As far as controllers without gauges, something that offers Analog output and CANBus output would be ideal. Using a LSU4.9 sensor usually avoids free-air calibration as well.
AEM appears to offer this X-Series:
https://www.vividracing.com/aem-xseries-inline-wideband-uego-afr-controller-lsu49-p-152414916.html
EcoTron InLine ($80 w/o Sensor):
https://www.ecotrons.com/accurate_lambda_meter/alm-inline/
Are there any other offerings out there that are similar?
Most of the other models we've been discussing are similar - 14point7 Spartan Lambda 2, Innovate LC2, Tech Edge 2J9.
Looks like EcoTron is the most cost effective for a controller that offers CAN output (although it is CAN or Analog output, one only). 14p7 Megatron is needed for CAN ($250), TechEdge does not appear to offer CAN output, Innovate only offers serial output.
WRT using a standalone with an MS or something like that- the other important key is how long it takes for the sensor to come alive. Right now, the nominal light off time for a production WB sensor at 70F is less than 10 seconds. Anyone have an idea how the various systems light off?
Haven't read all three pages, but you WANT to run a wideband all the time. Why would you get a standalone ECU and then give up one of the largest benefits? Some include:
-Significantly higher resolution (NB sensors generally output a 0-1V signal like an on-off switch vs. 0-5V Bosch LSU4.9. Meaning you know when you are rich and when you are lean. You do not know HOW rich or HOW lean you are which is obviously critical when tuning an engine)
-Enables you to run real closed loop control. This is another reason you got Megasquirt in the first place. (Tell me if you need a brief explanation of closed vs. open loop)
-Enables VE table autotuning. This is a great feature found in most new EFI's. You will not get good, if any results using a narrowband.
TLDR: Ditch the narrowband forever, and welcome to the modern age.
engiekev said:As far as controllers without gauges, something that offers Analog output and CANBus output would be ideal. Using a LSU4.9 sensor usually avoids free-air calibration as well.
AEM appears to offer this X-Series:
https://www.vividracing.com/aem-xseries-inline-wideband-uego-afr-controller-lsu49-p-152414916.html
EcoTron InLine ($80 w/o Sensor):
https://www.ecotrons.com/accurate_lambda_meter/alm-inline/
Are there any other offerings out there that are similar?
There might be, but honestly the AEM X Series is the best in the aftermarket. The response time blows away other controllers and again the CAN output is obviously extremely useful. Also AEM is very active in the motorsports community so supporting them in-turn supports us :)
I did a little investigation, here's what I was remembering:
O2 Sensor Controller (Bosch part # F 00H M00 187) linked to a Bosch 4.9 sensor. The part number comes up as a "Control unit, Lambda control Lambdatronic" on the bosch-automotive-catalog.com site, but I can't find any sort of spec sheet.
In reply to goingnowherefast :
Not everyone wants to run closed loop. 99% of the time yes, but not always.
In reply to Knurled. :
When would you not?
Knurled. said:In reply to goingnowherefast :
Not everyone wants to run closed loop. 99% of the time yes, but not always.
Closed loop is demonstrably better than open loop. Modern vehicles run closed loop for basically everything include WOT states.
alfadriver said:In reply to Knurled. :
When would you not?
Any engine that spends any appreciable amount of time misfiring due to radical cam timing (or port timing...) which would be one of the prime reasons to want an aftermarket computer in the first place.
If you want to make the engine calibration center of your brain shudder, look up the "ghost cam" tunes people do for Fords.
Knurled. said:alfadriver said:In reply to Knurled. :
When would you not?
Any engine that spends any appreciable amount of time misfiring due to radical cam timing (or port timing...) which would be one of the prime reasons to want an aftermarket computer in the first place.
If you want to make the engine calibration center of your brain shudder, look up the "ghost cam" tunes people do for Fords.
You do realize "ghost cam" tunes are running closed loop fuel at idle right? lol
The only engines that can't run closed loop nearly all the time are the ones that eat spark plugs as a rule. They also usually need a "rod in engine" sensor.
goingnowherefast said:Knurled. said:alfadriver said:In reply to Knurled. :
When would you not?
Any engine that spends any appreciable amount of time misfiring due to radical cam timing (or port timing...) which would be one of the prime reasons to want an aftermarket computer in the first place.
If you want to make the engine calibration center of your brain shudder, look up the "ghost cam" tunes people do for Fords.
You do realize "ghost cam" tunes are running closed loop fuel at idle right? lol
They also are not radically cammed engines.
When I have ever tuned an engine like that, it is impossible to tune for an arbitrary number, you just give it what it wants. The engine in my car now will actually idle smoothly, at 11:1 according to the MTX-L. Leaner than that and it starts stumbling. Likewise, under cruise it doesn't like to run leaner than 13:1ish dependent on RPM, although under higher loads it is perfectly happy to run as lean as 16:1. And it's all temperature dependent, too. How do you control that with closed loop, in a not-OE-complex engine controller? Or rather, why would you want to? You're not trying to make a cat happy, just trying to get good drivability.
My more radical engines would bounce between 15:1 and pegged lean under idle and cruise because of the constant stuttering.
Paul_VR6 said:The only engines that can't run closed loop nearly all the time are the ones that eat spark plugs as a rule.
That's a pretty apt way of condensing it.
Knurled. said:Paul_VR6 said:The only engines that can't run closed loop nearly all the time are the ones that eat spark plugs as a rule.
That's a pretty apt way of condensing it.
Occasionally I have these brief moments of clarity!
In your above example of your afr targets being temperature dependent you *could* always set up table mapping with CLT as the "w" axis mapping between AFR table blending. It's explained in the tunerstudio manual (there may still need to be some trickery using CLT as a generic input for fuel temp correction..haven't tried this in awhile). Then run closed loop anytime you don't think it's misfiring due to camshaft based shenanigans.
In reply to Paul_VR6 :
Given that the engines I generally play with are insensitive to tuning (with respect to finding enough pieces to glue back together and make an engine again) my main concern with tuning is to get it running as lean as possible without stumbling, because the generally high exhaust residuals mean you need a rich mixture for reliable combustion, and this results in lots of fuel dilution of the oil. The leaner I can get it to run, the less often I have to change the oil. Right now it's about 800-1000mi before the oil has gone from a quart underfull to a quart overfull and the 20W50 is a runny, stinky mess.
So, really, what I'd like is a mixture adjustment knob so I could pull fuel while on the highway until it just starts to stumble, and then add a little back. Could probably do this fairly easily, come to think of it, by telling MS that the trim pot returning a signal from 5vref is actually a 0-5v ethanol content sensor. (I'm on MS2/Extra, I don't think map switching/blending is something I can play with)
I run open loop on my Megasquirt MS3 when above about 80kpa... It works well, but does anyone run closed loop in boost with a megasquirt? I don't think I've ever seen anyone do that on a boosted Miata.
In reply to morello159 :
Why would you not run closed loop? Even if it's slightly wrong, it's still better than guessing. Run the specific a/f you want, accurately.
In reply to morello159 :
This is the way FAST XFI works, and it actually works really, really well. But the wideband controller is part of the hardware, meaning sensor lag is minimized, and they use NTK sensors which are far more reliable and robust.
I'd be leery of doing it with an Innovate and an analog input to a MS.
I think the fear was that the sensor couldn't respond quickly enough with the gas moving so quickly in boost, and if you had a sensor go bad, you might start pulling fuel unnecessarily. But pretty much everyone in MS land is running an analog wideband signal. I don't think TS lets you set asymmetric control authority limits - I would like for it to be able to add fuel if necessary, but above 100kPA (for example) never pull fuel.
I run my AEM over CAN, and I've heard the 4.9 sensors are a little quicker than 4.2... I may do some pulls in boost with it enabled and see how it does.
They also run speed density instead of MAF for some reason, but I suppose that's topic for another time...
morello159 said:I run open loop on my Megasquirt MS3 when above about 80kpa... It works well, but does anyone run closed loop in boost with a megasquirt? I don't think I've ever seen anyone do that on a boosted Miata.
We used to do it on the old Link systems around the turn of the century. Hydras and FM221s also ran closed loop in boost. So it's fair to say that a pretty healthy percentage of boosted Miatas have been running closed loop in boost for a long time. If the Link could do it, the MS3 should be able to do it.
morello159 said:I think the fear was that the sensor couldn't respond quickly enough with the gas moving so quickly in boost, and if you had a sensor go bad, you might start pulling fuel unnecessarily. But pretty much everyone in MS land is running an analog wideband signal. I don't think TS lets you set asymmetric control authority limits - I would like for it to be able to add fuel if necessary, but above 100kPA (for example) never pull fuel.
I run my AEM over CAN, and I've heard the 4.9 sensors are a little quicker than 4.2... I may do some pulls in boost with it enabled and see how it does.
They also run speed density instead of MAF for some reason, but I suppose that's topic for another time...
If your open loop calibration was good enough, the sensor does not have much correction to do, anyway. I also wonder if people are reading too much into the whole sensor response issue- I really have been using WB sensors my entire career, and have not really seen sensors that would break an engine due to their slow reaction. Heck, SO much of the tuning of cars is actually based on the sensor response, it's not even funny. And that's not even including how the controller currently reacts to that response. Honestly, every single car out there in production over the last 25-30 years has a transient fuel calibration totally based on what everyone here would call slow responding sensors. including turbos. I honestly can't remember what version of Bosch sensors we used for our 2010 production turbos- and they didn't cause any problems.
Speed/Density or MAF, I really think you are reading way too much into the sensor response issue.
BTW, your open loop calibration is pretty much based on the sensor, isn't it?
In reply to Keith Tanner :
MS3 can and will do it. So will MS1 for that matter.
BTW - In hindsight, "sensor lag" isn't a very good descriptor - more like "hardware lag". Having the wideband controller as part of the engine controller really speeds things up, instead of having a separate controller that has to generate an arbitrary signal (CAN or analog) and send it over to another computer to be unpacked again into useful data.
The nice thing about doing it with CAN is that at least you don't have to worry about varying ground levels screwing up your readings, like can happen with an analog signal.
In reply to Knurled. :
And I will stick to my point, I think that's being totally over thought. The "lag" of the system can easily be dealt with, or not even worried about.
Here's the thing- what "fault" causes problems? If there's a sudden a/f change that is unexpected, even a slow sensor reacting to it is better than nothing. If it's a known and expected one, the sensor will really do nothing about it anyway. All of the UEGO's we uses for tuning are all analog recordings- and they match up super well with the internal sensors, so even that, I don't think it's a big deal. If the sensor faults, then the system should have bands on how far it can correct.
And if there's a floating ground that causes a signal error, then it's not just a/f measurement that is going to suffer.
I really would have no problem using an analog signal wb sensor for any actual control.
You'll need to log in to post.
