Sending my car in for a 360kw setup next month, need advice.

[joshuaho96]

1 hour ago, Lithium said:

I've been seeing all the email notifications for this thread and kinda interesting to see a bunch of tuning chatter.   All seems pretty legit really, just thought I'd meander in and add 2c where it feels like they're not being so redundant. 

The more data you can get the better really, but it doesn't mean it's needed - really a lot of it is personal preference and there are a lot of different people who have different reasoning for different things which can all seem "reasonable".  I find it odd that a tuner would say they wouldn't touch something without a wideband and oil/fuel pressure though, sure both are sensible but realistically he should be able to tune without those... they are just good additional precautions.   

They definitely make the tuning and diagnostic process easier, I guess if he can be that bossy then good on him     I'd still not use the car's wideband for tuning, I prefer to use an external one (like on it's own power source etc) that I know I can trust and then verify the car's wideband against it to make sure everything is working right.  There are plenty of cars that have had widebands installed and potentially even have closed loop running etc off them where they've never been checked and are likely reading off and has potentially the tune isn't doing what people have intended from them.  A lot of variables at play with everything and as always, the more data the better.

It'd be great if everyone had EMAP, EGT, MAF, turbine speed, pre and post intercooler pressure, knock, coolant, oil and fuel pressure, hell cylinder pressure would be nice too  But theres are not NEEDED, but if we pick the right ones for what the car is doing and what fits the budget etc we can do a small subset that ideally ends up just being "nice to have" to help identify issues before they become serious, or help resolve issues after the fact. 

Hell, may be an unpopular opinion but if someone is trying to keep the amount they're spending down but want the car to have some safety then I tend to lean more towards fuel pressure than lambda for safety.  A lot of tune drift and lean out issues seem to come down to fuel delivery, so while monitoring lambda and adjusting or reporting on error is definitely handy - if the tune is dialled in well realistically fuel pressure is often more than capable, potentially better at catching most issues have where a lean out could happen.  I've rarely seen issues where uneven injector flow is the cause of a lean out (and almost any issues I do know of have been due to people using gas injectors, silly idea) and you could have a single injector lean out enough to hurt it's piston but not actually see a massive lean out at the wideband.  Ie, 11.5:1 on 5 cylinders and 13:1 on #6 = an average of 11.75, a closed loop correction on that will make 5 cylinders a bit rich and the 6th one still lean and simply mask an issue and I'm not sure if most people would do an engine protection on .25 off target AFR.

Another issue with relying primarily on closed loop is that exhaust leaks and misfires can set it off into weird places.   A decent exhaust leak will set closed loop off dumping a heap of fuel into an engine that doesn't need it (though monitoring trims will help pick up on these things so it IS a handy diagnostic) and things can get pretty weird if you get a misfire on a cylinder, closed loop can start trying to throw heaps of fuel at the thing and make the whole engine run like shit and make it hard to immediately work out that it all started on a single cylinder losing spark and adding to the mess by fouling all the plugs. 

With fuel pressure being the "main guy" the overall fuel model tends to be more accurate if set up correctly, and it can also step in if you start getting a pressure drop - both bumping up injector open time to compensate for would otherwise be a brief lean out, or going into engine protection before a lean out ever happens... while lambda based safety tends to rely on the engine already leaning out (and probably way too much) before doing anything about it.   I'm far from saying having O2 feedback is not worth it at all, just throwing other things into the discussion.  The more data the better, but data definitely needs context to be used optimally.

Now that I've mentioned misfires, another thing that surprisingly I've not noticed mention of so far is knock sensors, whether or not you use actual knock control.  That's another thing that potentially will pick up on dangerous operating condition as quickly as a wideband will (not that you want to wait for it to reach that point), with the added advantage of being able to use windowing to identify which cylinder is responsible.   Another neat thing about using windowed knock monitoring is a car I was involved with had fuel pressure, knock logging, lambda etc and it dropped a cylinder in a run.   Lambda went lean on a bank, fuel pressure was fine, and the noise from the knock sensor during one cylinder's "turn" dropped way off compared to the rest.   Replaced the coil on that cylinder, was running on all cylinders again and life was good.  Without that range of sensors and the ability to make sense of what they were doing that whole diagnostic process could have been a much bigger headfk.

< Misfire log

Not trying to be a dick, I actually rate using a MAF if it's an option - they're pretty cool for the reasons you've implied... only reason I'm contradicting you is there is more "interesting stuff" (well I find it interesting) in the details that contradict you there.   

Porsche, Ford and BMW (and others) don't use MAF for load reference on a lot of their current high end cars.  Some of them even still have a MAF sensor but it's more for diagnostics and other general control stuff, as earlier - more data is always good even if you're not relying on it.   Speed density tuning has often been considered less accurate but these days with more understanding, better sensors and better processing hardware it's damn hard to beat and also ties into load based tuning very very well as part of the calibration process you effectively are building a reference table of how the cylinders are filled under given conditions without needing to hit them, which means you can basically forecast what you need to give the engine to make x amount of torque before the fact and then hit that torque level very quickly.

This is particularly relevant to BMW and Porsche (the ones I know off hand at least) that don't actually use a traditional speed density setup where you have VE numbers in a MAP x RPM look up table, but they use a combination of boost pressure, throttle angle (and data on what the throttle's volumetric flow is at different angles), and MAP to calculate engine load/torque, either before or after the fact.   This means that they're able to demand a target boost and throttle position to hit a torque target with less messing around than you could hope to with a conventional speed density setup, let alone with a MAF based system.   The accuracy of the load calculation is very on point, way more responsive and accurate in transient load conditions and again - you can "plan ahead" in terms of controlling throttle and boost to hit precise torque targets very quickly.    If you've ever wondered how current generation Porkers are so ridiculously consistent and controlled with power delivery, that's absolutely part of it.

In terms of the grid of buttons, I don't have a lot of suggestions.  I think most of the time it seems nice to have but if you're struggling to think of things to use them for then you probably don't need to set them up right now.  The ones mentioned make sense enough (as much as pops & bangs etc make me cringe and feel for the hardware). 

I'd consider setting up a boost control mode that is effectively a torque limit per gear setup where you kinda have your cake and eat it too boost wise.   When I've tuned >500whp stock gearbox GTSts I've given them reduced power in first and second, third gear I hold a peak torque level that is maybe adventurous but not suicide - but hold that torque to redline (so you still make decent power) and send them to the moon in 4th upwards.   I personally reckon it's a really nice way boost targetting setup for RWD Skylines for various reasons, they don't really feel super held back but also are easier to drive fast and without breaking things.   So I guess you could have a "valet mode", a "fast but under control" mode, and a "dumb cnut" mode where you send it to the moon in every gear and situation.  

Anyway, sorry for the wall of text - thought I'd throw some of that into the yarns... hope it's not TL;DR or too off topic
  Be interesting to see where you end up with it.

Regarding closed loop fueling going all wonky with a misfire - I think that's an argument for better modeling and misfire detection in a standalone ECU no? Not sure anyone is going to ever bother with it but if we want standalones to potentially be viable replacements for things going into the CAN/GDI age the software complexity has to ramp up pretty dramatically to make it all work. I would consider most standalones to only be good enough for something like an RB26 that is stone age as far as a lot of the controls go.

Sounds like you're describing throttle mass flow when it comes to MAF/speed density replacements. I'd be curious to know how many cars actually use it and how well it actually works, from what I've seen if the pressure ratio across the throttle body is too close to unity modeling the flow rate becomes a challenge. Whatever the E9x M3s use sounds rather brittle as well, supposedly something as simple as an alternator bearing causing excess drag at idle will cause all kinds of running issues. I would be curious to know if someone has actually solved the problem of ITB mass flow measurement without a MAF. I have seen what alpha-N or blended alpha-N/speed density strategies are like and I'm not really a fan.

[Lithium]

3 hours ago, joshuaho96 said:

Regarding closed loop fueling going all wonky with a misfire - I think that's an argument for better modeling and misfire detection in a standalone ECU no? Not sure anyone is going to ever bother with it but if we want standalones to potentially be viable replacements for things going into the CAN/GDI age the software complexity has to ramp up pretty dramatically to make it all work. I would consider most standalones to only be good enough for something like an RB26 that is stone age as far as a lot of the controls go.

Sounds like you're describing throttle mass flow when it comes to MAF/speed density replacements. I'd be curious to know how many cars actually use it and how well it actually works, from what I've seen if the pressure ratio across the throttle body is too close to unity modeling the flow rate becomes a challenge. Whatever the E9x M3s use sounds rather brittle as well, supposedly something as simple as an alternator bearing causing excess drag at idle will cause all kinds of running issues. I would be curious to know if someone has actually solved the problem of ITB mass flow measurement without a MAF. I have seen what alpha-N or blended alpha-N/speed density strategies are like and I'm not really a fan.

Good shouts, because the quoting on this is annoying I'll try and address them bit by bit:

•  Misfire detection: Of course!  Would love that and have discussed this with people who do aftermarket ECU development and the likes.   That doesn't really negate anything I've said, but definitely a thing that would be cool.  
• Standalones improving to suit the CAN/GDI ages:  Tricky one, there is soooo much constant chassis and other bit bespoke development that it's hard to see how viable making something that can do everything would be without making it programmable, like Motec M1 Development.  

FFS, I pressed shift-enter hoping to make spaces between the points and it posted my comment part way through.  Fwiw I'm still working on it...

 

Edited August 16, 2023 by Lithium

[Lithium]

 

• Talking about TMF?:  I am talking about something along those lines, yes - though I assume you are talking about Emtron's TMF specifically?   That is one implementation of a speed density system that happens to use throttle as an input but I wasn't talking about their implementation.  In terms of the the pressure ratio across the throttle body concern, yeah I'd "day dreamed" about alternative ideas for determining airflow and had pondered on pressure across a throttle and ruled it out in my head because of that exact reason.   I've never used Emtron, let alone their TMF so I don't have a great understanding of how it works or how well it works so I can't really comment on it in context of that.  
   
• How many cars use "TMF" style speed density and how well it works: I don't think the E9x series BMWs use it though I'm not 100% sure.  And to be clear, I am far from claiming I am any authority on these things - much of it is what I remember from conversations over the years with people much smarter than myself so I'm kinda forwarding on my patchy recollection of how I interpreted what I was told about things.  In terms of how well it works, I feel like I already said that from what I understand the new Porsche's use this kind of system.  Do you doubt the 991 Series Porsches work well?  I've only been in a single 991.2 Turbo S and it went like a bat out of hell, but it very much felt sorted beyond "only WOT" type driving by my personal gauge.  Super super competent car all round.   Like almost too well behaved.
   
• Not a fan of Alpha-N/MAP strategies: Can you elaborate?   They can be a headfcuk to deal with but I'm not sure they are broken as such.   I definitely prefer the concept of pre-throttle pressure vs throttle angle versus an engine flow characterisation map and feel like that's probably the best answer though am not aware of an aftermarket ECU offering that.  Yet.

Now for the "my view of the world" ramble, again I am no expert so more than happy to be picked apart on this if there is stuff you (or anyone) knows beyond me so this is just my take and limited experience on this general topic.  Please excuse any terrible wording or thinking with this, its been a long day at work and if I were wiser I'd write this when my brain wasn't already done - but it's a fun topic so I'm doing it immediately

What I've gathered about at least the 991 series Porsches they use a combination of a boost sensor and throttle area vs angle calibration, though I could be wrong - am definitely riffing off bits and pieces I've been told.   I *think* it's the 991.1 that has just a pre-throttle pressure sensor and the 991.2 has one on both sides (the post throttle one being more for diagnostic purposes
?!) - I'm guessing from that there are possibly a bit more like how the Motec R35 GT-R package does things, where it uses pre-throttle pressure and throttle opening vs flow data and engine efficiency data to estimate manifold pressure on the other side of the throttle.   So if you think of an Emtron which (if I understand correctly anyway) is solving for mass flow using pre-throttle pressure, post throttle pressure and throttle angle - these guys are effectively solving for MAP by using pre-throttle pressure, throttle angle data and engine flow data.   I could be wrong, but if it's anything like that then in my head it's going to get around the situation where there isn't a large pressure differential across the butterfly, but also creates a reasonable amount of calibration work to make it work right.

In that sense it makes MAF a no brainer for the K.I.S.S philosophy, but then you were earlier criticising how limiting standalones are vs OEM ECUs.   The joy of modern OEMs (and high end stand alones like Motec M1s) is there is a fcuk tonne of calibration to be done to make it work right but if you do then you get something that is super consistent and responsive to transient conditions.   I believe much of the reason that the Motec M1 (and I assume modern Porsche ECUs) do this kind of thing is there is a latency between what the MAF is reading and what the engine is actually receiving.

I hope this kinda makes as much sense as it can and it doesn't end up looking like wordsalad when I read it tomorrow after a good night sleep. 

Edited August 16, 2023 by Lithium

[Komdotkom]

I hear that Emtron have copied the Motec M1 torque mapping model. So now it's a Link ECU running Motec software, great....

[GTSBoy]

On 8/13/2023 at 9:04 PM, IM-32-FK said:

you are wrong. Factory gtt doesn’t not run a n1 pump. That’s false information.

On 8/14/2023 at 7:43 AM, Dose Pipe Sutututu said:

I might be wrong

Dose is not wrong. The Neo 25 had N1 pump from factory. This has been known for.....maybe 20 years.

On 8/14/2023 at 11:04 AM, IM-32-FK said:

But my common sense says that since the wideband has a greater range of parameters than a narrowband, and it is much more diverse and comprehensive, then once you tune it and chuck the narrowband back on, the ecu won’t be getting the same values and readouts therefore the car may not run optimally. I might be wrong (probably), I’m not an expert

Um. No. A wideband has absolutely NOTHING that a narrowband does not have, except for what's in the name. The wideband works over a wider range of air fuel ratios, allowing you to see the actual air-fuel ratio (well, infer it from the exhaust oxygen mole fraction, anyway). The narrowband switches between 0V and 1V just slightly to either side of the stoichiometric point. The wideband is a better tuning tool for that reason. But you do not get some other mysterious parameters reported from it.

Also, you cannot connect a wideband to an ECU that expects a narrowband, and vice versa. So, when a tuner is using a wideband..... it is not connected to the ECU. It is a separate thing that he is looking at. The ECU should still have its own wide or narrrow band O2 sensor connected and doing what it does.

Most ECUs, including old stock Nissan ones, CANNOT handle a wideband. They must have narrowband, If you install a wideband O2 sensor into such a car, it either has to be separate (for display purposes only), or if you want to use it to feed the ECU, you have to use the narrowband emulation output on it to feed into the ECU.

 

[Lithium]

On 17/08/2023 at 7:40 PM, Komdotkom said:

I hear that Emtron have copied the Motec M1 torque mapping model. So now it's a Link ECU running Motec software, great....

Is that after the initial TMF implementation?   I had wondered how well that worked for the same reason @joshuaho96 mentioned - the impression I had from others is it basically used measured pressure drop across the throttle, but I wasn't certain that you'd get a reliable reading post-throttle in all situations.  I don't really know what it was (or is) now so treat the whole thing like I need to really know more about it - definitely consider anything I say on this bit as conjecture   I have been told the current Emtron setup works well.

Edited August 19, 2023 by Lithium

[Kinkstaah]

8 hours ago, GTSBoy said:

.....So, when a tuner is using a wideband..... it is not connected to the ECU. It is a separate thing that he is looking at. The ECU should still have its own wide or narrrow band O2 sensor connected and doing what it does.

Most ECUs, including old stock Nissan ones, CANNOT handle a wideband. They must have narrowband, If you install a wideband O2 sensor into such a car, it either has to be separate (for display purposes only), or if you want to use it to feed the ECU, you have to use the narrowband emulation output on it to feed into the ECU.

I mean yes what you are saying, but more modern ECU's do indeed use wideband. But you can't just employ a wideband where a narrowband is wanted and assume it's all going to be good (which is what you are saying)

Haltech, Link, and even the OEM LS1 ECU can indeed use a wideband (in addition to narrowbands) and utilize it/report on it, give you very useful tuning data based on it.




(examples being commanded AFR vs Actual/Measured AFR in % so you can then make changes to the base map etc.) This discussion with regards to closed loop O2 is making these changes on the fly in Haltech and similar ECU's.

The LS1 ECU I have to go and input the 'errors %' back into the base map as appropriate. Which is as easy as copy/pasting the resultant figures ontop of the base map as a percentage.

[joshuaho96]

On 8/16/2023 at 2:50 AM, Lithium said:

 

• Talking about TMF?:  I am talking about something along those lines, yes - though I assume you are talking about Emtron's TMF specifically?   That is one implementation of a speed density system that happens to use throttle as an input but I wasn't talking about their implementation.  In terms of the the pressure ratio across the throttle body concern, yeah I'd "day dreamed" about alternative ideas for determining airflow and had pondered on pressure across a throttle and ruled it out in my head because of that exact reason.   I've never used Emtron, let alone their TMF so I don't have a great understanding of how it works or how well it works so I can't really comment on it in context of that.  
   
• How many cars use "TMF" style speed density and how well it works: I don't think the E9x series BMWs use it though I'm not 100% sure.  And to be clear, I am far from claiming I am any authority on these things - much of it is what I remember from conversations over the years with people much smarter than myself so I'm kinda forwarding on my patchy recollection of how I interpreted what I was told about things.  In terms of how well it works, I feel like I already said that from what I understand the new Porsche's use this kind of system.  Do you doubt the 991 Series Porsches work well?  I've only been in a single 991.2 Turbo S and it went like a bat out of hell, but it very much felt sorted beyond "only WOT" type driving by my personal gauge.  Super super competent car all round.   Like almost too well behaved.
   
• Not a fan of Alpha-N/MAP strategies: Can you elaborate?   They can be a headfcuk to deal with but I'm not sure they are broken as such.   I definitely prefer the concept of pre-throttle pressure vs throttle angle versus an engine flow characterisation map and feel like that's probably the best answer though am not aware of an aftermarket ECU offering that.  Yet.

Now for the "my view of the world" ramble, again I am no expert so more than happy to be picked apart on this if there is stuff you (or anyone) knows beyond me so this is just my take and limited experience on this general topic.  Please excuse any terrible wording or thinking with this, its been a long day at work and if I were wiser I'd write this when my brain wasn't already done - but it's a fun topic so I'm doing it immediately

What I've gathered about at least the 991 series Porsches they use a combination of a boost sensor and throttle area vs angle calibration, though I could be wrong - am definitely riffing off bits and pieces I've been told.   I *think* it's the 991.1 that has just a pre-throttle pressure sensor and the 991.2 has one on both sides (the post throttle one being more for diagnostic purposes
?!) - I'm guessing from that there are possibly a bit more like how the Motec R35 GT-R package does things, where it uses pre-throttle pressure and throttle opening vs flow data and engine efficiency data to estimate manifold pressure on the other side of the throttle.   So if you think of an Emtron which (if I understand correctly anyway) is solving for mass flow using pre-throttle pressure, post throttle pressure and throttle angle - these guys are effectively solving for MAP by using pre-throttle pressure, throttle angle data and engine flow data.   I could be wrong, but if it's anything like that then in my head it's going to get around the situation where there isn't a large pressure differential across the butterfly, but also creates a reasonable amount of calibration work to make it work right.

In that sense it makes MAF a no brainer for the K.I.S.S philosophy, but then you were earlier criticising how limiting standalones are vs OEM ECUs.   The joy of modern OEMs (and high end stand alones like Motec M1s) is there is a fcuk tonne of calibration to be done to make it work right but if you do then you get something that is super consistent and responsive to transient conditions.   I believe much of the reason that the Motec M1 (and I assume modern Porsche ECUs) do this kind of thing is there is a latency between what the MAF is reading and what the engine is actually receiving.

I hope this kinda makes as much sense as it can and it doesn't end up looking like wordsalad when I read it tomorrow after a good night sleep. 

I can't really speak to modern cars, the calibration strategies are pretty opaque, probably because even the people cracking these things open don't seem to understand how any of it works. It seems viable, I just can't be bothered to take the risk at the moment trying to implement all of that on my end.

The objection I have to alpha-N and blended strategies on things like Haltech and Link is that while it's nominally a simple thing to map it's a very brittle load sensing strategy that relies on too many assumptions to stay constant that I don't think will stay constant and requires a huge amount of additional compensations to function in an ITB turbo configuration like the RB26. It's not actually modeling mass flow through the throttle body or anything like that. Something as simple as AC/power steering/alternator generating more load at idle just isn't really accounted for without extra tables and has to be compensated for with all kinds of weirdness. The use of manifold pressure as the primary load scale for timing also just seems weird to me because it's non-linear with respect to actual cylinder air mass so at part throttle you basically have no resolution in your timing tables in many areas. My conception of a TMF model is that it should be able to overcome all of these concerns once set up correctly while not being hilariously slow and generally unreliable in transients like MAFs stuck to the very front of an incredibly long intake tract.

I have no objection to speed density when applied appropriately either, I helped a friend tune a speed density setup on a Jaguar V12 recently and it works great when it isn't subjected to the weird idiosyncrasies of ITB turbos.

[Lithium]

On 20/08/2023 at 8:01 AM, joshuaho96 said:

The objection I have to alpha-N and blended strategies on things like Haltech and Link is that while it's nominally a simple thing to map it's a very brittle load sensing strategy that relies on too many assumptions to stay constant that I don't think will stay constant and requires a huge amount of additional compensations to function in an ITB turbo configuration like the RB26. It's not actually modeling mass flow through the throttle body or anything like that. Something as simple as AC/power steering/alternator generating more load at idle just isn't really accounted for without extra tables and has to be compensated for with all kinds of weirdness. The use of manifold pressure as the primary load scale for timing also just seems weird to me because it's non-linear with respect to actual cylinder air mass so at part throttle you basically have no resolution in your timing tables in many areas. My conception of a TMF model is that it should be able to overcome all of these concerns once set up correctly while not being hilariously slow and generally unreliable in transients like MAFs stuck to the very front of an incredibly long intake tract.

Yeah don't get me wrong, AlphaN I generally view of as kindof a "necessary evil" in almost all the cases where I have used it.  As you say, where there is unmetered load changes things can get a bit tricky. For me personally, "throttle based load" systems are best done when you can control the throttle with the ECU.  The most recent Alpha-N setup I tuned was a Toyota MR-S race car running a Honda K20A with E-throttle on a MaxxECU, and that works great.  The big thing is idle control is done via the ECU controlling the throttle, so you (so long as the tune is accurate) have very fine load updates when idle load changes.   As a lot of the things you have mentioned come down to, it will only work as well as the tuner and setup allow - a lot of these things introduce compromises, extra effort etc and really like everything when modifying cars you have to consider what is going to best suit the application.

Another good point you raise and one thing I've not really mentioned publicly about a lot of my tuning is that (depending on the application) I've often used estimated mass flow type values for axis on load and AFR tables then if there is anything dealing with temperature swings then that will be done with extra adjustment.    It makes AFR target tables (and traces through a pull depending on how you've done it) as well as ignition timing tables look quite different but potentially quite a bit more intuitive - like all things though, this kind of thing relies highly on the work you've put into it being accurate and complete.  One of my my mates finished a KP Starlet build a couple of years ago and has mega fussy standards overall and he wanted something that was going to be a mix of oldschool raw but also "sorted" so cold starts, driveability etc etc were like clockwork.   Like it had to be raw, but refined, and runs a Link ECU.    We had many yarns about how to do go about it hardware wise and while the ECU isn't necessarily the best suited to it, I decided to try wrestling into doing things I'd do with a MoTEC or whatever and basically tuned it with two MAP compensated AlphaN VE maps which blended between based off the airflow error between the minimum and maximum request idle air valve work and did the ignition table based off the calculated cyl filling.   I'm not saying this is a great idea or anything, buuuuut the car does exactly what he wanted.   It's economical, starts and drives better than a lot of factory cars despite having ~280deg cams and ITBs on a nearly 40 year old base engine so I'd argue that one way or another it's a combination which "can do the job".   It has no wideband (lots of wires and obvious sensors partly not in fitting of the feel of the car - cool wee car actually: Toyota Starlet | Kelford Cams) yet he has done road trips all around NZ in it covering a variety of altitudes etc with zero issues in terms of shitty running or anything so it can be done even with a crappy old Link ECU.   Throttle response is pretty electric, nothing stupid happens, you would not want a thing like this to be with an AFM only load input for all kinds of reasons.  Like everything, it partly relies on how much work you want to do to get it "right" and how well you understand an define things.   

Your comment on how non-linear the relationship between MAP and actual cylinder fill is when you're "in vacuum" is a whole other wild can of worms haha.  I fully agree, I feel like you can often tell if someone who has tuned a purely manifold pressure based speed density tune understands (or at least is as OCD as I am) about a lot of the fundamentals of this kind of thing when you look at a VE map.  This whole yarn really makes me think of a rant Paul Yaw (Mr Injector Dynamics) had years ago and is very relevant to especially this part of the conversation - if you haven't read it then this may be of interest: That Big Orange Heavy Thing | Injector Dynamics.   The relevant bit is "Math doesn't suck" and check it out if you're in an environment where you won't get awkward questions about looking at a webpage with pics of sexy school teachers on it.  One way or another it's worth a look.

On 20/08/2023 at 8:01 AM, joshuaho96 said:

I have no objection to speed density when applied appropriately either, I helped a friend tune a speed density setup on a Jaguar V12 recently and it works great when it isn't subjected to the weird idiosyncrasies of ITB turbos.

Nice     I did ask before but still curious what the specific ITB turbo idiosyncrasies you're talking about?   Another thing on the topic of speed density, ever tried IMAP/EMAP based load?

Edited August 25, 2023 by Lithium

[discopotato03]

I'd certainly like to know more about tuning with an EMAP input .

[Lithium]

I clearly easily get carried away, so I'm gonna restrain myself haha.   Basically you have MAP/EMAP as a axis - so the pressure ratio across the engine.  It flattens the hell out of the map as the effect on fluid flow from pressure ratio is part of what influences the VE map in the first place and also it effectively bakes in more tolerance for the effects of ambient pressure changes and also general outside influences on the tune (such as transient conditions caused by what the external wastegate is doing etc, can be very interesting looking at that ratio when the gate cracks).  

That is definitely a popular choice for serious hillclimb cars (like Pikes peak levels of altitude change) fwiw, where it isn't just ambient pressure on the intake side that changes for a given throttle angle but with turbo cars the EMAP will go higher and higher for a given manifold pressure target which can cause serious tune drift for something just focussing on intake air density for the fuel model.

Edited August 26, 2023 by Lithium

[joshuaho96]

On 8/24/2023 at 6:49 PM, Lithium said:

Yeah don't get me wrong, AlphaN I generally view of as kindof a "necessary evil" in almost all the cases where I have used it.  As you say, where there is unmetered load changes things can get a bit tricky. For me personally, "throttle based load" systems are best done when you can control the throttle with the ECU.  The most recent Alpha-N setup I tuned was a Toyota MR-S race car running a Honda K20A with E-throttle on a MaxxECU, and that works great.  The big thing is idle control is done via the ECU controlling the throttle, so you (so long as the tune is accurate) have very fine load updates when idle load changes.   As a lot of the things you have mentioned come down to, it will only work as well as the tuner and setup allow - a lot of these things introduce compromises, extra effort etc and really like everything when modifying cars you have to consider what is going to best suit the application.

Another good point you raise and one thing I've not really mentioned publicly about a lot of my tuning is that (depending on the application) I've often used estimated mass flow type values for axis on load and AFR tables then if there is anything dealing with temperature swings then that will be done with extra adjustment.    It makes AFR target tables (and traces through a pull depending on how you've done it) as well as ignition timing tables look quite different but potentially quite a bit more intuitive - like all things though, this kind of thing relies highly on the work you've put into it being accurate and complete.  One of my my mates finished a KP Starlet build a couple of years ago and has mega fussy standards overall and he wanted something that was going to be a mix of oldschool raw but also "sorted" so cold starts, driveability etc etc were like clockwork.   Like it had to be raw, but refined, and runs a Link ECU.    We had many yarns about how to do go about it hardware wise and while the ECU isn't necessarily the best suited to it, I decided to try wrestling into doing things I'd do with a MoTEC or whatever and basically tuned it with two MAP compensated AlphaN VE maps which blended between based off the airflow error between the minimum and maximum request idle air valve work and did the ignition table based off the calculated cyl filling.   I'm not saying this is a great idea or anything, buuuuut the car does exactly what he wanted.   It's economical, starts and drives better than a lot of factory cars despite having ~280deg cams and ITBs on a nearly 40 year old base engine so I'd argue that one way or another it's a combination which "can do the job".   It has no wideband (lots of wires and obvious sensors partly not in fitting of the feel of the car - cool wee car actually: Toyota Starlet | Kelford Cams) yet he has done road trips all around NZ in it covering a variety of altitudes etc with zero issues in terms of shitty running or anything so it can be done even with a crappy old Link ECU.   Throttle response is pretty electric, nothing stupid happens, you would not want a thing like this to be with an AFM only load input for all kinds of reasons.  Like everything, it partly relies on how much work you want to do to get it "right" and how well you understand an define things.   

Your comment on how non-linear the relationship between MAP and actual cylinder fill is when you're "in vacuum" is a whole other wild can of worms haha.  I fully agree, I feel like you can often tell if someone who has tuned a purely manifold pressure based speed density tune understands (or at least is as OCD as I am) about a lot of the fundamentals of this kind of thing when you look at a VE map.  This whole yarn really makes me think of a rant Paul Yaw (Mr Injector Dynamics) had years ago and is very relevant to especially this part of the conversation - if you haven't read it then this may be of interest: That Big Orange Heavy Thing | Injector Dynamics.   The relevant bit is "Math doesn't suck" and check it out if you're in an environment where you won't get awkward questions about looking at a webpage with pics of sexy school teachers on it.  One way or another it's worth a look.

Nice     I did ask before but still curious what the specific ITB turbo idiosyncrasies you're talking about?   Another thing on the topic of speed density, ever tried IMAP/EMAP based load?

For the RB26 ITB + turbo causes a lot of weird issues I think. Most of them are just ITB problems. At idle or low load IMAP is a very noisy signal because there's not that much vacuum volume. You can sidestep this with aggressive filtering or sampling based on crank angle but that makes it a slow signal. Then once you start using the throttle while I can clearly see vacuum during some part throttle cruise cases it easily saturates to atmospheric and spends a lot of time sitting around there for a solid half or more of the throttle travel until it spools the turbo. During cold start you also have to account for the fairly scattershot behavior of the cold start valve. The ECU doesn't even know of it's existence and it operates off of a very eyebrow-raising method of heating a bimetallic strip as soon as the fuel pump gets power to slowly close the shutter. It can get stuck open or start closing at a different rate with age so its relationship with actual engine warmup is loose to say the least. Then of course wastegate duty and all of that fun stuff affecting EPR also needs to be compensated for. My car has 4 cats on it from CA's emissions compliance requirements which has made a very, very noticeable effect on how much cylinder filling occurs despite running wastegate boost in both cases on an otherwise similar car. It used to hit the R&R corner of the map, now it actually doesn't exceed the factory intended load scale. I don't want to bother with tuning an engine twice in response to something as simple as a different exhaust.

I've never tried EPR-based load to see what happens there but maybe in the future I might be able to muster the effort to evaluate it against the MAF load data to see if it helps tame the non-linearity problems. For now I plan on just using the factory MAFs and building a MAF load tune, then going from there. It's coming along well but the final 15% to make it drive like a factory car is taking 85% of the effort.

[Lithium]

19 hours ago, joshuaho96 said:

For the RB26 ITB + turbo causes a lot of weird issues I think. Most of them are just ITB problems. At idle or low load IMAP is a very noisy signal because there's not that much vacuum volume. You can sidestep this with aggressive filtering or sampling based on crank angle but that makes it a slow signal. Then once you start using the throttle while I can clearly see vacuum during some part throttle cruise cases it easily saturates to atmospheric and spends a lot of time sitting around there for a solid half or more of the throttle travel until it spools the turbo. During cold start you also have to account for the fairly scattershot behavior of the cold start valve. The ECU doesn't even know of it's existence and it operates off of a very eyebrow-raising method of heating a bimetallic strip as soon as the fuel pump gets power to slowly close the shutter. It can get stuck open or start closing at a different rate with age so its relationship with actual engine warmup is loose to say the least. Then of course wastegate duty and all of that fun stuff affecting EPR also needs to be compensated for. My car has 4 cats on it from CA's emissions compliance requirements which has made a very, very noticeable effect on how much cylinder filling occurs despite running wastegate boost in both cases on an otherwise similar car. It used to hit the R&R corner of the map, now it actually doesn't exceed the factory intended load scale. I don't want to bother with tuning an engine twice in response to something as simple as a different exhaust.

I've never tried EPR-based load to see what happens there but maybe in the future I might be able to muster the effort to evaluate it against the MAF load data to see if it helps tame the non-linearity problems. For now I plan on just using the factory MAFs and building a MAF load tune, then going from there. It's coming along well but the final 15% to make it drive like a factory car is taking 85% of the effort.

I like this discussion. Yeah, those are fair points and some of them circle back to an earlier comment I made on Alpha-N type strategies - they're much nicer if you can actually control the throttles yourself, same goes for TMF type ones where you don't have to worry about anything beyond the throttle itself affecting airflow into the engine.   Ultimately seems like we're on a similar page in terms of things we've seen but potentially feel differently about what is acceptable.   I have been happy with my experiences with speed density mapping, am aware of it's shortcomings and strengths and it tends to suit the purposes of what I've dealt with but yes.  I can't ever see a situation where it's not going to need a lot more calibration effort than MAF does, but then MAF is never going to be as accurate in transient conditions.   So far the "best driving" cars I've driven are speed density based but I guess there is the other factor we touched on earlier, some OEMs actually do both to try and get the best of both worlds haha.

This definitely gets me wanting to play with M1 Dev again and try dumb ideas.  

[GTSBoy]

Is this not why the later Nissan ECUs (is Neo onwards) stuck with AFM for primary load index and added Alpha-N on for throttle transient handling?

I'm not sure that the Alpha-N implementation in the Nissan ECUs is equivalent to outright Alpha-N would be as the main load strategy, because they just seem to use it as a trigger to throw the main fuel map to the max load column, but I guess that any sort of blended strategy would always mean that one or the other of the strategies being used is bent or broken compared to when you use it alone.

[IM-32-FK]

This setup is costing me way too much that it’s kinda turning me off it. I just realised I’m gonna need a new twin plate clutch kit since my exedy single plate heavy duty clutch (610nm) won’t ever be able to hold 380kw. And now because it’s pull I gotta pay like $800 extra as opposed to a push type (pull to push conversion comes in the clutch kit). Does this sound normal? - $25k in parts only to reach 380kw power level? Plus $4k in labour and about 2k in tune? I swear to god I could’ve bought a brand new 2023 Mazda 3 for that price.

I got no choice though, as I’m too deep now. I got my eyes on an Xtreme organic solid centre twin plate (kni23582-2g) that has a 1210nm rating. Costs $2300 plus maybe $700 in labour. I mean i might as well get something that can hold something like 750hp in the future. I never used xtreme but are they good enough?

[Kinkstaah]

You can always turn the boost down.

This is why I say 320kw is the sweet spot for basically everything. Once you go above that you have to do effectively the same things for 350kw (reliably) as you do for 650kw ("reliably")

[admS15]

Most people around here use and recommend NPC clutches. @Dose Pipe Sutututuhad I believe their single plate carbotic and was holding 375rwkw for years with no issues. Not sure what he's running now. What are you running now Mr Dose?

 

I run their organic single plate and has been holding 315rwkw easily for around 6-7 years now. I've kept my setup at this power level as too much more gets exponentially more expensive.

[Dose Pipe Sutututu]

5 minutes ago, admS15 said:

Most people around here use and recommend NPC clutches. @Dose Pipe Sutututuhad I believe their single plate carbotic and was holding 375rwkw for years with no issues. Not sure what he's running now. What are you running now Mr Dose?

It's still the same clutch from 10 years ago, and 2x motors ago. Clutch has seen 3x motors in its lifetime 😂😂😂

It's holding 437kW at the rears and gets used on the track, albeit it's starting to give up 😂

If you quick shift between gears it slips

[admS15]

On 22/09/2023 at 9:08 PM, Dose Pipe Sutututu said:

It's still the same clutch from 10 years ago, and 2x motors ago LOL.

It's holding 437kW at the rears and gets used on the track, albeit it's starting to give up 😂

If you quick shift between gears it slips.

So U still use and recommend NPC or would you go something different next time?

[Dose Pipe Sutututu]

9 minutes ago, admS15 said:

So U still use and recommend NPC or would you go something different next time?

I might try out a Xtreme twin plate sprung centre organic.

I've spoken to a few GT-R owners and they said it quite friendly to drive and probably a bit less bitey than what I have on my car at the moment.