shutdown but still popping and backfiring and I thought for sure I was picking up bits of engine off the road but luckily it was all just noise and show. Everything was fine, engine restarted perfectly and I promptly drove it straight back to the garage and ripped that capacitor straight back out.

The injectors are actually maxing out at ~6000rpm at 1 bar as the afr leans right out. The duty cycle at this point is ~165% as well which means that the injectors are basically not closing at that point. With this in mind and the way the ecu behaves to the noise, I am wondering whether the cause of the noise is actually a consequence of the maxed injectors and the symptoms these will produce, such as a possible decrease in fuel pressure, certain cylinders leaning out more than others, detonation and so on.

I could try adding additional inline resistors but I doubt this will have much effect as there is already 50k of resistors wired in and although I can't remember what the resistance of the sensor is at this point I am pretty sure it is a lot less than this. I will double check this anyway.

I think I might just replace the injectors with some larger units first though since I really should have already done this anyway.  

[Mr Legacy]

Great read Acres. I'm having lots of dramas with mine. Almost ready for a single but will keep twin for now. I'm considering an aftermarket ecu. What programs are you using to log and reflash? What computer and OS are you using also? I have a Rev D ETune.

[duncanm]

Agreed - this is an awesome read.

 

I don't want to be the guy over your shoulder, as you're obviously putting the real work in here and I'm just being an armchair expert, but have you stuck a CRO on the sensor signals at the ECU ?

 

Lots of reasons those might be a bit shitty - dirty sensors (bits of metal on the magnets), crappy grounds, etc.

 

I note that the crank and cam share a shield past the bulkhead - maybe there's some crosstalk - it certainly sounds like you're getting some high freq noise in there (hacking the caps to filter false triggers)

[acres]

Thanks guys. I try to share as much as I can about the progress but I'm sure there is plenty I have missed.

 

Mr Legacy, ecu is a MegaSquirt and software is the accompanying TunerStudio and Mega Log Viewer. It is most definitely not a plug and play unit and does require quite a bit of modification to be use able. On the plus side it can do basically anything and is a fantastic learning experience. Computer is just my everyday laptop with the very non user friendly windows 8. What drama's are you having? Start up a thread, there is plenty of knowledge on this forum, maybe we could offer some advice.

 

Duncan, all advice, pointers, corrections and opinions are very, very welcome. For the most part I am learning as I go and have basically been winging it all through this entire project. Any advice, particularly in regards with the electronics/circuits, that anybody has to offer is most certainly appreciated.

 

I don't have a scope and don't know where I can get one that isn't at least $500 so haven't been able to trace the signals but I do have a schematic somewhere on the laptop for a DIY soundscope that just plugs into the microphone jack on the computer and uses some software to create a trace. I was planning on having a go at it a while back but just never got round to it so I think I might dig it out and have a crack at putting it together this arvo, see how well it works.

I don't think the sensor or wheel is dirty as it wasn't that long ago I had the heads off and the wheels were nice and clean, I guess some shit could have worked it's way in there but the false trigger is pretty reliable in that it only appears at the exact same rpm and cam wheel position. I would expect that if it was a sensor/wheel issue than the noise would be more random or at least less reliable in terms of popping up at the exact same rpm every time.

 

Yep the cam and crank grounds meet at the earth/shield junction and share a ground back to the ecu. The sensor shield's also meet at the earth/shield junction and share another earth back to the ecu along with shield's of other sensors. I actually poked around in the factory ecu and worked out that all earths coming into and out of the ecu share a common ground on the board. So for this reason I never altered any of the grounds, just made sure that every ground wire went back to the megasquirt on it's own pin. All ground pins are also common for the MS. I figured that if that's the way it came from factory than it is probably the best way to manage the grounds but maybe it could pay to separate the sensor grounds? Might help clean up some of the noise?

 

I also measured the resistance of the sensor again and it was about 2k, since I already have a 20k inline resistor and and a 30k shunt I don't think adding larger resistance will do too much. Not the shunt anyway, maybe a larger inline will help though?

With the current circuit though the engine can free rev to redline without losing sync, the problem is just coming up when the engine is loaded. I have ordered some fuel rails and injectors though so will be interesting to see if the problem clears itself once fueling is no longer maxed.  

[acres]

Think the trigger issue is solved. I didn't end up trying the DIY sound scope as I only had female stereo jacks on hand and needed a male one, I don't know how useful it would have really been anyway though as I googled it again and it seems that it was capped at less than 1V anyway otherwise you would fry the sound card in the laptop. So with time to kill I thought I would turn the dwell back down and see how it would respond, turned it back down to 3ms and took it for a run. There was some slight stumbling but no real solid misfire's but the big issue was the increase in sync losses. It had a few errors including a sync loss at a lower 5400rpm. I pulled the plug on cylinder 1 after the run and you can see it is starting to foul up over the insulator:

 

 

So turned dwell up to 4ms and took a tooth log while free revving the engine to the redline. There was no sync loss at all and I repeated this 3 times to make sure. This is the tooth log:

 

 

And this is an image of the logger showing bars with each trigger of the crank and cam wheel. Cam is the green bars up top, crank is the blue lines down the bottom:

 

 

Next I took it for a drive with the valves forced open, so running in parallel mode, with the boost limit lowered so it would run on spring pressure (still 1 bar though) and surprisingly it ran to the redline with zero sync loss! So switched it back to sequential mode and took it for another run and again no sync loss! Pulled the plug from cylinder 1 and it look's near perfect:

 

 

Took it for another run to record the tooth logger this time and this is the result:

 

 

Actually bounced it off the limiter this time by accident so I think that may be the reason for that single false trigger. 

 

Looking at the logs again and the battery voltage is actually dropping a bit too low while loaded up. It's around 13.3V at idle/low load but under throttle it is dropping as low as 12.7V. I will have to investigate this further but the lower voltage would surely be adding additional stress to the electrical system, increasing the incidence of the sync errors. I still think the larger injectors will help a lot as again with them maxing out it is adding additional load to the system, contributing to the sync error's. My theory on the dwell angle contributing to the errors, I presume that the lower time (3ms) is causing the sync errors at the higher power levels due to the additional energy required to create the spark. The coil was not receiving enough charge and was strained to deliver the spark, feeding noise back into the system. With the larger dwell (5ms) the coil became saturated from excessive charge time and delivered an unstable discharge for the spark. This saturation of the coil again fed noise back into the system.

Happy to hear all comments as to whether or not this theory would be correct.

 

I know it's a very narrow power band at this point but without the sync loss, once it hits 6000rpm - forking wooshka! Even with running way too lean and on spring pressure. That rate of acceleration is really exciting to see how well this all might end up working.

[allpaw4]

I think you'll probably know this already, but remember for a shield to work properly one end needs to be open (ie not connected to anything). If both ends of the shield are terminated, it will just act as another peice of wire. Might be worth going over the earths and checking them in comparison to the mega squirt diagrams, as the factory system may have differing shield termination points, and if both are connected, you'll run into dramas.

My apologies if you have already checked this and know this side of it, but I thought I should say something about how a shield is vastly different to an earth. The amount of times I've run into shields used as earths is unbelievable.
This really does make for some interesting reading, and it sounds like you're getting close to having it sorted out. Best of luck, and enjoy.

Cheers,
Ben

[Matt]

Sounds like your alternator is on the way out

Mine was dropping to 12.5v under load and above 4k was blowing out the spark a fair bit and causing a few issues

It also didnt give the fuel pump enough voltage when it needed it and made it look like we were running out of injector duty cycle, when it was just an under performing pump.
To cope with the drop in fuel pressure as the voltage dropped we had to wind the boost from 21psi to below 16psi between 4k and 7k rpm

$30 and a working alternator later and all is good....back to heaps of injector headroom...no spark breakdown and the fuel pumps working perfect

Going in for a retune in a week or two to make the most of the fixes

[acres]

Great info Ben! Too be honest I hadn't paid any attention to how the shields were grounded but that's probably because I left it as it was from factory. To wire in the MS I just made a patch loom by cutting the connector out of an oem ecu and soldering wires to the pins that stick out of the back of the connector. Each ground wire in the factory loom had its own separate wire and pin at the MS to try and keep the grounding as from factory.
Here is some photos from the book I pulled the schematics from:

Connector B139 is the shield/earth junction with pins 12 & 13 the cam and crank grounds. Pin 15 is the ground back to the ecu.


E11 & E3 are the cam and crank sensors. They share the same shield and ground at pin 10 on the earth/shield junction. I wonder if some of the noise in the system is because they share the same shield?

Funnily enough Matt, I was only reading your thread the other day about the grief you were getting from a low voltage and was the first thing I thought when I noticed the voltage in the 12's. Pretty sure I have another alternator floating around so I might switch it out and check for an improvement. I'll double check for any voltage drop across the wires too.

[acres]

Got the multimeter out and checked out the battery and alternator. With the engine off there is zero voltage drop across the terminals, alternator cable and starter cable. At the ecu though there is 0.1V drop. With the engine running the alternator reads between 13.9-14V with a drop of 0.1V at the battery, at the ecu though it drops a good 0.45V. After taking the car for a drive and getting everything up to temp, the alternator output drops to 13.75-13.85V with the same drop of 0.45V at the ecu.

So at idle the ecu is recording a voltage of 13.3-13.4V and this is confirmed with a multimeter but the actual alternator output is ~13.8V - this is within spec so that's a good result.

It's a bit difficult to check the multimeter while driving but when under load and the ecu is seeing 12.7-12.8V in the logs, factoring in the same drop of 0.45V, then the alternator should be putting out 13.15-13.25V. That seems within spec too though so I guess the alternator is probably fine?  

 

I guess the issue is, is the drop at the ecu acceptable? Looking at the wiring diagrams the power source is shared through a few relays to power things like the fuel pump and injectors so I guess a small drop would be expected?

[Matt]

Mick talked about the ecu aiming for 13.8v minimum under load?

My voltage at the ecu now stays spot on 13.8 when loaded..(scangauge reading)..dips down to maybe 13.6 max for a second then comes back

Id honestly just swap alternators and see if theres a difference as it sounds like the cabling is fine

But yep a drop should be expected due to supplying pumps and relays etc....but the regulator should up the voltage quickly to compensate

Its funny that ive now heard of two other people with my same issue lol...a local shop with a mint gc8 track car have it and a swapped alternator sorted it all and they were able to retune for a heap more power

[acres]

Well that makes it a huge drop at the ecu then if it should be up around 13.8V.
Just checked and there is a spare alternator in the shed so I'll switch it out tonight and check the difference.

[Matt]

Fingers crossed it helps!!

[natho]

any updates on this?

[acres]

Literally haven't touched it since the last update in this thread! Got busy with work and had a few other car related jobs that took priority in my little spare time. Sisters bug eye developed a massive oil leak from the input shaft of the gearbox which made it undrivable. Pulled the engine and box (to replace engine mounts and few other seals too) and it seemed the failed clutch bearings were the most likely cause of the failed input shaft seal. Split the box to replace the seal and replaced the 2 main bearings at the same time. This was a biggish job so took me a while to complete and get the car running again. At this time the clutch in the RS had finally given way for good (was slipping for at least the last 6 months) so took the opportunity to pull the motor and replace the many worn components (still running the original complete motor at 300k). It had a few leaky seals, some pretty tired hoses and the engine mounts in this were completely split in half. While it was on the stand I was going to pull the heads and either refresh them with some new lifters, or at least clean them, until I got a call during the day from my sister saying she was on the side of the road, the bug eye had overheated. The lower radiator hose had popped off and dumped all the water, end result it blew the headgaskets after only being back on the road for maybe 4 weeks. What a turd! So with that news I had to forget about refreshing the RS heads and just get it back together asap so she could borrow it while I got stuck back into the bug. The heads were machined yesterday so I am just waiting on the gaskets and a few other bits to turn up so I can get it back together. I'm also going to chuck a bearing kit through the turbo as it's a bit oily, shaft play is normal but it looks like it may have blown it's seals so I'll give that a refresh too while it's apart.

 

For the gen 2 though I ended up getting some top feed conversion rails, ID1000's and an FPR800. Just needed to order some lines and fittings and I will get these installed over the christmas break hopefully. I am looking to order a roll of hardline and just run a parallel setup under the manifold. Will use some appropriate AN fittings to join the lines to the rails and some AN barbs to terminate the hardlines in the factory position, so I can then just run the rubber lines to the filter and FPR. Will hopefully look nice and neat once done.

 

When ordering bits for the bug and RS I also ordered some brushes for the alternator, so I'll chuck these in and hopefully sort the voltage issue - I've had a flaky alternator in the past that was also fixed with some fresh brushes.

[acres]

Updates. 

Got the rails, fpr and injectors installed. The original plan was to run some hard line under the manifold and fit them all up with some nice AN fittings however I couldn't source a AN flaring tool locally at the time so ended up just getting some brass barbs and tees from the hose shop and knocking up a simple parallel setup with efi hose. It doesn't leak so will do for now until I get around to replacing it all again with the hardline. When I updated the ecu with deadtimes, etc of the new injectors and gave it a kick in the guts it soon became pretty clear that the original data I was using for the factory yellow tops was very, very off (not suprising since it was all just guesswork and trial & error). The fuel maps were wildly off so it took quite a bit of tweaking to get everything back to where it should be. The good thing is now that it has accurate data the fueling is much more reliable and repeatable, especially at the lower loads.

The alternator was stripped to replace the brushes but they were still in good condition so I left them as is (also because I couldn't see how to get them out easily) so I just scrubbed out all the carbon and general shit that was built up in there and gave the contacts a light sand before putting it back together. Voltage doesn't seem to be dropping off anymore so I think I'll leave it at that for now.

Another sync error reared it's ugly head. Checking the logs again this time it seemed to be appearing on the crank sensor instead of the cam. Messed around with some resistors to try and clean it up but that was just making it worse so taking the same approach that worked best with cam sensor I switched out the capacitor on the input circuit. First value I tried seemed to work perfectly, literally hasn't been one sync error in any of the many logs I have taken since making the change, not even while cranking where it would constantly have a few errors before firing up.

The starter motor had also been giving some grief for a long time now (I cooked it a little bit way back when while trying to get the engine to start when the ecu was first installed). It would struggle/refuse to spin if the battery voltage had dropped slightly or once the starter was hot after the engine was up to operating temp. So I stripped the starter to see if there was any obvious part that looked burnt out or excessively worn as I have a couple of auto starters that I could strip a few parts from but there was nothing that I could pick up on. Decided to just give it a clean, sand the contacts and re-grease the gears and see if that made any difference and surprisingly it hasn't had a problem since. 

So it now starts reliably (near on oem reliable), idles and drives smoothly and doesn't just randomly crack the shits and shut down from bullshit sync errors, I have made the exectutive decision to get the car prepped for roadworthy, re-registered and tuned professionally. For starter's I think I will get it tuned on primary only, so essentially running as single turbo, so that I have a good tune to a: work with while I play around with getting the secondary staging smoothly and b: I want/need to build the block before jamming in the bigger boost and the second turbo. It would be nice to have a good tune to break in a rebuilt engine.

 

I might start a build thread for the car too since after so many years of neglect it is a bit of a dog at the moment and will need a alot of work to get it back to being a nice car.

[natho]

two year update?

[Beckers]

two year update?

Haha 

But on a serious note, 

any news/update would be great.

[acres]

Lol. Have a couple of massive updates. Sometime in the last two years I replaced the right rear window when the mower threw a rock into it and labour day last week I cleaned a dead rat from under the bonnet when I jump started it to shift its position. What a machine this thing has turned out to be
I have had the worlds worst work/life balance the past few years so just dont have the time to work on it. Im sure Ill finish it one day though....

Pic to prove it still exists:

[Barbbachello]

Awesome to see its still alive