Deatschwerks 740cc Injector's Thumbs Up!

[bugblatterbeast]

This seems awfully close to what I'm getting after a few minutes of logging. I suspect latency will have to be tweaked as well as the scale seems to vary with IPW.

 

Assuming DW's injector dyno is consistent - what did your test sheet say for flow rate?

 

739 to 741

[bugblatterbeast]

Anyone tune tip-in with a TMIC?

 

yes.

[Deer Killer]

My experiment in latency based on bbb's method. Assuming IPW includes latency. It was unclear if AFR correction was applied before or after IPW is reported, so I did both. Common sense would dictate that latency would go up. The intercept below is predictable and seems sane. This is the second dataset, about 7000 datapoints after rigorous filter. It approximately agrees with my first round. Middle trend is without AFR correction. Where the red line cross the X axis is what's interesting. Average latency was 0.672ms, intercept of non-corrected trend is 0.682ms. So the apparent error of this method is small. Final number: 0.733ms. Probably should have grabbed voltage, but we can figure that out easily, 13.94V. This seems like a reasonable value.

 

I idle OK at ~0.8ms IPW, so if I were to use bbb's latency of 1.2ms, I think it would not work very well. Right?

 

Cold start appears to be fine, so I might as well just scale all the values by 8% and tune from there.

 

How I filtered: Log AFR, OL/CL status, throttle plate position, and wideband AFR. Filtered by CL status, then by 14-15 AFR, then by 14-15 wideband AFR, then to exclude large throttle inputs, and then manually trim outliers without enough datapoints after making the plot.

[bugblatterbeast]

Hey Deerkiller,

 

Your pm got me thinking so I dug up my notes as my current values have changed since the ones I posted. The initial posted values were done before I had a fpr problem and swapped the assembly.

 

I've since retuned. I'm on the road right now so I can't dig them up till the evening. I'll try to get them posted later.

 

I've also refined the logging method a little. Afr needs to be leaner than 14.4 for the data to be good as the enrichment threshold is set to 14.4. I didn't realize this till after I opened a dump of the maps in enginuity.

 

Bbb

 

My experiment in latency based on bbb's method. Assuming IPW includes latency. It was unclear if AFR correction was applied before or after IPW is reported, so I did both. Common sense would dictate that latency would go up. The intercept below is predictable and seems sane. This is the second dataset, about 7000 datapoints after rigorous filter. It approximately agrees with my first round. Middle trend is without AFR correction. Where the red line cross the X axis is what's interesting. Average latency was 0.672ms, intercept of non-corrected trend is 0.682ms. So the apparent error of this method is small. Final number: 0.733ms. Probably should have grabbed voltage, but we can figure that out easily, 13.94V. This seems like a reasonable value.

 

I idle OK at ~0.8ms IPW, so if I were to use bbb's latency of 1.2ms, I think it would not work very well. Right?

 

Cold start appears to be fine, so I might as well just scale all the values by 8% and tune from there.

 

How I filtered: Log AFR, OL/CL status, throttle plate position, and wideband AFR. Filtered by CL status, then by 14-15 AFR, then by 14-15 wideband AFR, then to exclude large throttle inputs, and then manually trim outliers without enough datapoints after making the plot.

[Deer Killer]

I've also refined the logging method a little. Afr needs to be leaner than 14.4 for the data to be good as the enrichment threshold is set to 14.4. I didn't realize this till after I opened a dump of the maps in enginuity.

 

Bbb

 

I'm not sure what you mean or how it effects this data, but if I trim the dataset, I lose about 3.5k/7k datapoints, however the final results are less that 0.7% different, 0.728ms.

[bugblatterbeast]

3654. 7.74/2.78/1.49/1.2/0.84

 

The new values were generated after fpr replacement and are an average of my winter and summer values. The low voltage latencies were deliberately left at the stock values as starting was easier with them lower than the real values.

[Deer Killer]

Your values are very close, but I think you're being conservative on the latency. At this point with 1.2cobb/0.78ms I'm pretty damn close, I could either re-adjust the MAF scale or just add a single point to the latency. The mathematical method came up conservative as usual because of noise, after the first run it was still a bit shy. Also when I get my latency this close, my injector scaler comes in slightly above 742cc/3648. All seems to make sense! Stock latency = wrong.

 

Cobb seems to have the scale right, but the latencies are wrong, hence they recommend adjusting the MAF scale. You should _not_ have to readjust the MAF scale for injectors assuming you already tuned it. Otherwise you're "doing it wrong"

 

I don't understand why higher latencies would cause hard starts. Unless it's just too rich.

[bugblatterbeast]

my car has always over fueled during cranking. even on stock injectors the car would bog slightly before ramping up to the cold idle speed. when the new injectors went in, I decided to correct the problem

[Deer Killer]

my car has always over fueled during cranking. even on stock injectors the car would bog slightly before ramping up to the cold idle speed. when the new injectors went in, I decided to correct the problem

My car sometimes cranks too long, probably twice as long as I'd expect. I thought this went away after tuning the intake (rich) but I suspect it's back after the injectors. I assume this mean lean, I have no other explanation.

[bugblatterbeast]

that's exactly what happened after I put in the injectors. It took a while to catch the hard starts with a datalog, but when it was all said and done, I found correlation between the supply voltage dropping (thus using the low voltage latencies) and the super tough starts. it was a bit of a chicken and egg problem. the car would start a little slower than usual and pull the battery voltage lower, causing the mix to go really rich (due to too large a latency setting), then it would take even longer to start and pull the voltage even lower.

[fweasel]

The hard starting issues sound like what I experience with my DW740's. I believe it to be voltage related and only seems to happen in the warmer months. After the car has sat for 1-2 hours, it may take as long as 15seconds to crank and start the car. Difficult for me to diagnose because I did not tune the car myself, and difficult for me to reproduce for my tuner due to the infrequent nature.

 

My tuner used the standard settings Cobb listed for these injectors as they were relatively new at the time. If I were to try a shotgun approach and ask him to change something in my tune and then live with it for a while to test the results, where would I start? All other aspects of the injectors and tune are spot on.

[Deer Killer]

The hard starting issues sound like what I experience with my DW740's. I believe it to be voltage related and only seems to happen in the warmer months. After the car has sat for 1-2 hours, it may take as long as 15seconds to crank and start the car. Difficult for me to diagnose because I did not tune the car myself, and difficult for me to reproduce for my tuner due to the infrequent nature.

 

My tuner used the standard settings Cobb listed for these injectors as they were relatively new at the time. If I were to try a shotgun approach and ask him to change something in my tune and then live with it for a while to test the results, where would I start? All other aspects of the injectors and tune are spot on.

This is _not_ the problem I am talking about. There are threads the long crank issue elsewhere about stock cars (long ~=twice as long new). This seems to have started after the 10% ethanol mandate, but that may be coincidence.

 

If you use the cobb settings for the DW740's and haven't artificially inflated the MAF scale < 1.21 volts, you will have a problem, and maybe even if you have.

[Deer Killer]

that's exactly what happened after I put in the injectors. It took a while to catch the hard starts with a datalog, but when it was all said and done, I found correlation between the supply voltage dropping (thus using the low voltage latencies) and the super tough starts. it was a bit of a chicken and egg problem. the car would start a little slower than usual and pull the battery voltage lower, causing the mix to go really rich (due to too large a latency setting), then it would take even longer to start and pull the voltage even lower.

Define rough, I image a rich start as "stumbly". I would assume this would be obvious to me as being rich.

 

Speaking of starts I've found the coolant temp enrichment to not do anything I expect, with TGV deletes and DW740's, the car starts fine, but has a small hiccup apparently a fixed time in the future when it moves into closed loop, but that's a different thread.

[SeeeeeYa]

This is _not_ the problem I am talking about. There are threads the long crank issue elsewhere about stock cars (long ~=twice as long new). This seems to have started after the 10% ethanol mandate, but that may be coincidence.

 

If you use the cobb settings for the DW740's and haven't artificially inflated the MAF scale < 1.21 volts, you will have a problem, and maybe even if you have.

 

I used the Cobb setting and I didn't/don't have any such problem. It seems to me after reading this and other threads the problem lies elsewhere. My suspicion is that it is exactly what the data indicates it is, a voltage problem.

 

Of course, if you already have an Optima battery that blows that theory away .

[bugblatterbeast]

This is _not_ the problem I am talking about. There are threads the long crank issue elsewhere about stock cars (long ~=twice as long new). This seems to have started after the 10% ethanol mandate, but that may be coincidence.

 

If you use the cobb settings for the DW740's and haven't artificially inflated the MAF scale < 1.21 volts, you will have a problem, and maybe even if you have.

 

I'm now tuned on 10% EtOH as that is the only high test fuel I can get consistently. The scale value changes a bit but not enough to cause huge startability problems. I can fill up on non oxgenated fuel and still start just as quickly. If I had to guess, I would pick worn spark plugs. Every car I've owned had trouble starting with the EtOH blend if the spark plugs were the least bit worn (I'm referring to < -30c winter starts)

[bugblatterbeast]

Define rough, I image a rich start as "stumbly". I would assume this would be obvious to me as being rich.

 

Speaking of starts I've found the coolant temp enrichment to not do anything I expect, with TGV deletes and DW740's, the car starts fine, but has a small hiccup apparently a fixed time in the future when it moves into closed loop, but that's a different thread.

 

Not quite stumbly but labouring a bit.

 

If you've deleted the TGVs, you shold remove the kink in the coolant temp trim curves around 68-86f by raising the values below and equal to 68f.

[Ridgeracer]

One more time...What does a stumbling cold start indicate??

[LittleBlueGT]

Not quite stumbly but labouring a bit.

 

If you've deleted the TGVs, you shold remove the kink in the coolant temp trim curves around 68-86f by raising the values below and equal to 68f.

 

You mean (in ST):

 

-Coolant Temp Fuel Enrichment

or

-Fuel Injector Coolant Temp Trim??????????

[Deer Killer]

You mean (in ST):

 

-Coolant Temp Fuel Enrichment

or

-Fuel Injector Coolant Temp Trim??????????

 

Yeah please verify which table in what units, etc. There are 3 tables available in enginuity, one in percent, two and AFR in debug mode only. The two in AFR seem to make no difference.. The car starts in the 11's and slowly creeps up until it decides to go into open loop. This does not seem to depend on coolant temp, but time. The one in percent makes no sense to me, the JDM/AUSDM tables are leaner apparently (measured in negative percent)

[Deer Killer]

This table? As you can see by the JDM curve, this really makes no sense to me. Your suggestion is "no tgv", bbb.

[bugblatterbeast]

This table? As you can see by the JDM curve, this really makes no sense to me. Your suggestion is "no tgv", bbb.

 

that's strange. the curves are inverted in enginuity. we might be talking about two different calibrations.

 

LBGT, it was the injector coolant trim that needed some fiddling with.

[LittleBlueGT]

that's strange. the curves are inverted in enginuity. we might be talking about two different calibrations.

 

LBGT, it was the injector coolant trim that needed some fiddling with.

 

This is a pic of my Fuel Injector Coolant Tamp Trim in STAD:

 

http://i248.photobucket.com/albums/gg194/littlebluegt/FuelInjectorCoolantTempTrim.jpg

 

 

What kink you talking about?

[Deer Killer]

None of this makes sense to me, *and* starting the car up today in 40F weather, it seems that the ECU decides on a delay value based on the starting temp, then gradually scales from ~11AFR to closed loop in this time period. I see no influence from these tables..

 

JDM appears to run leaner without TGV's. Maybe they just don't cold idle well?

 

Although, the the minimum open loop enrichment idle (coolant temp) table seems to work. I appear to now idle at 13.99AFR.. which means smoother. Copied in the AUS table

[Deer Killer]

I believe I had a eureka moment. It wasn't the startup enrichment that is the problem, it's entering closed loop too soon and can't get a good enough mixture to do this in a stable way. It's already too rich IMHO.

 

Left all the coolant temp enrichment tables alone except for "Minimum open loop enrichment idle (coolant temp)". I changed the 0.70 load row to 13.34 AFR up to the 68 degrees column. The car went from very rich to 13.34 AFR quickly, usually at this point I'd expect idle to start bouncing around, but instead it stayed at 13.34 AFR all the way until 93 degrees (~15 seconds after is crossed into unenriched vales). Smooth, no problems. You wouldn't guess it was modifed. Leave the other tables alone!

[LittleBlueGT]

I believe I had a eureka moment. It wasn't the startup enrichment that is the problem, it's entering closed loop too soon and can't get a good enough mixture to do this in a stable way. It's already too rich IMHO.

 

Left all the coolant temp enrichment tables alone except for "Minimum open loop enrichment idle (coolant temp)". I changed the 0.70 load row to 13.34 AFR up to the 68 degrees column. The car went from very rich to 13.34 AFR quickly, usually at this point I'd expect idle to start bouncing around, but instead it stayed at 13.34 AFR all the way until 93 degrees (~15 seconds after is crossed into unenriched vales). Smooth, no problems. You wouldn't guess it was modifed. Leave the other tables alone!

 

I never could figure out why that table was the way it was, I guess it is for the TGV, but once they are gone it is sub-optimal.