This is a good concise thread that will answer just about anyone's questions about the safety of PI if they search for it. Much easier than digging through ten "blew my engine" threads to find pieces of the convo strewn about. Thanks!
Port Injection causing engine failure
- Thread starter dyezak
- Start date
My engine blew Friday, logs are below.
Car felt funny at the top of a 2nd-3rd-4th gear pull. No CEL, codes, or map 4 triggered. Logged a 3rd gear pull and she popped.
-Went into open loop and afrs went lean
-PWM was abnormally high
-Trims bottomed out briefly post shift
-My restrictive exhaust was replaced with a single 3" straight pipe earlier in the day
http://datazap.me/u/samstryker/engine-failure?log=0&data=1-4
Btw I'm Sam and I'm new here sooo Hey
Car felt funny at the top of a 2nd-3rd-4th gear pull. No CEL, codes, or map 4 triggered. Logged a 3rd gear pull and she popped.
-Went into open loop and afrs went lean
-PWM was abnormally high
-Trims bottomed out briefly post shift
-My restrictive exhaust was replaced with a single 3" straight pipe earlier in the day
http://datazap.me/u/samstryker/engine-failure?log=0&data=1-4
Btw I'm Sam and I'm new here sooo Hey
I hate to beat up BMS on this...but their big soapbox they stand on for sales is the safeties that the JB4 adds to the mix. You would think that it would be easy, even a no brainer, to state that if you are WOT and not in map 4 or 0, and the trims show open loop that shit went wrong and you need to trigger a safety.
If that's your big selling point (safety), then seeing logs like this on a popped motor are akin to a black eye on your logic....and if your logic is this thin on something this important where else is the logic lacking?
Looking at problem cases always helps improve the logic. Here "before it blew up" I see AFR leaning out and it switching to map4. That specific log has an ios bug where the timestamps are compressed so I can't tell how long it took to trigger map4 but seems like around a second after AFR went over 13:1. Which is what it's programmed for. Maybe that could be faster like 500ms. The key there is to also avoid false triggers in case AFR jumps lean during a shift and jumps right back. The safety triggered was lean run not bank to bank trim variance. On the twin turbo firmware the open loop = 0 safety isn't active as there were a lot of false triggers on manual cars. Probably would not be a bad idea to enable that safety on automatic only cars, though. Catching real problems while avoiding very annoying false triggers is always a balancing act.
Then in the run labeled "it blew" I don't see any obvious signs of distress. Trims normal, AFR bank to bank normal, timing OK. First sign of an issue is boost dropping off but that would not normally trigger a safety. If something unhealthy was going on during combustion we'd see it in the AFR.
What was the actual damage? Blown piston ringland? Was the car running stock knock sensor tables and misfire detection enabled?
Last edited:
I am all for safeties, but the JB safeties are no where near fast enough to save any motor from detonation or misfire damage. I think it take 4 seconds of over boost to kick the car into map 4, same for lean condition over a certain range. Correct me if I am wrong T. Last time I ran the JB I have logs of an entire 1/4 mile pass with the AFR between 14-17:1, I was leaning on the JB safety to kick in, if we had issues, never happened. There is nothing wrong with a JB4, I am going to run one on our new GTI to test our upcoming turbo, but once you get to 650+WHP in my opinion Piggy backs are just asking for issues. Lying to sensors at those levels, not having proper HPFP reading, etc, etc do not offset the convenience of being able to switch maps, etc. Under 650WHP JB is a great tool, and ads quite a bit of convenience to the car. I recommend them to customers all the time. My point here is, getting on Terry for any motor failure is ridiculous. He programs safeties into the JB, but no safety outside of the factory DME is going to be quick enough to stop engine damage if things go really wrong quickly with lean conditions. The DME has saved my butt countless times, but I have also blown up more motors than anyone testing, SO... Bottom line, you want to push past 650WHP, do not blame anyone, or any product for your engine failures. Set up the car properly, enough fuel system, proper tuning, and proper testing will go a lot further than trying to blame PI, or BMS for your engine failure....
The over boost safety is set at 1 second, same as the lean run. The bank to bank fuel trim variance safety is set to 1.5 second. 4 seconds would be way too long, lol. The DME's safety systems are focused around knock and pulling timing and it does that quite well. The JB4 safeties are focused around spotting obvious hardware issues and alerting the driver/reducing boost BEFORE it gets to the point that the DME safeties need to kick in. If you run your high power car long enough at 16:1 it's going to melt down and the DME won't stop it. So the JB4 needs to. Same goes for an over boost situation.
If you have a log where a specific safety should be kicking on and isn't by all means lets see it so we can correct that. If you come up with an idea for a new safety that would be helpful we can program those in quickly. I've had it save me countless times at the track as have hundreds of our customers. The better customer data we get particularly issues where things went sideways the better we can make those safety systems operate!
With a back end flash loaded the sensors all see the true values including the high pressure pump so it sounds like you don't really understand how the JB4 is implemented. Seems to be common among some within the industry. Some "tuners" really don't know and others do know but pretend they don't so they can continue to spread false information.
If you have a log where a specific safety should be kicking on and isn't by all means lets see it so we can correct that. If you come up with an idea for a new safety that would be helpful we can program those in quickly. I've had it save me countless times at the track as have hundreds of our customers. The better customer data we get particularly issues where things went sideways the better we can make those safety systems operate!
With a back end flash loaded the sensors all see the true values including the high pressure pump so it sounds like you don't really understand how the JB4 is implemented. Seems to be common among some within the industry. Some "tuners" really don't know and others do know but pretend they don't so they can continue to spread false information.
Last edited:
Not pointing any fingers here. Just want to find out what went wrong and prevent it next time. I'm targeting 26psi mid range and only hitting about 24 up top, I wouldn't really say I'm sitting that close to 650whp. I also have plenty of extra room in my fuel system. I agree that a piggy back can become less ideal at higher power levels, but for my purpose (600ish hp) it should suit my needs. Logs have always looked clean up until this point.
If I went into a safety map on the 2-3-4 gear pull, how was I targeting full boost on the next 3rd gear pull? I didn't switch maps or change anything between the logs. It seems like I went lean in 4th gear for a long time (up to 17:1), boost target never changed until I let off either. I definitely want to implement open loop safety. I'd rather trip false safety map switches than pop another engine.
Went overkill on my fuel system, tapped my PCV ports, constantly reviewed my logs, and kept things pretty conservative. Not sure there was too much more I could have done on my end.
Any feedback is greatly appreciated. Looking to learn from this and come back stronger.
-Sam
If I went into a safety map on the 2-3-4 gear pull, how was I targeting full boost on the next 3rd gear pull? I didn't switch maps or change anything between the logs. It seems like I went lean in 4th gear for a long time (up to 17:1), boost target never changed until I let off either. I definitely want to implement open loop safety. I'd rather trip false safety map switches than pop another engine.
Went overkill on my fuel system, tapped my PCV ports, constantly reviewed my logs, and kept things pretty conservative. Not sure there was too much more I could have done on my end.
Any feedback is greatly appreciated. Looking to learn from this and come back stronger.
-Sam
The bold part is what I saw too. Your trims were flatlined for 3.5sec indicating open loop and your AFR was incredibly lean for the last 3sec of that.
The reality is we shouldn't be relying on open loop at all under boost and WOT....if that's detected we should be tripping a safety. I'm just not sure the JB4 has the capability to do that reasonably quickly to save things in this situation.
Yes same here, always looking to make things better.
I see it switching to map4 in the log. How did it switch back? Did you restart the car? Once it changes to map4 it flashes the SES light a few times (based on safety triggered) and stays on map4 until you restart the car OR change the map manually.
I see it switching to map4 in the log. How did it switch back? Did you restart the car? Once it changes to map4 it flashes the SES light a few times (based on safety triggered) and stays on map4 until you restart the car OR change the map manually.
We can trigger it after 1/10th of a second but some open loop is expected during WOT shifts, especially with manuals. Normally with an engine failure you want to determine what physically failed and try to work backwards to surmise a cause, and then evaluate conditions and safeties that were and/or could have been in place to address that cause.
I do see the map 4 change but it didn't occur until the very end of my pull (5460rpm in 4th gear) by then my AFRs had climbed up to 17:1 and the damage was probably already done. What's weird is I'm 100% certain I never shut off the car or changed the map between the two pulls, but on the next log I'm back in map 6... Obviously I'll be running the open loop safety when I'm back on the road, any other suggestions? It might be smart to lower my AFR safety trigger if possible as well.
I'll make sure to post my findings after the tear down
I'll make sure to post my findings after the tear down
Right, comes up about a second and a half after it goes lean. The timestamps are compressed in that first log. There is also a minor 1/2 second delay in the testing during the actual shifting we could tighten up. It's hard to imagine that a couple seconds of leaning out just one time on conservative tuning killed it although it's certainly possible. Normally takes being that lean through a few gears to do it.
What actually failed? Which cylinder broke? #1 had the worst timing. Didn't see answer re: knock sensor sensitization and misfire detection?
Something else we should look at is why it's going rich during the shift. Maybe that is what triggered the open loop in the first place. We made changes to the PI algorithm between v32.5 and the latest v32.8 when adding the E85 sensor and virtual sensor scaling that might already take care of that.
I'm looking over the older v32.5 source code and don't see any reason for it not to sway on map4 until reset. The only change we could make there is to make the map change stored (e.g. not only until restart).
What actually failed? Which cylinder broke? #1 had the worst timing. Didn't see answer re: knock sensor sensitization and misfire detection?
Something else we should look at is why it's going rich during the shift. Maybe that is what triggered the open loop in the first place. We made changes to the PI algorithm between v32.5 and the latest v32.8 when adding the E85 sensor and virtual sensor scaling that might already take care of that.
I'm looking over the older v32.5 source code and don't see any reason for it not to sway on map4 until reset. The only change we could make there is to make the map change stored (e.g. not only until restart).
In my experience the JB4 safeties are nice addition, and better than nothing to a point. When the car goes into open loop and runs lean, the JB4 safety can pick it up. I don't think it always picks it up, but not always a fault in programming. A 3rd gear log can go by really fast... hell a 4th gear log can go quick at high power, so when you stretch it out, a car can go halfway through the gear or all the way though a gear with no safety engaging. It's better than nothing, but what's even better is being attuned to the car enough to feel when it's running poor and being able to let off and figure it out. What's also better is being able to spot problems in logs, before they become problems... or catch them as they begin to develop. Which leads me to my biggest issue with the piggybacks.
The biggest issue I've had of late with the JB4 isn't the safeties, it's just diagnostic ability. I really do wish there was a way to get full logging on cars running it but I realize it's impossible. STFT 1 and 2, LTFT 1 and 2, torque limiters, higher Hz refresh, channels labeled correctly so I don't constantly have to remind myself what the abbreviations mean, that sort of thing would be a welcome add. The JB4 channels are great for a quick shapshot and for the bluetooth stuff, and strictly speaking it's good enough, but it does make certain things more difficult. As far as tuning on them, the hardest thing isn't understanding how to work with the piggyback and the DME properly, it's learning and keeping up with changes on how to program the JB4 to do what you want with all the FF, AFR/duty bias, DWP, AFR Bias etc abbreviations and how everything is scaled and how the logic behaves (PI biased on LPFP, the new E85 firmware, etc). If the tuning interface was more advanced it would actually make things a lot easier.
The biggest issue I've had of late with the JB4 isn't the safeties, it's just diagnostic ability. I really do wish there was a way to get full logging on cars running it but I realize it's impossible. STFT 1 and 2, LTFT 1 and 2, torque limiters, higher Hz refresh, channels labeled correctly so I don't constantly have to remind myself what the abbreviations mean, that sort of thing would be a welcome add. The JB4 channels are great for a quick shapshot and for the bluetooth stuff, and strictly speaking it's good enough, but it does make certain things more difficult. As far as tuning on them, the hardest thing isn't understanding how to work with the piggyback and the DME properly, it's learning and keeping up with changes on how to program the JB4 to do what you want with all the FF, AFR/duty bias, DWP, AFR Bias etc abbreviations and how everything is scaled and how the logic behaves (PI biased on LPFP, the new E85 firmware, etc). If the tuning interface was more advanced it would actually make things a lot easier.
On the safety systems logs showing problems are really invaluable for evaluating and refining those systems. There is no reason to have to go half way through a gear on a problem if we all can look at a log and spot it sooner. That's the benefit of having a programmed system watching all the data coming in real time. It can spot the trend and pull the trigger on a safety real time. Not after the run or after lots of runs when you look at logs later (maybe for the purposes of finding out why your new build is suddenly smoking badly after its initial runs). It's the same reason the JB4's in dash gauges are so useful. Having an eye on boost and ign timing in all 6 cylinders real time all the time helps you spot any issue as it develops.
We just need to continue to look for and evaluate markers keeping in mind some can be "false" like open loop mode. A well tuned car can run in open loop with AFR perfectly on target, and well tuned cars will fall in to open loop during NLS, DTC, and under certain normal conditions. But open loop can also predict if a car is going to run lean in some situations sooner than when you see it under AFR.
On the JB4 we're slowly standardizing names. Anything we can do to reduce the number of questions and confusion is worthwhile IMHO. For example we adjusted PWM to WGDC for that purpose. We've given E85 % a dedicated channel and soon STFT bank2 will have it's own dedicated channel (rather than sharing with IAVG), to simply log reading. The analog sensors spit out values at 250hz but CAN stuff is currently set to 10hz. There is an interface setting on the windows interface called "raw sampling" that let's you get 40hz analog with 10hz CANbus... Logs look wonky but if you need the high speed to look for very subtle boost or fuel pressure changes its there. Ultimately I see no reason we can't get 40hz CANbus since we're out on the PT can directly. Working on it.
On understanding the PI mapping we've designed it to be really easy. In the log you look at "fuel enrichment" which is a 0-100% dutycycle to the PI injectors. And then you can see how it changes based on other parameters in your log like low fuel pressure, intake temperature, boost, engine speed, etc. The "fuel bias" values give you a means for ranging part of the mapping in a 0-100% way by RPM. So if you need more fuel at higher RPM all else being equal you raise fuel bias at that RPM so the algorithm scales up at that engine speed. We've recently added a scaling for virtual and physical flex fuel sensor which was the last missing component on that system.
On understanding WGDC, FF (feed forward) & dutycycle by RPM, default wastegate position, pid gain, etc, I've tried to make it as simple and easy to understand as possible but always like to hear ideas for improvement.
We just need to continue to look for and evaluate markers keeping in mind some can be "false" like open loop mode. A well tuned car can run in open loop with AFR perfectly on target, and well tuned cars will fall in to open loop during NLS, DTC, and under certain normal conditions. But open loop can also predict if a car is going to run lean in some situations sooner than when you see it under AFR.
On the JB4 we're slowly standardizing names. Anything we can do to reduce the number of questions and confusion is worthwhile IMHO. For example we adjusted PWM to WGDC for that purpose. We've given E85 % a dedicated channel and soon STFT bank2 will have it's own dedicated channel (rather than sharing with IAVG), to simply log reading. The analog sensors spit out values at 250hz but CAN stuff is currently set to 10hz. There is an interface setting on the windows interface called "raw sampling" that let's you get 40hz analog with 10hz CANbus... Logs look wonky but if you need the high speed to look for very subtle boost or fuel pressure changes its there. Ultimately I see no reason we can't get 40hz CANbus since we're out on the PT can directly. Working on it.
On understanding the PI mapping we've designed it to be really easy. In the log you look at "fuel enrichment" which is a 0-100% dutycycle to the PI injectors. And then you can see how it changes based on other parameters in your log like low fuel pressure, intake temperature, boost, engine speed, etc. The "fuel bias" values give you a means for ranging part of the mapping in a 0-100% way by RPM. So if you need more fuel at higher RPM all else being equal you raise fuel bias at that RPM so the algorithm scales up at that engine speed. We've recently added a scaling for virtual and physical flex fuel sensor which was the last missing component on that system.
On understanding WGDC, FF (feed forward) & dutycycle by RPM, default wastegate position, pid gain, etc, I've tried to make it as simple and easy to understand as possible but always like to hear ideas for improvement.
Last edited:
Knock sensors were not desensitized. I was running the OTS BB hybrid flash with no changes made to it. Car has been sitting at the shop. Haven't had the chance to leak/tear down yet.
I'm running full e85, no ethanol sensor but that shouldn't make a difference.
I do see the rich post shift AFR values. Would an auto shift fuel redux help? Simmilar to what we have for duty bias, but for the fuel bias's?
Couldn't agree with this more. I should have let off sooner and could have prevented this. I thought it was a turbo giving out so I didn't think much about it until I saw milky oil in my exhaust...
The port injection algorithm was reworked in v32.8 so we'll have to see how those logical changes effect that specific symptom if we can come across another car with the same condition.
Sorry I'm crashing the beat up the jb4 party... And I'm late the pick back up on the discussion.
But if PI is becoming so feared as an engine melter why not disable the cylinder fuel cut on misfire or or try to add an ignition cut to the culprit cylinder.
The fuel cut is only there to protect the cat. Which I've heard a few guys don't have anyways
Not as a permanent solution but a smaller bandaid to help hold on out bigger bandaid lol
But if PI is becoming so feared as an engine melter why not disable the cylinder fuel cut on misfire or or try to add an ignition cut to the culprit cylinder.
The fuel cut is only there to protect the cat. Which I've heard a few guys don't have anyways
Not as a permanent solution but a smaller bandaid to help hold on out bigger bandaid lol
Similar threads
Technical
reflex plus and zietronic guage