Leonh

Im going to keep a thread going of my cal and it's updates as and when I put them in. Feel free to throw in any advise or critic

I run:
T72 turbonetics hybridised turbo (T4 body, T3 shafts.)
4x 850cc
twin fuel pumps (rewired)
SARD FPR - base pressure at 42psi.
Custom Vmount.
RIAT airtemp sensor.
AEM boost solenoid
OMP + Premix
Streetported block using Pineapple templates.

Extras:
2x EGT sensors (sensor in each exhaust runner)
1x EGT POST turbo
AEM AFR gauge wired back into the ECU.

The cal is dog rough at the moment but it was able to drive me from the UK to Germany and back. its running pig rich above 4k at the moment (By design at the moment.)

THIS CAL IS NOT FINISHED AND IS NOT INTENDED TO BE USED ON BOOST OR AT ALL. IF YOU USE IT ON YOUR CAR ITS AT YOUR OWN RISK.

sorry disclamer to ward of stupid

Leonh

Issues with the map so far:
4k+ way too rich, on boost and off.
Idle does not always settle down to 900.
Ignition timings not confirmed via EGT's yet.

arghx

iTrader: (3)

Not a fan of the timing maps at all. Look at what people have been sucessfully running on Haltech and Power FC for years. You've got too little leading advance in vacuum and lower boost and for pump fuel too much timing at higher boost levels and in the mid rpm range.

I've said in other threads that AEM's base calibrations have poor timing maps; it very much looks like they just tweaked maps for piston engines. And they have very little understanding about how split maps are done on turbo engines for street applications.

Your entire fuel map is controlled by the boost compensation table with the way it is currently configured. While that may be quicker for tuning, you lose a lot of resolution that way... might as well be jetting a carb.

You are on the right track for your base boost control duty cycle table. However I would lower it from 60% to about 30% to start off; this will reduce the chance of something going wrong. Read here about using the AEM boost control: https://www.rx7club.com/3rd-generation-specific-1993-2002-16/electronic-boost-controller-comparison-chart-952767/page2/

Under options--> engine start try setting your cranking advance from about 6 degrees to 15 or even 20. There's a good chance it will start faster.

As for the idle... there are a lot of parameters here. I suggest you log a bunch of the idle related parameters before you make any changes. Then you can see what's going on. Since I don't have the car in front of me I also suggest you try the idle tuning procedure explained in the AEM guide--it is pdf page 87 on my version.

Leonh

Thank you very much for the feedback <3 I really do appreciate it.

You make a good point with the boost comp map. I did have a LOT of problems getting it dialled in for vacuum. (too rich/too lean with the % of adjustment available) I will change it back to a normal map as the way it is sounds just like a "quick way" to tune a car. I'd rather do it right.

Ignition map:
I think I took somebody else's as a base and adjusted to start with but if this is a bad example could you please nudge me at one that looks to be along the right path? I'm wanting to get the EGT sensors wired in properly this weekend so I can log what exactly the temps are out of the exhaust runners.

Maybe the easy option would be to take the car to a professional tuner but part of me wants to do this myself. As I know I would spend the extra hours on the finer details.

Leonh

Made some changes:

Ignition map - Advanced in vacuum and taken some out under boost.
One area I've not touched is the idle area would this be worth changing?
Fuel map - using "maths" taken out the % added into fuel from boost comp. (Hell I would be surprised if it starts with this map) msec kept the same.
Boost - base control lowered duty down to 30%
Crank Adv - Changed to 20 will do some testing on car and lower it to suit.

none of this has been tested on the car yet.

arghx

iTrader: (3)

Here's the thing. If somebody doesn't have a lot of experience tuning turbo rotary engines, they are going to start with the maps provided by AEM and tweak them. And those maps provided by AEM are WAY off what is widely considered good practice on basically all other tuning platforms for this car. For examples of what other basemaps look like on another tuning platform, check out this thread: https://www.rx7club.com/haltech-forum-62/converting-power-fc-timing-maps-haltech-969764/

So with the AEM EMS one guy copies another guy's basemap who copies another, and each only makes minor changes--some motors survive, some don't, it's hard to say what caused it so the maps perpetuate.

Now, when somebody who has tuned rotaries on more popular Rx-7 systems (Haltech, Power FC) sees the AEM base maps they just discard them completely. A user on here, Aeka GSR, is part of the Power FC tuning group and switched to AEM for the additional features. I'm pretty sure he just translated his maps over.

If you take it somewhere, it is mostly out of your hands. A lot of people don't want to deal with it and they want somebody who will provide a service to them so that they can focus more on driving the car. However, even if the tuner is very very knowledgeable of both AEM EMS and FD specific stuff, there is only so much time they can realistically spend on it. So a lot of the little things like the driveability and various compensation tables can't be fully tuned by a professional in one session.

Main leading Ignition map is improved from the first one you posted; trailing split needs work. The idle ignition timing is an interesting thing--as far as standalones go, the AEM EMS is very powerful in the way it lets you set up the idle timing. It has full adjustability of a closed loop correction system.

Go under Advanced igntion --> Ignition trims --> ignition vs Idle rpm table

Generally speaking your best bet is to turn off all the various feedback systems in the EMS, including ignition timing and idle air control valve feedback, then get the idle stable by adjusting the throttlebody. If you are happy with it that way, just keep it and let it be simple like that (like an old carb'd engine with a dizzy). If you want more advanced smoothness and driveability, one approach is to dial your open loop idle air control, then switch closed loop idle air control, then set up closed loop idle ignition control. All of these systems are built into the stock ECU.

Boost duty cycle looks ok for now, pending real-world testing.

It's hard to say anything about the fuel at this point, except that for best results I would not run boost comp at all in vacuum/low load.

Cranking timing is something you have to play with, but as I said I have had good results with 15-20 degrees timing advance. The stock ECU actually runs 5.

Leonh

I can see how that would happen and think I fell into that trap

New Map:
Fuel map redone. (maths version FAILED misserably, It now overfues but this gives me safety to lower it gradually than a lean condition.)
Ignition split changed.
Target boost lowered
Few idle tweaks.

arghx

iTrader: (3)

Timing maps are looking much better. I see you adjusted your idle ignition feedback table (Ign vs Idle rpm). You may still need to tweak that a little bit. You're jumping to a 7 degree compensation so keep an eye out for some idle surging. You really should consider setting up your Air Temp Ign and coolant temp retard table for protection . The tuning group has some possible settings in the example maps found there. Air temperature fuel compensation is also going to be critical; I see you have some values in that table. You will probably need to tweak them more.

You also have the option of setting up (or at least experimenting with) knock control. We can talk about that later, you're not there yet.

I'm not sure how your tip-in response is, but you may want to start playing with the dTPS Accel table under acceleration fuel.

Leonh

Thanks for the feedback Arghx the car is driving a load better all over I've made a few more changes to the fuelmap, ignition retard due to air/water temps and a few other small bits. I've also set the o2 feedback values for when I start to use that function.

Cant really do much more on the fuel map right now as over the pond here its a little damp. Gives me time to read a lot of the info up on the group and boy is there a lot

General route im going to do with the fuelmap is to do some cruise runs, some 0psi runs and then a few runs at minimum boost 7.5 psi. log all of these and make changes to suit.

arghx

iTrader: (3)

Your map has come a long way. Keep up the good work and please keep me updated. I am looking to add example AEM EMS maps for the group, so as you get the high load driving dialed in it would be nice to have a copy of your latest map.

Setting up closed loop boost control (using throttle or gear-based), O2 sensor feedback possibly, and knock control would be some things to have. They will take a while to get working though.

Btw, if you are having stalling issues on decel (possibly due to a lightweight flywheel), you can adjust the dashpot on your throttlebody. You can also set up the "Hi Idle rpm offset" under Idle --> Options to set a temporarily higher idle speed when you pop in the clutch as you decelerate.

arghx

iTrader: (3)

updates? got a more recent map to post?

Leonh

No real updates sadly weathers been all over the place, need to configure the far gauge gain as it's out of sync with the ecu again. Mostly been playing with the fueling and had made some progress out of boost. Main stopper is the radiator fans stopped working, it's the fans or the wires relays tested. Also going abroad in a week I'll post what I have later

EB Turbo

iTrader: (1)

Arghx, what is your theory and practice with fuel map strategy? Speed density, speed density+ boost comp or, alphaN + boost comp?

I started using speed density and then as I learned more about the EMS and tuning stratigies in general I am now tuning exclusively with alphaN + boost comp. As well as having the most resolution in the EMS it is also the easiest to tune. Well less time consuming then speed density. I look forward to your comments and will gladly go into more detail if necessary.

EB Turbo

arghx

iTrader: (3)

I haven't messed around much with alphaN so I can't say too much about it from firsthand experience. I can see how combining it with boost comp would give you a combination of simplicity and flexibility, but I don't have enough seat time with it to say too much about it.

Overall I do prefer just regular old speed density without boost comp for a few reasons. First, I have the most experience with it from other engine management systems including reprogrammed OEM and standalones. SD does come in two basic flavors (volumetric efficiency table and directly mapping pulsewidth like the EMS and Power FC do) but you get used to each one.

While speed density only is more time consuming, it measures engine load better under the full range of driving conditions. Under low loads a lot of non-linearities are reflected in change of manifold pressure that rpm and throttle position can't reflect. That's why Alpha-N is normally used only on race cars. If you look at how Mazda has done things, they had airflow-based engine management on all the rotaries except the FD. They did speed density on the FD and Alpha-N on their race cars (787B).

The strength and the weakness of boost comp is that it is essentially a 2 dimension table with limited resolution. It's quick to adjust which is nice. However with spreadsheet tools and other techniques you can effectively accomplish a "boost compensation" effect on the main fuel map by multiplying cells a certain way. However the resolution is kind of low on the main map in terms of the total number of cells which is in itself a drawback... use of boost comp really depends on the application.

I know I'm jumping around a bit but let me say the one thing that I think has huge drawbacks is running speed density and controlling EVERYTHING with boost comp. You are giving up so much flexibility for driveability tuning. Might as well turn some screws on a carb.

EB Turbo

iTrader: (1)

You loose no resolution from using boost comp. Essentially all it is a fuel multiplier as manifold pressure changes. If you took your properly tuned fuel table applied the 1:1 values to the boost comp table and then properly scaled your fuel table you would know no difference(try tuning boost comp that way). The added benefit of boost comp is being able to have a low Microbit/sec number. Boost comp is essential because of the strategy of the EMS and the way it calculates actual pulsewidth via the Microbit/sec value. With input values being 0-255 in AEMPro the Microbit/sec value sets a scaled window for allowable pulsewidth in the main fuel table. High Microbit/sec values(80+) when making individual cell changes may change AFR up to .15-.25. This can be frustrating when trying to dial in steady AFR's. Low Microbit/sec numbers 20's-30's with one individual cell change you will see AFR changes of .025-.05.

Your concern is with the overall resolution(amount of breakpoints). I would agree it is definitely not a lot but is has a lot more then most.

The issue I have with speed density with aspirated cars is 10psi @ 50% throttle is completely different then 10psi @100% throttle. AlphaN with boost comp takes into consideration all of those events. plus the added benefit of being able to having a low Microbit/sec number. My cars have made more power, better fuel economy and I am a lot more confident in my overall open loop fuel table. A bit short on time at the moment, but I will be glad to answer questions or discuss this further if anyone would like. I think there are a lot of misconceptions and bad practices with the AEM. Im just trying to shed some light on those areas. Im glad that there is a good knowledge base on here already.

EB Turbo

arghx

iTrader: (3)

The way it is normally implemented, you have a bunch of cells all set to the same value (say all the positive manifold pressure rows in the main fuel map) and then boost comp adds in on top of that. Say I have, for argument's sake, 8msec of injector pulsewidth at 6000rpm at 120kpa in my main fuel map. Then I add more fuel to it based on the boost comp table. This is simple and easy, and as you've pointed out it works around the AEM EMS's quirks with microbits. However, it is still a multiple of 8msec. You lose the ability to change the slop of fuel vs rpm with respect to boost. To put it in calculus terms... you have to look at the partial derivatives.

Now, I will admit that this is a bit of a quibble on my part. You can certainly make it work in these applications using the typical boost comp method. I would prefer that the boost comp table had more cells in it though.
Yes, this is a pain in the ***. It comes with the territory on aftermarket engine computers. How much it matters really just depends on the application and whatever tolerances you may have.
Of course this depends on what you are comparing it to. Older Megasquirts have hardly any resolution. A newer GM, like a C6 Z06 has 32x32 timing maps (although the stock calibration doesn't really utilize all that resolution). Power FC is 20x20 for fuel and timing, which is good enough.

I agree with you here on a practical level. I'll put it to you this way: I wish the boost comp table were an actual 3 dimensional (say 16x16) map as opposed to 2D. I consider your strategy a successful work-around to limitations of this particular computer. Every system has its quirks and limitations.

Getting academic for a moment...

The OEM controllers tend to use a lot more interacting tables. If you look at old GM speed density (like on a Syclone) they have a bunch of VE tables that all add together to account for different conditions. Since that hardware is so old they are limited by the small maps but they make it work. Newer speed density may have a separate power enrichment factor or WOT fuel target tables. One example of this is Chrysler speed density architecture as implemented on the SRT-4 and PT Cruiser.

Bosch and a lot of other systems use what's called a "torque request" and "airmass request" strategy with modern speed density. So we use the accelerator pedal position sensor and other inputs to calculate the engine output torque requested by the driver in newton-meters. Then a series of functions and tables translates the torque request into an airmass request. The airmass request is converted into a target manifold pressure (since no MAF sensor is used) assuming full spark advance. Then the actuators (wastegate, electronic throttle, cam phaser) are manipulated to achieve the target airmass.

The engine output torque is then modeled in realtime and delivered by a CAN to the driving dynamics control modules (traction control, stability control, etc). These control modules can then request a torque reduction which feeds back into the control loop, ultimately resulting in less spark advance or manifold pressure. Actual AFR may be done in open loop with an enrichment factor or target AFR, or it could be done in continuous closed loop with really advanced stuff like a neural net (which is what you'd find on say a BMW 335 turbo model).

EB Turbo

iTrader: (1)

What axis do you want to add to the boost comp table? RPM? Where are you not able to modify fuel v rpm? That's what the main fuel table is for. I think you are just used tho the PowerFC maps that have a base fuel table and then the correction table. Two tables like that are not necessary. Especially when they have the same axes.

Boost comp is a mathematical way of tuning. I have tuned a few Rotaries with MoTec M4's. They were all set up to run a boost compensation strategy with speed density the way I received them. Motec still uses the same types of strategies in their latest ECU's. Our Mazda 6's use AlphaN with a box pressure comp.

What you have described is has nothing to do with actual fueling or fuel calculation. Those systems are for targeting boost and torque output. Those systems are MAF based and use a VE table strategies. Mazdaspeed3/6 has throttle is a function torque targeting strategy as well as a COBB implemented closed loop boost targeting system. This can be very difficult to tune when actual manifold pressure is critical. Lotus Elise/Exige has a throttle based torque targeting system. Because it is an NA engine is uses cam advance, high/low lift, timing and fuel to target torque values. Mazda RX-8's have calculated load limits but they are have no closed loop functions as far as timing or fueling control.

The explanations given from AEM on boost comp tuning do not properly describe how to understand and tune boost comp properly. If you like I will be more then happy to explain over the phone.

EB Turbo

Leonh

I've got somethings to mull over when I'm
Back in the uk it seems but for me boost comp always felt very swingy in the vac areas as by my original cal, many be I did it wrong but I liketo learn from experience..

Speaking of that just say a stray dog while on holiday in Moscow that knows how to use the tube system 0.o ineedmore vodka

arghx

iTrader: (3)

I am actually getting ready to move right now and I don't have any AEM-equipped cars I'm actively working on, but I will keep your offer in mind. Thanks.

Leonh

Okay so I've finally got the time to start working on this again, but the laptop battery crapped out.

Been working on the idle recently and I've hit a snag, The idle target is being stupid and not actually going for what Its been set to 80% of the time, it will try and hover around 1200 - then eventually come back down to 850. This is causing the car to hunt and generally be a bugger to get the idle dialled in. Any ideas what other variables could be causing the target to be higher than what it should be?

EB Turbo

iTrader: (1)

There are many things that will set the target idle speed. Idle Target base, High Idle RPM offset, RPM Offset vs Start Table, RPM Offset vs TPS Table. If your TPS is not calibrated properly and is showing 0<-5% at closed throttle, The RPM Offset vs TPS Table may be adding some target RPM. This may not be correct for you but can be a possible cause. Can you post screen shots of your idle template and your advanced idle template.

Another thing to watch is for stable AFRs at idle. Are you watching your Idle Learned Value? Where does it normally fall? 5, -10, 20? You also want to know where your idle adjustment screw is. I like it to be 2.5 turns out. This will roughly put my idle% v target table in the 30%-60%.

EB Turbo

Leonh

Been pondering this and I think it is the high idle target, I think the VSS is wrong, I seem to remember taking a log from the run and laughing at the speed the ECU thinks I was going.

Thanks for the input, didn't realise that option was hidden away by default. Will do some tests next time I'm round at the car.

When setting the idle screw (under TB) do you disconnect the IAC and aim for a set RPM range or just go roughly 2.5 turns?

EB Turbo

iTrader: (1)

The High Idle for VSS has a time limit. Unless your time limit is really high this shouldn't be too much of an issue. Watch the parameter idle target and see if that moves around a lot. Also make sure in your RPM Offset vs TPS Table that between TPS position 1-5 that the RPM is set to zero.

EB Turbo

Leonh

Right spent some time tonight and got it idle sweet. Seems the information in the AEM Tuner help for the "Idle FB dead band-" and "Idle FB Dead band+" are the wrong way round. I flipped them as per compared calibrations and now the RPM drops to set level. I've got the Idle learned value to a few percent either way of 0.

Now the lowest I can take it on a stable, non bogging down AFR is around 13.4 is that about what I should expect?

EB Turbo

iTrader: (1)

You want to have the Idle Learned Value at around -7%. This will over shoot the initial idle target and allow the closed loop to bring it down. That number you can play with. You may not need to go as low as -7%.

This depends on what emission control devices are on and working. That should be a good steady number for idle anyway.

EB Turbo