Cold start seems a bit...off SD

[Jim Blackwood]

Right on board with everything you just said. I migrated over to HPT from Megasquirt, was there from ver.1.0 so I get what you are saying, and my VE and advance tables are really pretty good. I've been logging IAC since the beginning, it is steady now and runs in the 20-40 range at idle. Had to do a little tweaking in the PID section to get it there but hey that whole thing is a mind twist but I can sort it if I study it long enough. Adaptive ign timing was another challenge but is working fairly well now. I drove the car all last summer, got not great mileage but plenty good enough for this stage and ran smooth if just a smidge rich. I'll trim that down in due course. Starting has never been good but was tolerable. With your suggestions it's now clear that it can be as near perfect as it gets.

Which leaves warmup and here we are. So...

Studying the datalog (attached as screen print) I see that the injectors went WAY rich for a very short interval at initial crank and the engine fired off immediately, which is good. Then they shut down almost completely which is not so good. The engine ran long enough to burn up the initial flood of gas, idle dropped off, injectors came back, but not fast enough to prevent a stall. This was the result from setting Startup Airflow Decay [ECM] 2172 to .400 from .800 and tomorrow's first test will be to set it to 1.200.
My best guess is that it will change that 3 second drop off to several closely spaced pulses and the engine will continue to run but probably roughly. Or it could smooth out. Find out tomorrow I guess. I will whittle down that initial pulse once I have the next few seconds smoothed out and it is free from oscillation. That's really the focus at this point. btw, I have extended the adjustments that gave me that nice fast light-off to the rest of the 3 tables involved. They basically appear to be leaner than your comparable tables though so I'm not going to get in a hurry about that, I think I'm in the ballpark.

Jim

[kingtal0n]

Once the engine fires your primary source of enrichment to keep the engine going is the stoich table. Multiply larger values 1.10 to 1.20 range into that table for cold ECT situation will run the engine

There are other maps which can support cold start like the ECT adders and so forth elsewhere but most are timed or in relation to crankshaft revolutions and utilize a decay. Which is all well and good but for fine tuning later- for now just use the global stoich to keep the engine running because it is simple and easy

actually it looks like your injector is going to 0 after the engine started. Perhaps it is hitting a DFCO or something? I would increase resolution of injector graph and take notice of the exact ms the injector is being told, and work that out how it is arriving to that value. Your resolution is too low for me to tell what is really happening.

Fix the scale on all those data collections. For example wideband a/f should be say 20 to 10, not down to 6.
Injector ms should not be 60, either. That is very strange. I Know sometimes hptuners reports weird values but 60 seems ridiculous. What is the ms of the injectors while at idle? cruise? Should be near like 5ms for a cruise and like 1-2ms for idle.

[Jim Blackwood]

Here is the datalog (screenshot) of the last test with the 1.200 value. Note the oscillations in the PW, and note the area where it is different from the last one above, all of which is from changing the Startup Airflow Decay. That isn't the right parameter to use in correcting this, given the results. I changed it back to the original figure of .8 after trying various values. Note that the PW varies from full off to full on basically. That is the region of the startup trace that I am now trying to influence. But what I do now know is that decreasing the Startup Airflow Decay below about .5 causes the pw to drop off long enough for the engine to die. Anything above that gives me those ultra-rich pulses and larger numbers stretch out the overall time. Your Friction airflow decay is about .25. So, if I set that value lower, I have to add in fuel somewhere else. That may well be the solution, if so then the question is where and how much.

Startup airflow is about 1.5 and about 6.0 for friction airflow, if this means anything to you. Your values are about 2.1 and 8. Doesn't seem very far off. The delay figures seem reasonably close.
All this leads me to think perhaps I should try to turn off this function entirely for now or decrease it as above, and I just wonder what other automatic functions I could do the same with until the base parameters are set properly. This one at least I know for certain has a dramatic influence and not in a good way.


Your reference to a stoich table is confusing to me unless this is the adders vs ect and iat and maybe some other related tables. Can't be the main VE table since there's no temp info. I'll be happy to do that, maybe in conjunction with the above. Can't see success happening as long as something else is interfering. I've long been looking for a global cold compensation table but don't seem to be able to find it for some reason.

I am running 26Lb OEM Delphi injectors for '02 Silverado/Camaro. The injector times looked pretty reasonable to me when the engine is in steady state running. (which you haven't seen yet) Not during this startup cycle though. I'm willing to accept the need for greatly excess quantities of fuel during startup and even initial warmup for now, but expect to be able to trim those back once I have stable operation. Not there yet. Getting it to light off quickly was a big step.


As for changing the scanner units, yes that's something I need to do but it's going to take 2 or 3 days to get it done. I already have gone to g/s in the VCM editor and my scanner needs to be made consistent, no argument. But I'd like to put that off while this cold snap is in town, at least until I've got the engine starting properly. I will look at it though. The AFR range was set to encompass my AEM WB

Any guidance on a global temp table?

Jim

[Jim Blackwood]

I had hoped that by zeroing out the airflow Initials and the airflow decays (both startup and friction), and then raising the levels of the Afterstart Enrichment Initial Adders vs ECT and IAT that I could raise the floor on the idle inj pw and do away with the rich spikes. As the attached trace shows, it wasn't to be. Last preceeding test I had gotten the pw up to 1.3 but shown here it is back down to .3 and spikes to... I show it going as high as 476, I didn't even think that was possible. Ny AFR with these pulsations is now rich but clearly that is all coming from the rich spikes. I'm at about 2.0 on my ect and iat multipliers.

I'm kinda stumped. You mentioned DFCO and I thought I had that disabled but maybe not. Any suggestions there or anything else that might be causing this which I could disable if only to check what's doing it?

Thanks,
Jim

[kingtal0n]

I had hoped that by zeroing out the airflow Initials and the airflow decays (both startup and friction), and then raising the levels of the Afterstart Enrichment Initial Adders vs ECT and IAT that I could raise the floor on the idle inj pw and do away with the rich spikes. As the attached trace shows, it wasn't to be. Last preceeding test I had gotten the pw up to 1.3 but shown here it is back down to .3 and spikes to... I show it going as high as 476, I didn't even think that was possible. Ny AFR with these pulsations is now rich but clearly that is all coming from the rich spikes. I'm at about 2.0 on my ect and iat multipliers.

I'm kinda stumped. You mentioned DFCO and I thought I had that disabled but maybe not. Any suggestions there or anything else that might be causing this which I could disable if only to check what's doing it?

Thanks,
Jim

Airflow stuff is airflow. It should not change injector on-time.

You seem to have a setup error somewhere. Did the maf fully disable? Please tell me there is no maf sensor.

The fueling should be very simple situation. Car starts, looks at VE table, there is your primary source of fuel. The VE table needs to make sense. Is your Idle VE like 50%, it should be. You can temporarily disable all the cold start stuff and jack up the VE table to get a cold start if you wanted.

Next the modifiers, cold start mods are slight. Values slightly over 1, like 1.04 or 1.12 are useful. Adding 10 or 15% fuel overall to the VE table.

That sets the dynamic range for cold situation, VE table + cold start modifiers. You must have a 2.0 somewhere instead of a 1.2 for example. Some huge number in the wrong spot maybe.

Another possibility is a corrupt file or incorrect segment swap type of situation. I don't know the specifics but I do know when I load old tune files sometimes my file corrupts and i Get weird behaviors like that. If you don't see anything obviously wrong in the file maybe try starting over with a new file or somehow ensure the segment is correctly swapped. You could also try 'read entire' or 'write entire' to make sure what you are looking at actually matches what the computer is thinking.

Your injectors are very very small so the tuning should be super easy. I am surprised to see this kind of difficulty. It makes me think there is a major setup error somewhere or a wrong value like 'wayyyyy wrong' misplaced. 400+ injector on-time is like a bug almost. Look for negative values in your enrichment tables, I had a negative valve of -0.02 that caused 4000% Injector duty cycle once.

See where it says 'air fuel ratio commanded: 3' that should never happen. Why it is commanding 3:1 air fuel ratio? There is a very bad wrong number somewhere or a corrupted data.

Also your previous post says "heres a picture" But there was no picture to look at.

Make sure DFCO is disabled. Make sure Closed loop is disabled. Turn all that stuff off for now. Find the Open EQ table (open loop EQ) or whatever its called.

[kingtal0n]

Upload your current tune file and I'll try to look it over today

Perform "read ENtire" before you send it though, if you can.

a word of caution also: Whenever doing "Read" or "write" and especially when "entire" You need to make sure
1. battery is powerful, charged, both of them (laptop & car)
2. The cables are secure, fully attached and stable, good quality

and finally most importantly,
once the transfer begins,

3. Remain absolutely still and do not touch anything!

Don't mess with the brake pedal or touch the steering wheel or get out of the car or shake the car or move the laptop or anything at all. Just sit there remain motionless or gtfo of the car before you even begin the transfer. I set mine to 'go' and watch it from 10 feet away until its done.

These type of electronics are very sensitive in my experience, to every minor perturbation. The newer ones are not so bad; but back in the day, 10-20 years ago man you could tap the little box or move the cable and wind up with serious corruption in the files.

[smokeshow]

Lol. Well you don't see that every day. Put your transient fueling back to stock...that impact factor is causing the spikes. Current cal is telling the PCM that 100% of the fuel is impacting the port wall below 50?F and above 50kpa. So naturally its compensating for it. Have to be careful with that table...the way it is calculated makes it sensitive to large adjustments.

[Jim Blackwood]

Guys, I can't thank you enough for the help you have given me with this. Check out the latest trace. Pretty normal startup for a change. Needs some help, it's too rich and I had a double bump to fire off, but so much closer!

For that run I had leaned out the ECT/IAT curves considerably, down to .245@45 degrees and it's still rich so I'll go a few more steps before looking at anything else. For the double bump I hope I can use the prime pulses to fix that but I'll wait until the afterstart is leaned out.

SUCH a dramatic improvement. Can't tell you how good it feels.

Jim

[Jim Blackwood]

Here's the latest startup. Nice quick light off, settled down to a steady idle pretty quickly, I like it. 37 degrees ect so that's promising.

King, I have questions about the Cranking FA Multiplier tables, or table. I can't do a direct comparison with yours since I have 2 tables and you have 1. (Wouldn't think that had anything to do with the 2 bar tune I'm using but who knows?) So I'd like to ask you about the upper reaches of the table and some other fine points.

-On my tables above 176 degrees the numbers drop to 0.62 and that populates the table from there up and all the way down the vertical axis as well. Is this normal? I suppose numbers less than 1 decrease startup fueling but why would it drop that much?

-At 32 degrees I'm showing a multiplier of 16 and that seems to work just fine but at -4 it's up to 24, does that seem like reasonable numbers to you? At -40 I'm showing 30.

-From what I read there is often a cyclic character to the vertical axis. As I understand it, as cranking progresses the numbers decline, then climb again, possibly to avoid flooding the engine. Have you seen this? I wonder how fast it should cycle and how far? I have no way to know how long the ECM uses each line before moving down. Should the cycle take 8 lines, 4 lines, etc? and should the multipliers vary by 10% or so? 20%? Just asking for your thoughts on this. Obviously since I have the double tables I can put in twice the number of cycles. But as long as it fires right off it'll only ever use the first couple of lines.

Jim

[kingtal0n]

Here's the latest startup. Nice quick light off, settled down to a steady idle pretty quickly, I like it. 37 degrees ect so that's promising.

King, I have questions about the Cranking FA Multiplier tables, or table. I can't do a direct comparison with yours since I have 2 tables and you have 1. (Wouldn't think that had anything to do with the 2 bar tune I'm using but who knows?) So I'd like to ask you about the upper reaches of the table and some other fine points.

-On my tables above 176 degrees the numbers drop to 0.62 and that populates the table from there up and all the way down the vertical axis as well. Is this normal? I suppose numbers less than 1 decrease startup fueling but why would it drop that much?

-At 32 degrees I'm showing a multiplier of 16 and that seems to work just fine but at -4 it's up to 24, does that seem like reasonable numbers to you? At -40 I'm showing 30.

-From what I read there is often a cyclic character to the vertical axis. As I understand it, as cranking progresses the numbers decline, then climb again, possibly to avoid flooding the engine. Have you seen this? I wonder how fast it should cycle and how far? I have no way to know how long the ECM uses each line before moving down. Should the cycle take 8 lines, 4 lines, etc? and should the multipliers vary by 10% or so? 20%? Just asking for your thoughts on this. Obviously since I have the double tables I can put in twice the number of cycles. But as long as it fires right off it'll only ever use the first couple of lines.

Jim

I noticed the huge numbers for cold temps I would guess they are there because the cold is also going to affect cranking voltage (injector influence) and fuel vaporization so they just want to make sure plenty of fuel will go in. But i've never had cold temp that low and never tried it.

I Dont remember If I have top row numbers less than 1.00 that is kind of weird to me. I haven't looked in the file for months. You would think 1.0 from the VE table + the prime pulses would be the 'right' amount of fuel, at least I would think so. But you have to watch out because there are like 10 different tables influencing fuel for cranking, theres a Cranking VE table specifically to look for, that tells like for 200rpm 300rpm 400rpm what sort of VE multiplier to use.
Also remember when the engine is hot and run recently there will be fuel leftover for a while hanging around somewhere. Often you can get an engine to briefly sputter once its been run and just shut off, even without injectors plugged in, know what im saying? So wherever you are seeing .62 or whatever its likely a high enough temp the motor was just run that it has some fuel sitting around and that fuel + the 62% pulse = 1.xx


The table vertical axis is a mystery what the units are. I think its in seconds though, cranking seconds perhaps. You could try putting 0 's for the first 1, 2, and 3 seconds and then crank and see if it takes 4 seconds to get the engine fired if you want. I just never cared that much to test it.

Yes I set mine up to cycle more and then less and then more fuel just in case it keeps cranking, you don't want to flood the engine forever with all large numbers IMO. Of course there is a clear-flood protocol, all ECU have that, but if you clear flood and get a pop from the engine then let go back to cranking and it's got some big number it will just flood again. You can put way more fuel into a motor than you can match to air just by cranking with an open throttle body, so generally once the engines been cranking for 4 or more seconds you (the programmer) have to assume something is wrong and the engine didn't light off for some reason, If that reason is because it was rich then you want those .4's and .6's to roll even after a clear flood. And if not from being rich then the following vertical cell has a number larger than 1 or near 1 to give it the full juice.
Very cold situations cause fuels to posses a low vapor pressure, which means the rate of fuel turning to a gas (fuel is injected as a liquid to cold engine parts will sit around as a liquid more often) is slow, and this factoring into cold start because if you have a very low vaporization rate (think of ethanol at -20*F) then you need to put in a very high surface area of fuel liquid to coat every engine part quickly so it can evaporate fast. A pump shot of alcohol and then wait 2 minutes for it vaporize for example and gas state fill cylinders and intake ports. Then finally crank the engine and get heat input for the next cycle. As heat builds up the rate of vaporization of fuel increases. Fuel is cooling the valve and intake and head, and it can get so cold as to form ice on the intake parts when alcohol is vaporizing rapidly. You probably already knew that though


But ideally it will fire up instantly so there is no reason to even use those numbers unless something happens like a sensor going bad or fuel pump dying or some weird thing where fuel pulse or maybe bad gas, when fuel supply becomes erratic and you just need to make it home. That kind of what I feel those extra lines are really for, getting home and figuring out why it made it past the first line. It's like from 2 down means "something is wrong" so now you have to think like a diagnostic and not just trying to start the engine anymore with a perfect number, there is no ideal or perfect number once the engine failed to start immediately because every situation will be different for when that happens. Best just don't flood it though, or think about cyclic in nature to avoid flooding, is what I thought they meant by it. When an engine failed to start, the battery is in jeopardy, I will prefer only put a final new battery in a car once cranking fuel is setup because trying to figure it out can tax a battery life span.

I have many of the same questions you do and never really get any answers, but it doesn't matter because you can still find the ideal setting for instant starting once you play around with it for a while.

To really dial in the engine RPM climb and recede behavior after a startup you would use those friction and delay tables for airflow mostly. You need the engine to start and rev up kinda quick like all OEM engines tend to do to get the oil flowing really quickly, build oil pressure fast. So when starting don't try to limit the air to make it idle fast, let it rev up nice and come down gradually.

[Jim Blackwood]

Thanks, that all kinda makes sense. Delving into unknown areas always has been a challenge, of course these controllers have a LOT of unknown areas for many of us. Maybe most of us. I'm certainly in that group. But hey, I've just got to say that as it sits right now my startup behavior is WAY ahead of anything I was EVER able to get with any version of MS or even with the old SN95/TwEECer. And it can only get better from here.

Next I'll be looking at the 2014 ECM used in the Caddy CTSV-Sport. THAT should be interesting.

Jim

[smokeshow]

The crank fuel mult y axis values are piston cycles/power pulses. They represent how many cylinders have received fuel following crank sync.

[kingtal0n]

The crank fuel mult y axis values are piston cycles/power pulses. They represent how many cylinders have received fuel following crank sync.

When first power the ECU and crank the engine it has no cam location signal. Cranking with no cam sensor for example.

The computer will still inject fuel and guess a cylinder, causing a backfire when it guesses incorrectly. Or maybe it wanted to explode fuel on the exhaust stroke for some other reason (it never seems to guess 'correctly')


I always figured the ECU would resort to batch fire mode during cam-loss cranking before startup.
Because how can it specify which injector to fire if it doesn't know where the cam is? Wrong cylinder might receive fuel and it wants to test a spark.

After thinking about this some more, maybe the ECU makes a guess which specific cyls to spark and fuel but if that guess was wrong it would just be filling the exhaust with fuel on the wrong cylinders for an untold number of revolutions. I suppose this is why you get the magnificent backfire at the end when it tries a spark to the exhaust stroke. It also begs the question of why not waste-spark strategy? And whether the fuel injected sequentially to wrong initial cylinders is enough to quickly start the engine once the correct spark is sent on the second cranking event, (crank -> Backfire -> pause -> crank -> Start) at which point is the y-axis starting over from #1 on the table on the 2nd crank attempt? Or does it remember previous event input somewhere down on line #8 or whatever from when it backfired.


Fueling strategy leads to a slew of questions regarding the y-axis of a table such as this.
Batch fire for example means 4 cylinders will receive the same amount of fuel after the first crank trigger.
Does that mean it will jump from #1 to #5 on the y-axis? Seems unlikely.

Next question is maybe instead of batch fire it simply injects fuel to both possible cylinders (One on exhaust and one on compression).
It must at least do this because otherwise we would not get a backfire during the guess spark.
Therefore, at least two cylinders receive fuel after 1 single crank trigger event.
Again, does that jump us from #1 to #3 on the y-axis?

So we could be anywhere from #1 to #2 to #3 to #5 after a single crank trigger event depending how it regards fueling to each specific cylinder.
Furthermore it could be dividing the fuel by both cylinders, say it jumps from 1 to 2 but the value in 1 is "8" maybe it will send "4" to each of two cylinders that could be about to have a compression stroke.

And even more possibilities once the cam signal is detected, each cylinder can then sequentially receive a specific dose from the table. But how many y-axis values did it move before finally resorting to sequential mode?

So many ways it can happen I am still unclear

[Jim Blackwood]

Also really unsure about that. Based on how quickly it lights off when it is right maybe they gave fuel to both cylinders and fired both? I mean the raw fuel in the off cylinder wouldn't matter if the right one fired and it should speed things up. Once the cam comes around it could drop off the wrong ones. That's how I'd do it but what do I know.

I could use a little more help on that transient table I just copied King's table, this is: Engine /Fuel/Transient/Transient Fuel Mass Gain/ Fuel To Wall Impact Factor. Please understand this is my first time delving into this table, I wasn't the one who set it up with numbers>1. Anyway right now the engine runs great and I've been leaning out the VE table and fine tuning the Engine/Fuel/General/Cranking Fuel/Fuel Air Multiplier table and the Engine/Fuel/OpenLoop-Base/Cold-Warmup/After Start/Initial Adders vs.ECT curve and that is going just fine. Better than I hoped for really except for one thing.

Below 1600rpm and above 80kpa the engine suddenly cuts out. Now this only happens in a high gear at near full throttle and very slow speed so it's not a serious driveability problem, but it IS within the normal operating envelope for this engine and I'd like to get it right. I've tried going both richer and leaner on the ve table with no real change so I'm beginning to suspect the transients table. Does that make sense? It seems to be the same sort of cutting out issue with the injector pw acting up. Values in that area of the transients table are .750 and above. I'll try decreasing that and see what it does. I'm including a screen shot of the scan and the transient chart. This tune originated from a modded tune by a 3rd party so it is likely to have a few weirdnesses about it, but it seems to be getting pretty close to what I need.

Anyway I'd appreciate your thoughts on the transient numbers.

Jim

[Jim Blackwood]

I changed the values in the bottom three rows to 0.650 (matching the 70kpa row above) and that did it, problem went away. I can now lug it down to under 300rpm and recover just fine. So a question:

What does it do if the value is set too low?

Also, along the same line, I have a stumble right off idle when I let out the clutch to get moving, well maybe not a stumble so much as a failure to immediately take the throttle like it does everywhere else. Is that also the transient table doing that? It's under very light application of the throttle, just a gentle caress in fact, but in the datalog I'm seeing a very slight dip in both PW and IAC. Thoughts?

Thanks,

Jim

[kingtal0n]

from the log looks like pulse is going to 0.4 and spark is going to 0 that seems like a DFCO timing and fuel remove rated

your changing the fuel tables wouldn't have timing dropping down to 0, am I reading that right?

[Jim Blackwood]

Let me post an expanded shot. Right at the cursor line you can see a small bump in the TPS beginning. At the same time there is a dip in the IAC. Spark advance is climbing from 8 to about 27 and eventually 31. PW dips slightly from 5.5 to 5.0 before starting back up. This occurs at time 12.830 to 13.003 where the IAC also starts back up. So we're only talking about a bit over a tenth of a second here.

My best guess is that something is causing the PW to dip when it should be going up.

Jim

[smokeshow]

As I already mentioned...set transient fueling back to stock. Just go do it. You literally risk engine damage with the table set with those values anywhere near 1.0.

[Jim Blackwood]

Thanks Smoke, that seems to have done the trick. I had to search back and find the original Silverado file, didn't realize King's table had been edited. Numbers look better and it drives better. Of course, this isn't an LS so it might not be ideal and I would pay close attention if you were so kind as to tell me what symptoms numbers that are too low can cause, etc.

Would your advice be the same regarding the DFCO settings? I disabled those while sorting out the original problem but now find an odd stutter when I suddenly let off the gas.

Jim

[smokeshow]

I'd set DFCO back to stock as well...DFCO doesn't enable suddenly with stock cals so its probably not causing that issue. Better to leave things alone if they're not related to the issue.