When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
I had previously tuned my car in 70f weather. It's a '99 1.8L with FlowForce 640s with an otherwise stock fuel system. GM IAT is just after the intercooler on the first upward elbow. I was originally using a MAT correction curve similar to Brain's table as shown below, but I have since made the higher cells all 100% correction. Whenever I am driving my car to work in the morning it is very rich (12.5 cruise, 12.0 idle) even though it's maybe 65f outside. Later in the day when it's 90f out with intake temps close to 95-100f that makes the fuel correction 100%. Despite this, the car is uncomfortably lean (15.4 cruise, 15.0 idle) even after the "warm" fuel should've been flushed out, and even worse on hot starts. After researching, I discovered that NBs had a returnless fuel system with higher pressures for some emission reason which makes fuel heat soak pretty bad in my experience. I've previously tried adding ASE at the higher temps but that just seems like a poor bandaid fix and it doesn't address the lean cruise condition. I've considered doing a returnless fuel system, but that would only lessen the effects, and it would probably still be there to some degree, and here in Tennessee weather can swing a good bit between seasons. Now that I am getting things more dialed in fueling-wise, I feel like it is almost impossible to get things satisfactory without addressing the injector heat soak/temperature issue.
Additionally, larger injectors are more prone to heat soak. Another user explained this very well.
Originally Posted by stefanst
The current mainstream theory seems to be that what you're seeing at idle is injector heat-soak. To break it down:
- At higher loads (longer injector opening times) and higher temperatures, without any IAT correction, you experience a mixture richer in fuel than under the same conditions at lower temperatures. This agrees with the laws of physics that warmer air is less dense and contains therefore fewer molecules of oxygen in the same volume. Since MAP-based tuning estimates required fuel from volumetric efficiency, we need to correct for the less dense charge.
- At lower loads (shorter injector opening times) and higher temperatures, even without any IAT correction, you experience a mixture leaner in fuel than under the same conditions at lower temperatures. This is explained by the change of injector behavior with temperature. Warmer injector coils have higher resistance and therefore have less current going through them. This means a weaker magnetic field and the injector opens more slowly. So now our dead-time is longer than at lower temperatures and our dead-time compensation is wrong.
Why doesn't injector temperature seem affect the high-load scenario? It does, but it is drowned out by the effect of the lower charge density.
Example: If your injector is supposed to open for 1ms and your expected dead-time is also 1ms, the ECU commands power to the injector for 2ms. If the dead-time is off by only 0.2ms because of injector temperature, we now have an effective opening time of only 0.8ms. This means approx. 20% less fuel and the car running lean. Even if air-density is only 95% of what it is at lower temperatures, your still missing around 15% fuel.
Now picture the same exact scenario for higher loads. Lets say under full throttle your injector is supposed to be open for 10ms. We deduct the same 0.2ms as before for incorrect dead-time. This means we effectively only open for 9.8ms. We're running 2% leaner than planned. Now add in the fact that we only have 95% charge density at higher temp, so we need actually 5% less fuel than at lower temperature and all the sudden we're running 3% richer.
So running rich at high load and lean close to idle can be explained by air-density change and temperature-dependent injector behavior occurring at the same time.
Then another user in the forum suggested maybe adding a "thermocouple to the fuel rail, measure that temperature, and add a compensation table"
Originally Posted by codrus
I've had a bunch of issues with injector heat soak, it's a problem inherent to big injectors on a small motor because the longer the dead time relative to the idle pulse width, the bigger the issue.
So the megasquirt doesn't have a table for injector heat soak dead time compensation. That's a solvable problem (it's just software and the code is available, albeit not freely redistributable), but you need a good measure of the injector temperature and we don't have that.
As I see it, there are three main temperature ranges that are relevant: Cold start, normal running temperature, and heat soak after being parked for 10 minutes. The best I've come up with is to program the megasquirt's dead time numbers for the normal running temp and live with it being off in the other two. A richer idle at cold start is fine, it's not enough of an issue (at least, not with my id1000s) to really hurt anything. Lean idle after being parked and heat soaked is more of an issue, but I've just lived with it by giving it a bit more throttle for the few minutes it takes to cool down. As you flow fuel through the injectors it cools them.
My NB is still a non-return system. A return fuel system should cool the injectors down faster, because there's more fuel flowing through the rail.
One could use a thermal paste to attach a thermocouple to the fuel rail, measure that temperature, and add a compensation table. That's probably better than nothing, but it's really the temperature of the injectors we care about, not that of the rail.
Theoretically, it's possible to measure the current through the injectors and determine the resistance to build a better compensation table. A friend and I experimented with this a bit, but didn't get anywhere yet (our initial measurements were too susceptible to noise, so it needs a current measurement on the board rather than hacked up with loose wires).
--Ian
So... would this be feasible? I was considering getting a closed element sensor from DIYAutotune and using some sort of epoxy/JBWeld to attach the element tip to the back of the fuel rail? I suggest the back of the fuel rail so that the fuel flow is essentially that of one single injector. The injectors have a rubber cushion seal that should isolate the injectors somewhat from the intake manifold's direct heat. Most of the heat is probably radiant so if the sensor was just close enough it may give a somewhat useable number. The biggest issue would be making the fuel temp signal actually useable with the ECU. It would require altering the source code and adding something like (FuelTempCorr/100) into the fuel calculation table similar to how barometric correction exists.
I've considered messing with the MAT/CLT correction table to add more fuel at high clt temps and low loads, however, I'm thinking that with a thermostat keeping my clt temp near 192f this wouldn't affect the difference between cold and warm weather. It might be more useful without a thermostat but since I try to daily the car year-round, that's not possible. If the addition of the fuel temp sensor is too much, maybe I'll attempt a mixture of the MAT/CLT table, a return-style fuel system, and a phenolic intake manifold gasket. Thoughts?
So... would this be feasible? I was considering getting a closed element sensor from DIYAutotune and using some sort of epoxy/JBWeld to attach the element tip to the back of the fuel rail? I suggest the back of the fuel rail so that the fuel flow is essentially that of one single injector. The injectors have a rubber cushion seal that should isolate the injectors somewhat from the intake manifold's direct heat. Most of the heat is probably radiant so if the sensor was just close enough it may give a somewhat useable number. The biggest issue would be making the fuel temp signal actually useable with the ECU. It would require altering the source code and adding something like (FuelTempCorr/100) into the fuel calculation table similar to how barometric correction exists.
I never got around to making any more progress on that project, alas.
The actual code changes aren't that hard (at least not if you've got some experience with embedded system development in C), but from a practical standpoint the challenge is that the license terms don't allow redistribution of the changed code. To comply with it anyone who wanted to use this feature would need to be able to make those changes himself, one couldn't just get a pre-built modified image from someone else. You also need to be able to set those values, which takes some changes in TunerStudio. It's been a while since I looked at how TS represented data to load and I don't remember if the data description stuff was sufficiently general to allow you to do things like add a completely new table without needing to change the TS source.
I don't know if the injector rail temp itself is close enough. It might be but I can also see a bunch of reasons why it might not be, and I suspect the only way to answer that question is to try it.
Return conversion will not help. BTDT.
Messing with MAT corrections will not help. Set MAT correction to make AFR's match your tuning temp (MAT at which you tuned your VE table) at approximately 100kPa, not at idle.
It is not fuel temp, it is injector winding temp... Although there may be some level of correlation. So, having a temp sensor could help... but I do not know how you would wish to use the information.
See my attached tune and look at ASE / ASE Taper (in 0.1 seconds, not cycles) / EGO authority in idle area / Possibly MAT correction as a reference for your rich cruising.
These things will 99.9% fix your issue.
While you are there, you can also see WUE and delayed temp for start of EGO. This allows extra fuel during warm-up, instead of EGO forcing AFR to what is needed when fully warmed up.
Validate your VE table when MAT is at your tuning temp (I think you said 70F)
IThe biggest issue would be making the fuel temp signal actually useable with the ECU. It would require altering the source code and adding something like (FuelTempCorr/100) into the fuel calculation table similar to how barometric correction exists.
Some corrections for the returnless fuel system and for the fuel rail temperature are already implemented starting from 1.5.x firmware
07-17-2021, 11:39 AM
Temp Sensor on Fuel Rail to combat heatsoak?
0
