I've been working on characterizing the injectors on my TPI setup and tuning for the "hot soak" enleanment that seems to plague the TPI setup. I'm just wondering if anyone has a theory to explain the ridiculous open-loop hot-start enrichment these require. I find that after a 20 minute heat soak I need about 20% enrichment that has to taper off after a couple minutes back to my normal fuel map. I've looked at stock $6E and $8D, and both have ridiculous amounts of post start (open loop) enrichment to counter this effect. The strange thing is that it doesn't seem related to hot intake air - I've monitored MAT and found that it actually will increase after startup. Say the hot start MAT is 140 F - by the time I'm down to normal fuel map levels the MAT is up to 153 F..... !!!

Here's tables in a stock $8D ROM that show how (at 92C) we have 21.88% enrichment for a minimum of 102 seconds before closed loop is allowed:




Is this related to fuel rail/injector temps? I'm just wondering what the underlying science is that's creating the need for this much extra fuel to get the AFR's under control. I don't see this effect on other types of engines with different style manifolds, etc.

In my LINK ECU I have a 4D fuel table setup based on MAT vs. Engine Run Time and this works well for controlling it. I just want to understand what the underlying cause is.

I may go put some thermocouples on the fuel rail and possibly around the injectors and see if there's significant changes in temp during this first few minutes of start after heat soak.

GD

 

You could try thermally insulating the fuel rails and see if that impacts anything...

https://www.thirdgen.org/forums/alte...ting-fuel.html

I never really quantified the effects to the degree you're attempting to do, but it did result in noticeably more consistent operation (seat of the pants feel) in terms of when the engine is "cool" vs "hot".

 

Does the miniram require similar hot-soak enrichment?

I'm skeptical on insulating the rail. You would think that if that helped significantly, GM would have thrown in some fiber or phenolic washers rather than waste fuel (and, more importantly emissions) with a 22% minute-and-a-half startup enrichment.

Not discounting your results in any way of course. I'm more thinking that it just wouldn't help the stock rails as much as it does on the minram with them being shrouded by the runners on the stock TPI. I think they just pickup a ton of radiant heat.

GD

 

Might be because it is batch fire with pooling of fuel in the intake. I mean it can't actually be burning 20% more fuel, right?

 

I doubt it's actually burning 20% more fuel - I would say it's probably the fuel system is delivering 20% less......

Pooling in a hot intake seems unlikely. Especially in a heat soaked intake that's at 20 in/Hg of vacuum. Any pooled fuel would boil I should think.

I am going to look at at couple things. Fuel pressure, and rail temperature after the heat soak...... There's definitely something to be learned here and applied to a greater or lesser extent to other applications.

GD

 

Like i said I never really quantified it in terms of changes to fuel enrichment. I do know the hot starts improved along with the running consistency... again it was more seat of the pants feel. It was mainly a ln effort to get the car to run more the same when really hot as it does when it first gets into closed loop operation (which is where it felt the strongest, and having exhausted the avenues dealing with MAT related stuff)... the hot start improvement was kind of a side benefit.

It could be the case that the lack of a front crossover tube made the Miniram more of a candidate for this kind of improvement (despite the return paths built into the fuel rails).

Its something that's really cheap to try anyway...

I'm kinda skeptical about the radiant heat thing to be honest... the conducted heat from the intake manifold would likely be the dominant heat transfer mechanism. Especially when the fuel is sitting stagnant in there after the car has been turned off.

If radiant heat was a large factor I'd suspect the rails would still be getting somewhat hot since the amount of surface area of rail and manifold facing eachother is still very large and the distance between them is so small. But I'm not seeing that actually happen, so I tend to discount it. Even convective heat transfer doesnt seem to be significant despite being surrounded by very hot air under the hood.

Like I said in my thread too... consider how the fuel is absorbing heat during operation and unused fuel carrying that heat to the fuel tank via the return line. Over time the fuel in the tank warms up and picks up even more heat when it gets back into the rails. It compounds like interest... the warmer fuel probably generates higher vapor pressure in the tank which can also result in running issues with the CCP system. The insulated rails seem to mitigate a lot of that.

 

The temperature delts between the rails vs the intake manifold was striking after the insulating bolts and washers (to the touch anyway as I never quantified it). You come back from a drive and can literally just put your finger on the rail indefinitely whereas you'll burn your finger in a second or two if you touch the manifold. It's actually not surprising that I saw a difference in performance based on having removed that much heat from the fuel.

 

Yeah it's certainly worth a shot. Easy enough and cheap enough to do. I'll order up some appropriate hardware from McMaster. Going to have to use some pretty thin washers....

I'm curious - if you are running some variant of stock code with your miniram - do you have those stock enrichment tables still adding 22% in open loop?

GD

 

Someone can correct me I'm wrong, but I don't think you can just look at open loop vs coolant by itself and compare it to the enrichment vs coolant, since the open loop AFR calculation also needs the open loop vs MAP table to arrive at the final open loop enrichment value for any given MAP and coolant value.

if the open loop vs MAP value is A and the open loop vs coolant value is B, then the formula for calculating the commanded AFR for any given MAP and coolant value is

14.7/(1 + A/100 + B/100)

I run open loop everywhere below 1800 rpm, so I tuned around that. My commanded AFR values are significantly leaner than factory AUJP since the car seems to be happier with those values. But for idle I richened up the OL idle %change by 4% (looks like you did 3%). Overall at idle, the Miniram seems to like about 13.75 to 14.0:1 as measured on my WB, regardless of whether it's just after start up or at normal operating conditions.

In terms of the startup enrichment vs coolant temp, I'm actually running the factory AUJP tables. It seems to run just fine like that. In fact all of my start up and crank fuel tables are factory.

 

The pics I included in my first post are 100% stock values from a virgin unmodified $8D BIN file. I am not running a factory ECU at all so I've had to mimic this effect without having these specific tables. I created a 4D fuel table that has MAT on one axis, and engine run time on the other - I have the table taper off to 0 enrichment at about 270 seconds.

I would say those tables are a big reason you have had to lean out your fuel map in open loop. The stock BIN files run crazy rich in open loop. They don't taper off they just stay pig rich waiting for it to go into closed loop.

GD

 

I cant comment on how and why there seems to be agood bit of enrichment in open loop areas and startup enrichment in stock tunes.

There may seem to be alot at 92 deg c but compared to rest of the bin at lower temps, its not near as much. Stock these things seem to need fuel but i havent played too much with them stock as it seemed to work fine as is. If it is way rich open loop i correct it however, as i generally tune open loop first

i know on my 383 stealth ram intaked car, $6E AUJL based, i did remove fuel from the open loop afr vs coolant and lv8 tables (not a lot but few % points), and removed crank pulsewidth adder for startup to get a more crisp quick start. Maf table 1 changed for idle. Other than that it ran very well in open loop.

 

After a longer drive I notice that in open loop the AFR's start to lean out and after checking the fuel tank temp it's pretty clear that these engines do a real good job as fuel heaters. I am now firmly convinced that this is all due to fuel temperature. As the entire tank heats up you get a (much less dramatic) leaning of the mixture due to the fuel being less dense. When the engine sits and heat soaks the rails, it takes some time after startup to get the rails back down to the temp of the rest of the fuel - which leads to the initial leaning of the mixture.

With my LINK ECU I can map in a fuel temp sensor and correct for this effect pretty easily just based on fuel rail temp. I'm definitely going to incorporate this into the fuel rail of my FIRST intake but I'll prototype it on the stock TPI and report my results.

Obviously there is no correction for this in the factory ECU's so instead they rely on a huge amount of enrichment on startup and then getting the fuel system into closed loop as quickly as possible.

GD

 

Interesting. Seems to somewhat correlate with my qualitative findings after having thermally insulating the fuel rails.

It'd also be interesting if you also insulated your rails and then re-ran your test with the fuel temp measurement.

 

At the moment I'm waiting on my fuel temp sensor..... being I have 86 fuel rails I have an unused 9th injector port on the back of the DS rail. Plan is to tap that for the 1/8" sensor. Pull sensor ground from TPS/MAP, and use probably the EGR control wire as an input to my LINK ECU and see how that stabilizes the AFR and counteracts the heat soak.

As of now I don't have any plans to insulate the rails if a fuel temp input to the fuel calculation takes care of it. My ECU, in the absence of fuel temp, assumes 68 F for the calculation. And funny enough that's exactly the point where I don't need any additional startup enrichment.

I'll be going to a FIRST intake with my Vortec swap so I'm not going to spend the time to insulate the factory rails.

GD

 

Well - fuel temp sensor installed. The rails definitely heat soak and actually got hotter than the plenum by about 20 degrees, but the addition of the fuel temp input to the the ECU made basically NO DIFFERENCE to the hot-soak enleanment. The fuel rails do cool off once the engine starts while the MAT did not. The fuel rails dropped about 60 degrees while the MAT gained about 10.

I'm open to any suggestions on what else may be causing the heat soak enleanment. Clearly I don't yet have a handle on the exact cause. I can say that it's not the plenum temp (that actually rises), and telling the ECU the fuel temp doesn't have enough of an effect to significantly reduce it.

When I say enleanment - here's the temps I'm seeing after about a 15-20 minute hot soak in the garage (ambient temps are about 50 F), and the resulting startup AFR:

Coolant: 190 F
MAT: 145 F
Fuel rail: 168 F
Startup AFR: 19:1

I need about 30% enrichment, tapering to 0% at about 3 minutes from these temps.

Anyone have a theory?

GD

 

You are saying it needs 30% more fuel after start to maintain normal afr after a long heat soak stop? Then it runs fine in open loop after that 3 min?

 

Exactly correct. It mirrors what the factory ECU does with open loop enrichment, or hot restart, etc - depending on the ECU.

GD

 

I didnt think the restart enrichment lasted that long on stock ecu. On aujl 6e i see the decay delay amount as 50 injections. I believe thats 1 fire per crank revolution in double fire mode. So 50 injections is 50 revs. At 600-800 revs a min idle, that isnt alot of time.

this is perplexing to me cuz I haven’t experienced leanout like that once i enabled open loop entirely. It only happened on closed loop so i assumed it was a o2 timer thing

every efi system seems to have a initial startup enrichment value. Theres a cranking shot to get it to start and then an enrichment that usually follows a quick decay when warm and longer when cold. Not sure the physics behind it other than perhaps the cylinders have cooled down from being off, and fuel needs to increase to have a better chance to atomize and burn for few seconds before running normal commanded fuel and the valves/piston get up to normal temp?

 

Well - take a look at AUM ($32). There is a table - "Open Loop Fuel/Air Ratio Vs. Coolant Temp". This doesn't decay away as far as I can tell - it's in play whenever it's in open loop. There is also a couple more tables - "Open Loop Fuel/Air Ratio Vs. Load", and "Startup Fuel/Air Ratio %Change Vs. Coolant Temp"

At 92 C, and Load Value of 0 (idle, hot restart), you get:

Open Loop Fuel/Air Ratio Vs. Coolant Temp = 25%

Open Loop Fuel/Air Ratio Vs. Load = 3.91%

Startup Fuel/Air Ratio %Change Vs. Coolant Temp = 7.81%

For a total startup enrichment of 36.72% on startup.

I believe that some of this probably decay's away (that startup one for sure), but DAMN that's a LOT of fuel. My engine/intake is 1986 components so this would be the factory tune for mine.

How much cooling of the cylinders could occur at 192 F coolant temp? And wouldn't they come up to normal temp a lot faster than 2-3 minutes? I'm just struggling to understand the physics involved here. I'm hoping someone has a theory I could test.

GD

 

yeah these are how you run and tune open loop imo. I adjust these to get the air fuel correct before ever engaging closed loop. Its like the maf cant do it on its own and these tables have to exist to compensate. I have not fully studied any maf system hacks so im not sure but just going by experience of using them. Maybe you can zero some of those out and try compensating in maf tables but i never did that.

So it seems the initial startup % is only 7% or so which goes away eventually.

 

Well yes you can use them for tuning but that's NOT what the factory intended for them. The factory intended the engine to be in closed loop as quickly as possible and STAY in closed loop except in PE mode. But for whatever reason, they had to add HUGE significant enrichment to the open loop startup (even when hot) to ensure the engine doesn't stall out from going lean before the O2 can begin closed-loop control. Remember these factory ECU's are not supposed to stay in open loop. If you did want to stay in open loop you either need to zero these and use the MAF/VE tables, or use them as a coarse adjustment for the entire map....

If you look at it from the perspective of what the factory ECU is doing..... you have two effects here. One is to counter the enleanment of a hot restart. The other is to drive the O2 sensor rich (above 700ish millivolts) so the ECU knows it can begin closed loop operation.

You can look at it from the other end also. I'll rephrase the question from the other perspective - Why are the MAF/VE tables 30% leaner when in closed loop and how is the ECU able to pull out 30% fuel and still hit BLM/INT target of 128?

GD

 

The way i figure it is your base fueling needs to be configured to near target air fuel before closed loop happens, else you will have to correct alot of fueling, which stock cars i’m not sure they are doing that if blm and int are close to 128 as it transitions to closed loop. I wish i had some stock cars to check this with wideband. I will need to read the hacks to see if we can identify exactly what the fuel calcs are doing. But to me it seems like afr would or should be close to 14’s while idling in open loop at operating temp waiting for o2’s to be ready and when they are, it goes closed and fueling does not change because afr target is correct. Blm int would be close to 128. If everything is working as it should that is. You should be able to see this with a wideband. If open loop enrichment was 30%, and dropped out immediately upon going closed, the engine would stumble and you’d see corrections. Maf tables i have always used to tune closed loop blms. The open loop tables might be used all the time as a type of base modifier even in closed loop. I would have to read the arap maf hack for certain. If i had a stock car i would just try adding a boatload of fuel in open loop tables and see what happens when it transitions to closed loop. I dont see how i could make it idle at 13:1 afr then go into closed loop and see 14.6:1 afr and still have blm near 128. Its been awhile since i had a maf car in my hands. I forget alot but RBob would know if hes still following

If you enable open loop, my cars have not leaned out like you said, so my open loop fueling has been ok. My 383 car was the only weird one where closed loop had the issue, again making me think it was some kind of o2 issue. But any car where i disabled closed loop and installed a wideband, the air fuels have been pretty close to target and car ran fine.

 

I understand what you're saying. It doesn't make any sense to me either. But the fact remains that the open loop fueling in $32 AUM is 28-36% richer than whatever is found in the MAF table on a hot start.

It is artificially rich for sure - to drive the O2 above the millivolt threshold for entering closed loop. You won't see a lean stumble because the ECU waits to enter closed loop for both a set period of seconds (I believe 50 seconds or so) as well as waiting till the O2 rises above 700 millivolts - indicating that the open loop enrichment has successfully pulled the AFR rich - then the closed loop can step in and maintain the situation. In the time that all this takes the enleanment has vanished and closed loop can step in an handle things as normal. From what I have seen, the enleanment vanishes after 2-3 minutes and I can taper out the enrichment in my 4D fueling table that's based on MAT and runtime.

It may also be that the problem is reduced if you have a modified engine with a crazy cam that needs open loop fueling. Probably a higher and significantly richer idle anyway..... Since I don't yet know the underlying cause, I can't say what might affect it.

GD

 

To be more clear about this issue, the BIG enleanment really only occurs for about the first 60-90 seconds. Once I get to 90 second in my 4D table the enrichment is down to about 8% for a hot soak restart.

GD

 

Startup enrichment and decay is one or the more complicated areas of the code. Combine that with spark startup decay and it's a mess. But have to agree with RocZ. Using $8d here, but doubt that other masks are significantly different. For any coolant temp >= 111*F, $8d requires 50 injections from the Table at 0x441 before startup enrichment decay can begin. Because injections are RPM based, decay is slower at lower RPMs.

Assuming 1000 RPM after startup, there will be 1 injector firing every 60ms, so the 50 injections will occur in 3000ms (3 seconds). After that, enrichment decay begins, but it does not happen all at once. Decay only occurs every 4 injections (Scalar 0x41E), until the enrichment has been decayed to =0.

The amount of decay comes from the Table at 0x44F and is all manner of different values at different temps (really don't understand why). But the decay values are large relative to the enrichment added from Tables 0x431 and 0x46F. All in all, at 1000 RPM and 200*F coolant at MAT, the decay occurs very quickly after startup, say in 3-4 seconds total -- 3000ms waiting for 50 injects, and 240ms required per reduction (decay) of the enrichment.

Also, when looking at this, can't really rely on the converted values in the tables because each has different conversions to their underlying values because the inputs to the table lookups may be scaled differently. Have to actually look at the underlying values for things to make sense because that's what the code uses. For example, at 198*F, the startup enrichment added is:
From 0x431: 0x0d (13d = Table = 5.08% AFR Change. Coolant is used for lookup)
From 0x46F: 0x4c (76d = Table = 30.0% AFR Change. Scaled MAT is used for lookup = different scaling than for 0x431 coolant lookup)
Total:--------0x59 = 89d Enrichment added

From the Table at 0x44F, each decay reduces the enrichment by (0xa4 = 164d). This is much larger than the added enrichment (0x59 = 89d), so the enrichment will be decayed in one 240ms pass (0xa4 - 0x59 < 0), causing the enrichment to =0 in 3.24 seconds (3000ms wait time before decay + one 240ms decay pass .

Sorry for the wonky hex stuff, but that's how it works.
Elky

Stock $8d items referenced above:
0x441


0x41E


0x44F


Left = 0x431 Right = 0x46F

 

Sure - but that's just the startup enrichment. Which is relatively small compared to the open loop enrichment that does not decay away.

You still have ~27% of open loop enrichment.

My hot soak 4D table starts out at 30% and by 90 seconds is down to around 8-10%. I'm still struggling to understand the physics that cause this. Don't typically see this on other types of engines.

GD

 

Admittedly then despite having read all the posts, I don't understand what exactly you're referring to as "open loop enrichment". Once past startup, all the code does in OL is:
- Compute a commanded AFR
- Set INT =128 to prevent BPW corrections
- Get VE
- Calculate Speed Density BPW using VE, injector flow rate, ideal gas law and AFR
- Compensate BPW for low PW and voltage and send to injectors
So any richness has to be from an external source unless the injector flow rate is way off causing the BPW calculation to be rich. Is this "OL enrichment" showing up in reported WB AFR?

 

My assumption is that the open loop enrichment is added to the commanded AFR. Take a look at the tables I posted above.

"Open Loop Fuel/Air Ratio Vs. Coolant Temp"

and

"Open Loop Fuel/Air Ratio Vs. Load"

As for it showing up in the wideband - depends. If it's not heat soaked then yes it shows somewhat rich. If it is heat soaked then the enrichment counters it.

GD

 

Because XDF titles are so different, any chance you could post the hex addresses of those tables. I'm curious to understand this. Tks.

 

Sure - the "Open Loop Fuel/Air Ratio Vs. Coolant Temp" is at:

0x41F

And the "Open Loop Fuel/Air Ratio Vs. Load" is at:

0x3D3

This is $32 that I pulled the addresses from just for reference and because my Trans Am is an 86 that originally ran this code.

GD

 

I thought you were initially referring to $8d with which I'm familiar -- $32, not so much. Regardless, assembly language makes my head hurt, so will opt to hurt with $8d. Based on other similarities, $8d is likely not materially different than other TPI masks.

Everything below considers post-startup and after decay of startup enrichment. Bottom line, the 2 tables you referenced are indeed used after startup to compute OL AFR, but their impact is inconsequential unless extreme temps or kPa. Note that the impact on computed AFR is shown and not on BPW. But that's not an issue because the relationship is approximately inversely linear. That is, if AFR increases 10%, BPW decreases approximately 10%.
Under the covers, here's how OL AFR is computed:
Sum:The sum of these 3 items is stored as Interim OL AFR #1.
A value from the table OL AFR % Change .vs. MAP ($32=3d3, $8d=420) is added to the Interim OL AFR #1 to produce Interim OL AFR #2.

Interim OL AFR #2 is multiplied by Stoich AFR to yield Final OL AFR.

$8d code simulation shows the following (see attached spreadsheet for details):
OL AFR does not materially change from Stoich unless MAP is < 45 or > 70, or unless coolant temp is < 20*F

So what GM has done with these tables makes sense: Lower temps = richer; higher kPa = richer. Otherwise no change to AFR and thus BPW.

Not sure where you additional fuel is coming from, but it's not from the 2 tables. Again, can't use the underlying decimal values or the converted values from the tables. Must use the underlying hex to make the calculations.

 

Interesting - and yet one more reason to hate working with the stock ECM's like the 7165 - with the XDF's so poorly documented and poorly translated you almost have no idea what it is your are changing or what it will affect without crawling through the assembly.

So looking at S_AUJP - because that seems to be at least mostly correctly documented and the table values properly translated in the XDF...... you have 9.77% OL enrichment in the Loop Open Param, AFR % Change .vs. Coolant under most conditions (>46 F).

Not disputing your spreadsheet but wouldn't it need to be somewhat richer than 14.7 to drive the O2 rich so it can enter closed loop? I assumed that's what the 9.77% open loop enrichment was supposed to do. Also if your spreadsheet is correct then that S_AUJP XDF isn't correct because it shows 9.77 % enrichment in open loop. But again the closed loop system is watching for the O2 to spike so it kind of makes more sense to me that the XDF is accurately reflecting that.

Then there's the Startup Enrichment Offset .vs. MAT that seems exclusively to deal with heat-soaking and is adding 30% (to something - the Startup Enrichment .vs. Coolant table? - it's not clear) above 154 F MAT.

And as for the Startup Enrichment that decay's after 50 "injects"...... how long does 50 injects take at idle I wonder?

Like I said I'm throwing about 30% additional fuel at the engine when I do a hot soaked restart, but much of that decay's away quickly. Down to about 12% by 60 seconds. I'm wondering how long "50 Injects" is exactly under those conditions. Not something I've tracked.

Between the larger number of injection events before startup enrichment decay at higher coolant temp, and the MAT offset at higher MAT temps I do see a theme here that GM was aware of the enleanment happening on hot soak restarts. To a greater or lesser extent anyway. Maybe my particular setup is worse than most (or worse than stock, though I don't clearly see how that would be the case), or in stock form they allow it to go significantly leaner without correction, but the complete picture is still not entirely clear for me.

GD

 

Update - testing out another theory, I have bypassed the TPI fuel rail mounted regulator in favor of an external regulator (will be needing it for my FIRST intake and rails anyway so one less thing when I swap) and I will report back if it makes any significant difference in hot restart. I had an adjustable regulator hat and replacement regulator (both from SouthBay) on the TPI rail and I'm not entirely convinced it could hold accurate pressure under all temps, etc. The Deatschwerks regulator I replaced it with is excellent and being mounted on the strut tower with it's own gauge it is much easier to work with.

GD

 

Move here from https://www.thirdgen.org/forums/diy-...ml#post6349257 where it was posted in error.

They are not my values. They are from the Factory GM AUJP Calibration.

The spreadsheet is correct. The Factory table values were merely used in the code simulation to execute what GM designed. Regarding CL, the o2 sensor is declared active once voltage is detected outside 300 or 700mv. Even before CL is invoked, o2 voltage is used to determine FAST o2 direction. CL is invoked after the sensor has been declared active, and engine run-time, coolant temperature and BLM Learn thresholds have been exceeded.

It's certainly possible that the XDF conversion for 0x4FE is incorrect. It's a product of the first AUJP XDF development. But, again, that's irrelevant. The underlying hex value is from GM, and that's what the code uses and therefore used in the simulation.

We're now dealing with startup. The above discussion had to do with post-startup after the startup enrichment had been decayed.
The startup enrichment .vs. MAT is added to startup enrichment .vs. coolant. This gets decayed out. See Post #25 above.

Post #25 above.