i was on the dyno Jan 16 and encountered a one data point boost spike during two back to back runs which were the last of the day. our boost was set and fairly steady at 24.2 PSI.

all of a sudden the MAP sensor maxxed out. only for one data point. (4 per second). i run defensive settings at P20 so timing registered at 3 degrees and the fuel injectors show 100% duty cycle. AFR was 10.1 and knock was normal.

the next to last run the spike occurred at 7629 rpm and on the last run it happened at 7309. a diffuser went thru the engine immediately after...

generally i believe my instruments. the MAP sensor has been fully tested and is working perfectly. it registered max values on both of the spikes.

a board member pointed out that i had had a spike in the other run (Nov 2010) when i also lost a primary injector diffuser.

since then i have a new Power FC and Datalogit.

the diffuser loss may or may not be connected to the spike. my tuner, Luke indicated that the only time he has seen such a brief boost change is when there is a backfire in the UIM...

any ideas?

i am happy to supply more data. i would just say that all my engine metrics other than boost look fine to me...

next to last run Jan 16 2012




last run Jan 16, motor failure due to broken diffuser




Nov 2010 run w boost spike and loss of diffuser


here's a MAP for another perspective

 

That is weird, I have never seen large spikes like that only small ones which I guess is normal. I wonder if there could be an issue with the boost controller for the waste gate or the hose going to the bottom of the wastegate, but that is not likely to cause a spike like that.

Why would the diffusers break? Maybe the boost spike broke them? How old were they and how common is it for them to break?

it is strange for the power fc to pick up a boost spike like that, because my boost datalogging of the pressure signal output from my aem wi controller picks up little boost spikes of about 2 psi that do not show up on the pim chart, the boost curve from my wi controller is much more erratic than the pim curve so I'm not sure which one is more accurate, I thought the map sensor would kind of buffer/smoothen out the readings or just not react fast enough to pick up a large boost spike. Have you ever datalogged another boost pressure source?

 

Am I seeing this correctly, there is a much larger spike in the injSc curve? That is even more strange..

 

Possible for a glitch in the processor or a glitch in your sensor. Which would throw off everything else. I doubt the turbo outputed 10psi more for a nano second. More likely and logical you had a sensor glitch or ecu glitch. Could've been from rf noise or vibration. I work with extremely precise and temperate electronics and controllers for a living and it isn't all uncommon for a large vibration or rf noise to mess with clock speeds on processors.

 

i just did a complete audit of the data from Notepad.

first off, i am getting 11 samples per second not 4.

and yes, the spike is there.

the PIM is 29960 on both runs however the PIM voltage is 4.387 on one run and 4.356 on the other.

supposedly the GM 3 bar reads V to 4.914 average max.

PSI on 29960 is 27.90

boost prior to the spike had been steady. the dyno reading psi every 100 rpm averages 24.2

the next to last run featured the boost spike and ended w no damage. the following run the boost spike occurred and immediately thereafter i lost the diffuser and the motor/run ended.

so it appears that the spike voltage did not max out the sensor range.

as far as the IDC, on the next reading after the spike it registered 100 and it also registered 100 on the final run 2 datapoints after the spike PIM reading. timing also decreased to 3 IGL and -8 IGT.

interestingly (at least to me) the IDC did not curve up to 100 on the last run. it was normal for a following single data point, spiked to 100 on the following point and then was back to normal on the next. timing did the same.

i posted a PIM chart that shows a spike reading from my Nov 2010 run. that was w a different Power FC and Datalogit.

rxspeed7 i appreciate your comments re the glitch/noise/etc and the fact that you have experience here, after all the weird things that have happened over the last 1.5 years i am just trying to be extracautious and cover all possibilities. prior to the project my 500 hp FD ran perfectly for 5 years. w no boost spike readings,

the Dynojet dyno did log boost, about 4 samples per second and shows no spike.

after the spike the next datapoint is at 24.

Luke mentioned the only very brief spikes he has seen are from a backfire in the manifold. i did not break an apex seal in Nov 2010 nor on the next to last run this Jan so it is hard to imagine plus i would think a backfire would last longer than less than a tenth of a second.

thanks for your interest

hc

 

The backfire theory is extremely possible. Could've very well happened from pre-ignition and that pressure being pushed back into the manifold which then would've resulted in A; broken diffuser mashing the seal or B; detonation resulting to broken seal. I know you don't believe that it detonated because of your pride and I fully understand but **** happens man. It's all a learning experience.

What do you run for split at the upper boost levels?

 

I'm still wondering about the huge spike in the injSc curve, and the fact that there was no corresponding spike in the injPr curve, that looks like a glitch. I've seen some short spikes in my rpm curve before which would indicate a glitch, it wasn't tire spin.

 

If the dyno logs 4 points per second, it's hard to imagine that it missed a backfire.
Has anyone here logged a backfire? It would be intresting to see how long it actually lasts and compare.

 

Pm'd you

 

here's some data...


time IDC RPM IGL IGT knock AFR FP EGT PIM

43.602 71.2 6970 13 2 32 12.014 65.5 1520 27920

43.696 71.6 7035 13 2 26 12.135 66 1522 28032

43.789 72.6 7109 13 2 25 12.105 65.7 1526 28016

43.883 73.3 7183 13 2 22 12.24 65.5 1530 28016

43.997 73.5 7246 14 3 37 12.36 65.2 1524 28192

44.07 74.3 7331 14 3 54 11.92 65.9 1522 28248

44.164 76 7403 14 3 57 11.262 66.1 1522 29960

44.257 74.6 7425 13 2 50 10.796 65.7 1520 28704

44.351 100 7360 3 -8 55 11.262 66.1 1523 28200

44.46 71.1 7500 13 2 64 10.51 67.5 1488 26464

44.585 72.1 7515 15 4 1 10.616 62.1 1475 20448

44.694 73.3 7530 13 2 59 10.255 64.3 1484 26320

AFRs read 1/2 of one full point lean in relation to the dyno which is correct. i am replacing the FJO WB unit. while knock rose, it often was around 50 on many of the runs and i do not think it is an issue as the EGTs are cool and we have alot of meth in the motor and, given that, the timing is conservative.

this is the run where the diffuser wnet into the motor around 7300.

 

At 7246 rpm your AFR is 12.36 (little lean), then your knock rises to 54, pretty high IMO (since mine is at only 2 at 7300 rpm), then it eats the diffuser and then boost spike with 57 knock and then engine failure, seems like it could have been too lean and too much knock causing whatever happened maybe?

Just comparing your data against mine with 24 psi, your knock is high, compared to mine. I just made two pulls on the dyno last night, so I'm comparing data. mine was 24 psi, 528 hp again and 532 hp (then we found uim gasket boost leak). My AFR was a little lean in the low 12s like yours for the first run, but my highest knock reading was 14 after 5000 rpm. Timing is similar.

My wideband was reading about .5 AFR leaner than the dynojet's wideband, I was thinking mine was correct and not the dyno. Why do you think the dyno's readings are correct and not yours?

post edited a couple times.

 

Yea it really seams like all things together, knock and detonation was you problems. Do you have the graph from the Dyno? Those lines just aren't for showing power. They tell a story in their own.

 

I did some searching and reading and it looks like many people have had broken diffusers and some engine damage from it. So maybe it's a combination of a few different things that caused the engine failure. Still no great explanation for the spikes in the data especially the InjFrSc

Tomorrow I'm gonna remove my diffusers so they don't break on me, I remember replacing a couple with some from japan so they are not old originals, or just cut the legs off of them so nothing can break off into the engine. good thread here about it all https://www.rx7club.com/showthread.p...light=diffuser

 

here's some Dyno data:

run 26 is the run before the failure and the two runs were the first 2 runs at that higher boost setting.


AFR

i am less confident re my FJO generated AFRs than the dyno. that does not mean that the dyno is correct so there is that possibility. my Datalogit AFRs are approx a half of a full higher. for example on run 27 at 7174 the dyno AFR is 11.52 and the Datalogit shows 11.98.

it is worth noting that on the last run Luke leaned it a bit around 6000, from 10.02 to 10.92 using dyno AFRs and went from 427 hp to 461, so the motor liked moving in that direction. as you can see from the AFRs the motor was by no means tuned out at the 24 psi level.

if the dyno AFR is correct i see a significantly rich tune in most of the areas.




Boost

no boost spikes recorded.



Torque

it looks like runs 26 and 27 started to diverge at 7000



as to knock. i ran 1400 CC of methanol from 2004 thru most of 2010 at 20 psi and 500 SAE with not one problem. no failures of any kind. same motor build from me as current.

nov 2010 i switched to larger solenoids that delivered 2000 CC. (we are talking theoretical max delivery so it was probably a bit less in each case). on 93 pump.

this was w my twin TO4E 46 trim turbo setup.

my new single setup is different as to knock readings. where i used to get a declining to under 10 at peak power.

with my single setup:

Advanced Knock (which i used in the past and the above knock numbers reference)

run 26

P18 (around 24 psi) starting at N12

23 22 20 29 32 30 24 22 27

p 20 boost spike reading at N19 was 17

Basic Knock

24 24 24 30 35 32 30 30 35

P20 N19 59

so the knock numbers are different V my experience 04-10. using the higher scale (Basic) i noticed these numbers on earlier runs where we were running one bar boost and were rich and figured it was just the sensor or setup because i do not believe we were getting significant knock at that point. the numbers did not change as we raised the boost and leaned it a bit.

anything is possible here, i have zero agenda other than am looking for answers whatever they may be. i do think the plugs (NGK 6725 10.5) looked a bit on the lean side and when you get down to it all the computer data while helpful needs to take a back seat to plug readings as they are on the actual frontline and don't lie.

similarly pictures don't lie. if you look at the pics of my front rotor you will see that all 3 apex seals were damaged at one end. this is where the diffuser met the rotor. i do believe the diffuser caused the motor failure. i do not know at this moment what actually was the cause of the diffuser failure.

hc

 

here's a comparison as to my FJO WB and the dyno using all data points

 

Do you have an ECU voltage log?
I notice that your EGT on the one log you show it goes flat starting from the spike. Did you log EGTs for the other two events?

Here is what I think is possibly happening:

1) Boost spike happens either due to bad sensor reading or flow effect in the intake, most likely just a bad data point.

2) ECU sees high boost, and due to your 'defensive' map, dumps way too much fuel/meth.

3) Cooling from fuel and meth makes diffuser brittle, force from max fuel flow breaks it off, motor eats it.

You could try to simulate this effect manually, but I can't see any way to test it that wouldn't likely cost you a motor. Honestly, you are better off getting rid of your 'defensive' setup, try to make it scaled appropriately, and set up a fuel/ignition cut at a certain boost limit. The nice thing about this is that it saves your motor, but also alerts you any time there is an event, so you can back off and troubleshoot.

Additionally, you can just get rid of the diffusers... Still a giant spike in flow of fluid with high cooling properties could have a quenching effect on other parts too...

 

patman you make some good points...

certainly a possibility

thnx

 

In your runs you go pretty high into the rpm range, even though you loose significant amount of power. I don't really see the point in revving the motor that high. Usually I try to stick to 7500-8000 for turbocharged street cars.

In your run 27 at around 7200 you start to loose significant power then have a few spikes and plateaus. Usually that would signify a misfire or knock then immediately you loose all power. Seen that same trace before when I've had motors knock on the Dyno.

I personally don't like flooding motors at high rpms in a effort to safeguard, reason being is you introduce to much fuel and get misfires due to either blowing out the spark or a big change in cylinder pressure which can wreak havoc on a motor. I've always been taught that running excessively rich is sometimes just as bad as running lean.

Just my .02 cents and you can take it with a grain of salt but I have built and tuned quite a few cars and have not had a failure yet.

Thanks
Zack

 

I also responded in the big thread in the single turbo section, but I will repost that here...

I took a look through the data. It's not completely clear what happened due to limited information and sample rate. You were logging everything under the monitor window, which increases the amount of data available but decreases the number of samples per second. If you had chosen only "Advance" and "Aux" you might have gotten another 5-10 samples per second which could have helped. It's hard to say exactly how high the spark got, what the knock sensor reading was, what the actual spark and EGT were. Adding 5 more samples per second would be about a 50% increase in sample rate if you think about it. I displayed the same log with "Basic" and "Advance" rpm/timing/etc. Those are sampled at different times (Boost and PIM use different units but they both relate to manifold pressure). It's all a tradeoff though.

Basic:



Advance:



It's the same basic data, but you can see that sampling rates makes this whole thing a little murky because the spike happened fast.

It's hard to tell if the engine knocked. You have a very well-instrumented car for what you are doing, but all the really nice stuff is used by big-budget facilities. The ATI Vision system for example has a 20khz sample rate. That's literally 20,000 samples per second that can log combustion chamber pressure traces, crank and cam position, and injector current down to half a degree crank angle. It costs big, big dollars though. If we had combustion chamber pressure traces we could much more easily determine. I'm only saying that to point out that you are doing a good job with what you have, and you can't be expected to fully get to the bottom of this without data from a gazillion dollars worth of equipment.

Overboost Fuel Cut

If I could give you a piece of advice, advice that a lot of people don't agree with, it's to put overboost fuel cut on your next engine. Some people feel that cutting fuel could be dangerous to the engine, based on the idea that lean = dangerous. I can tell you right now that basically every OEM uses fuel cut for overrev, overboost, traction control, or some combination thereof. The PFC's implementation of it isn't perfect, and I'm not saying that it is a panacea, but I think you would benefit from properly setting it up on your next engine. There's no way for sure to know if your overboost countermeasures in your map did the job (not enough timing retard? not fast enough?), and I'll admit there's no way to say 100% that setting up fuel cut could have saved your motor.

What I can offer you is 1) my personal anecdotal experience and 2) detailed information on how fuel cut is used in an actual OEM system. I use overboost fuel cut on my own car and it has protected my engine from boost spikes in the winter or boost control mistakes. I've probably hit it a few dozen times over the past three years that I've had it configured and it hasn't caused a noticeable problem.

Now, on most modern engine management systems the architecture is based on engine torque. To simplify: there is a calculated torque request from the accelerator pedal, a series of commands (air/fuel/spark) to generate that torque, and a calculated engine output torque used for feedback. I am going to use the Bosch Motronic ME7 system (late 90s VW etc, they are now on version 9) here but GM's and many other manufacturer's strategies are similar.



If the calculated output torque exceeds a torque limit, or calculated airflow/load/vehicle speed/engine speed or some other limitation is reached, the computer generates a torque reduction request. Torque reduction requests are also used during normal operation, such as during shifting of automatic transmissions or during cylinder deactivation on engines equipped with it.





The torque reduction request system figures out how much to reduce engine output torque and more importantly for this discussion, how fast to reduce it. For purposes of discussion, torque can be reduced, in order of slowest to fastest: wastegate position (for boosted engines), throttle position, ignition timing, and fuel cut (either selective cylinders or all cylinders).



I know that's a lot of stuff I'm throwing out here but basically what I'm saying is that fuel cut is the fastest-responding form of overboost protection (if it's configured right for the application). If the overboost was the root cause, proper fuel cut may have been the best countermeasure.

 

since this a tech interested part of the board i thought i would throw up a number of items.....

run 26 (run before failure)... as Tom indicates the secondary injector is reading 11.5 Ms ontime while the primary is at 6. the primary and secondary should be very close together as to ontime. note that the secondary injector spikes w the boost spike.

further, note Fuel Pressure... 66. (66 - 43.5 = 22.5 about right for the non spike level) not effected at this moment by the boost spike. perhaps mechanical lag? however note the next chart. Fuel Pressure at 46! (?) Fuel Pressure had been stable around 66 nominal (43.5 static) prior.



also in the weird column is injector duty cycle... it shows going to 100, as per P20 setting but after the boost spike. injector duty cycle readings are purely electric and as such should be immediate, as in the speed of light. i do have an FJO Peak and Hold module that further drives my secondaries.



here's another perspective. this is not the operative MAP as to P19 line. Luke had made another base fuel adjustment in line P19 (leaning it out a bit V line 18.) since i had a P19 reading on the log of 26.98 and the log is generally a point lower than the MAP let's use 28 Ms as the actual updated MAP P19 setting.

my PIM settings are

p18 28,500
P19 29,500
P20 30,000

so even considering we weren't at a full 30,000, the 29,960 spike reading is very close to P20 and it's 49.24 setting... but the injector ontime reads

primary 6.024
secondary 11.508

double the readings and you get only 35.06 rather than 46.77 (95% of P20 setting)



here's an eye opener... this is the boost reading from the dyno greatly magnified for run 26 at the point of boost spike.

the Power FC/Datalogit MAP shows boost spike at 7629.

note how the dyno shows two rpm reversals!

while it doesn't pick up a boost spike it it shows RPM reversals at 7607 and 7696.

the throttle was 100% and the motor did not continue to make power for those two instants. the probable cause was way too rich on the AFRs which were plummeting and were 10.6.



the second rpm reversal (7696)




run 27 differs as no RPM reversal at the boost spike



let's look at knock.

i have always used advanced and it generally has dropped w boost and as methanol enters the motor. i took another look at a 507 SAE 20 psi run i made in 2009 and the last few readings turned out to be in the same area as the Jan 16, 2012 runs.

i consider this knock level to be fine. i ran these numbers for four years, removed my motor simply to thoroughly spec it. forensics on a completely healthy motor. the motor was pulling the highest idle vacuum in 4 years when i removed it. the faces of the Atkins apex seals were pristine and the sideseal clearance had maybe opened up one thousandth. so i have no problem w the knock level based on my experience.

here's 2009 w my twin TO4 setup at 20 psi



moving to 2012, note the similarity in the 3 waves of knock! BTW, the top line is EGT... max 1600



the peaks and troughs of knock roughly correspond to swings in AFR. lower knock, lower AFR.

my guess is the early knock wave is due to pushing the turbo w the wastegate shut heading to target boost. when the WG opens there is less strain and knock reduces. eventually as the motor does more work (hp) the knock picks up. an interesting hypothesis that i wouldn't put lots of money behind

another consideration as to knock being a factor in either of the runs, and i am leaning towards it not being the culprit, is the fact i run a J&S Safeguard system. the J&S works only in boost and will retard IGL 10 degrees if knock is encountered. it will retard it on the NEXT rotor face! a Bosch knock sensor is employed and is mounted in my rear rotor housing. i have used the system since 04 and it has lit up on rare and appropriate occasions.

the point is my logs show no IGL retard and i greatly respect the unit.

the comment was made re running my engine out RPM wise wasn't getting a payback... actually looking at runs 26, 25 and 24 i see what is happening... the tune goes way rich and this is exactly where the engine power flattens and eventually declines, as well as becoming erratic. it is just way rich. i do think that the BW once properly tuned will make 550 and have a bit more midrange.

running 3000 CC of methanol, we can easily run 11.5 + AFR and we are below that towards the end of the runs. the only reason is we are in an early phase of tuning at these boost levels. maybe 3 runs. EGTs are also nowhere near where they will finish.

here's a run w zero smoothing at 19 psi which supports the rich AFRs as causing the power curve breakdown. the motor should never be below 11.5 and should probably be a tad higher.

looking at the two runs here at 19 psi:

run 24 10.95 442 hp

run 25 10.65 397 hp





howard

 

"If I could give you a piece of advice, advice that a lot of people don't agree with, it's to put overboost fuel cut on your next engine."

i appreciate your look at my MAPS LOGS...

i do think i will change P20 and go w a fuel cut. i had it setup that way a couple of years ago and while initially running my GT4094 hit the fuel cut at 30 psi.

zero problem w the motor.

howard

 

Howard
What plugs are in it?
Thinking with an aggressive port and hot plugs this could happen at those intake pressures and rich mixture.
Tom

 

NGK 6725 10.5 in all four holes.

each lead plug has an FC3000 amp and a MSD 8253 coil.

Magnecor 10 mm wires.

msd 8207 coil on each trailing plug.

we should be able to burn water w the setup

hc

 

Hmm very envious... Possibly the plugs are setting off the fuel mixture, with the arc welder you are set up with the plugs may be glowing well after the spark event.... Possibly try a couple settings colder..

 

Hello Howard,

first of all I have to excuse that I`ve not read all of the posts above..

but I want to tell you an interesting thing I´ve discovered...


As you know there are a lot of Software functions in the PFC we don`t know 100% what they are doing. Unfortunately we only have the "code to text" translated information in the FC-Edit software which tells us at least a little bit about what we are changing...

f.e. IGN vs. AirT

well we think it reatrds timing the value we put in at a temperature we write there... and well... maybe someone really tested it if it works... BUT we don`t have the C-Code to have the evidence that it`s working like how we think or how it Exactely works...

As an Engineer @ Merc, working on ECU Software algorithms I´m used to work with software someone else (bosch, etc.) has written... but at least we have a documentation how it really works...

the PFC doesen`t give us such a Documentation...



Enough text... here are the informations.


I´ve mapped a friends Car (stock twins, 2x550 2x850 injectors, 4x9er (colder) plugs, 3"exhaust, airfilter, larger intercooler and AI )

We first tuned cruise and idle then we made a nice 1turbo tune.
perfect boost builup for his desired 0.75bar boost.

Then we went to take a look at the second turbo and... holy **** we had boost spikes and a really really oscillating boost and WG% values ...

so I´ve changed duty and boost values... but there was no real success...


and then as you know... If you don´t know what to do, you´ll make things you are 100% sure that it won`t help... and then it helps...

I`ve used 2. boost setting in the FC-Edit instead of 1. with THE SAME bost and duty settings..

And all of a sudden... it worked..

so my conclusion is that it might not only be a different setup you can store and change with a button, like a race or street setup.. it seems to me that there are different boost controlling PID parameters which are used on setting 1 and on setting 2...


It seems to me that setting 1 has higher P-values and with a fast reacting setup this controller will be unstable..


maybe you encountered such a problem..

good luck!

regards
Marc