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 am new to ECU remapping, I have working knowledge of how an engine works etc but no knowledge of mapping. I am a software developer so can easily learn about software etc.
I am looking at training courses available but they are either very expensive £1800 for 4 days. Or very cheap £400 for 2 days, but the cheap ones seem to only cover how to read and flash maps other people have made!
I want to be able to write maps from scratch and eventually do road / dyno maps.
I would start by trying on my own car, but my car has an ECUTek map which I believe will now be locked?
So my question is, for people who remap cars, where did you start? How did you learn? What tools do you have? Best way to start?
Quite a crowded market. With a lot of slave tuners flashing diesels cheaply.
To differentiate yourself and work on nice cars for good money will need expense, time, results, learning. Proper insurance, premises, support staff etc cost money. Tuning budget stuff gets frustrating due to customer attitudes regarding their responsibility to present a car in good condition for tuning and that you cannot just boost a TD04 to 40 PSI without consequences.
I started with tuning Subarus, but found more of a niche in adding flash tuning features for tuners, for which I've done commercial projects on GT-R and BMW and open source for Evo and BMW.
Look at your skills and interests and think a bit beyond the obvious furrow many try to plough would be my advice.
It is an industry that rewards results over qualifications which is a blessing and a curse. It is also an industry that markets training offerings to people that want to tune.
What sort of software developer are you?
Last edited by john banks; 24 September 2016 at 08:18 PM.
Have a look into esl. I was the same in regards I am very mechanically minded and have motorpor engone Erin degree but completely new to remapping. Spoke to Andy from enduring solutions and booked onto one of their courses for a mere £300. Can honestly say I was very much the confused newbie in the room but they took it at a perfectly slow pace for me. The lads will go through every parameter and how it will alter and affect each and every other parameter and performace. I'm not a dales rep by any means but from my personal experience was worth every penny and have since done plenty of data logging and small changed to have a much smoother car which pulls a lot harder than previously.
Completely agree John, it does seem like there are plenty of "cheap remaps" out there, where people just flash on prewritten maps on common cars. Many of the courses I have seen seems to teach you how to do just this. Pointless IMO. To be honest I think I am more interested in working on my own car and learning about it than I am doing it as a business
I have been doing loads of research on all aspects of tuning, AFR, knock, timing, boost etc. So hopefully I can get logging soon and start to understand the data.
I am a .net software developer, I mainly create database apps so nothing to do with cars at all but what does help is my understanding of things such as memory addresses, hex values, XML etc.
The Euduring Solutions course sounds good at £195 + VAT looking at their website but it is a little far away. But I will look more into this. Do they teach you how to use Romraider / ECUFlash?
Just had another closer read on their website, they seem to specialise in classics, I will be tuning my newage. But I suppose the principle of mapping will be the same.
To be honest I think I am more interested in working on my own car and learning about it than I am doing it as a business
That is always a good place to start, but look ahead and think about where the volume and opportunities might lie.
Tuning a Subaru is very different to tuning other stuff, can expand on this if you wish.
If you just want to tune using someone else's product or software, there are various providers. "EFI University" seems popular with some. OTOH, you could have a (real) degree in automotive engineering and be a poor tuner without the right experience.
I am a .net software developer, I mainly create database apps so nothing to do with cars at all but what does help is my understanding of things such as memory addresses, hex values, XML etc.
Are you into embedded, disassembly and reverse engineering?
Can you please expand on how tuning a Subaru is different to others?
I think the plan is to learn to tune a subaru first as there are open source tools available. I just need a tactrix cable. Then when I am comfortable, move onto other stuff, possibly VAG? Seems like tools such as Alientech / KESS stuff is good to tune most cars?
Had a look at EFI University, also seen HPAcademy which seem to offer similar online courses. Any of them worth doing do you think?
Whilst the basic method is measuring tooth times and deriving angles from rotating crank and cam triggers, incorporating a variety of analog sensor readings, making some lookups and calculations and scheduling the timing of ignition and injection (direct or indirect), most of the processes in between and the tuning approach between Japanese and European ECUs is very different, although there is some recent convergence.
Mostly it surrounds the depths to which torque (and everything else possible it seems) is modeled and controlled in the dominant Bosch and Siemens ECUs. Even a fairly recent Subaru ECU might have a fairly simple pedal to torque request but it quickly gets converted to items a tuner is more familiar with like load, throttle position, boost target etc. The Germans model everything right down to the wastegate spring, exhaust gas temperature, pressure and flow and the power required to drive the compressor with many 1d and 2d tables as a result of model analysis to break down the engine into a virtual model. It can seem elegant or frustrating depending on what you are trying to do.
To take an extreme example to caricature the situation, you could take an old Subaru with cable throttle and find that to tune it you spend time with tables that have boost targets, wastegate duty cycles, boost limits, airflow meter scaling, injector scaling, ignition timing and fueling. Often you will be leaning out the rich factory fueling. It is dumb enough that it doesn't properly correct knock, it has no idea if it is running lean, and doesn't really care what boost you put through it if you simply unplug one vacuum line for example and it won't complain about it. You are on your own pretty much.
On the other hand you can have an MED17 on a German petrol engine from the last 7 or 8 years and find that you are primarily altering tables that limit requested torque or engine load. To get more torque you might have to tell it that the gearbox is capable of more torque in a particular gear, you might have to alter a model of the surge or choke line of the compressor wheel. If anything you might richen the fuel target but once you have the closed loop widebands will just run it, but only if the modeled EGT is considered high enough to need it all. The ignition timing self tunes with continuous knock control and has theoretical models of the optimum timing it could run and the torque that would be related. Your job is more to see whether these are appropriate for the situation. The throttle is now adjustable intake valve lift with a separate process or even processor to monitor that it is doing what it should. The bypass valves are electronic, often the wastegates are too or operated by vacuum intead of boost. There are hundreds of checks and balances. It might not have a MAF sensor at all but it doesn't run speed density, it has a complete model of the intake system.
In terms of tuning, for much Japanese stuff there are well known companies that make tuning suites that have all the ECUs and their tables defined for you, and example maps and dataloggers. For VAG, to flash the latest cars through the port instead of having to take the ECU out (sometimes they stick it under the inlet manifold for "cooling") and break open seals, you often have to have £5 to 15K software. Then you still don't have the usually German functional descriptions and map descriptions or software to edit them, all of which you have to buy or swap. Then you have to be sure you are correcting checksums properly or the car won't start or you could even brick the ECU and have to take the inlet manifold off.
The functional descriptions for recent German ECUs running not very exotic cars are now 20000 pages. It can take a long time to understand how they work.
As another example, you can run a BMW at full boost at 14.7:1 and it will be quite happy. You could do the same with an old Subaru and add a few faults into the mix and it will probably put a hole in the top of the piston by the time you've done a lap. You can listen to a German engine not hear knock but find the ECU is pulling out 8 degrees of timing. You can clamp the pressure sensors on a BMW with the right hardware and a few tricks run 600 HP instead of the 300 HP it came with without even flashing the ECU.
Flexibility in thinking is gained from doing many different platforms, but I find I have to concentrate on one at a time for a minimum of a few months to understand it. Sometimes a few years.
Last edited by john banks; 25 September 2016 at 02:12 PM.
thanks for the info! Didn't know that about VAG stuff, this is why I prefer Japanese cars to German stuff. Everything VAG is just over complicated, over engineered, millions of sensors. But I suppose their factory tunes have to be clever to be able to get 300bhp out of cars but still return 40mpg.
Looking at this, I think I will definitely stick to Japanese stuff for now. I have a concept of how I would make more power with a Subaru by starting to modify the timing maps, boost maps etc.
Even though I am only going to look at Subarus for now, but I am curious, say if you were to start tuning a VAG car with the newer ECU you mentioned. What would be your process? What tables would you modify in what order?
Great reply John, just goes to show that alot of these guys reflashing ecus etc probably dont actually understand what they are doing. Its literally a man with a laptop loading someone elses rom that may or may not be any good.
I suppose until it blows up the average joe just wont ever know if the map is any good or not!
I found that 99/00 JECS with incomplete information and controls had knock control that was deaf once you were about 50% above stock torque. Just when you needed it. Or it let it knock long enough to hear it audibly in the cabin quite easily. Very few other factory systems have I experienced this and have disassembled the logic in the Evo and GT-R enough to trust them. Later Subarus are not a problem I gather.
What tables to adjust depends on the specific VAG. There is some good info in Cobb's tuning guides. Ecuconnections is an interesting site too.
Many tuners do not do more than apply someone else's map. On some cars this can work really well.
Since my last post, I have done a LOAD of reading and research on this topic, learning as much as I can. Installed a wideband to my car, bought a Tactrix cable and done a load of logging. This weekend I have mapped my car for the first time, it took me about 2 days and 100 miles worth of driving as I only wanted to make very small changes each time.
The results are fantastic! Car feels so good to drive, spools fast, more power midrange and holds power all the way to redline! Good start I think
right so - looking here we can see that your slightly overshooting then slightly undershooting your target boost:
Boost plot over RPM - manifold and target absolute boost
relative pressure might be easier to work with here but so long as target boost and your boost reading are the same type it's no big deal you can still see where your off. For the areas where you go past your target, your wastegate duty is too high so you'll have to reduce it. For the areas where your under target you need to increase the duty to up the pressure. Try and adjust your min and max tables together so that overall the balance and shape is the same.
Looking at your timing plotted out you can see where you have some work to do:
You can see that your timing drops (as it should) whilst the turbo spools up, however around 5000-5800 it kind of levels out, then drops around 6000rpm and picks up again. You don't have any fine or feedback knock recorded here so most likely you just need to look at your engine load on the timing table to see which cells your hitting and smooth things out better. You want the timing to increase as the RPM's climb up - the big dips and spikes will be dips in torque that you will likely feel ever so slightly in the car.
The AFR needs some work too looking at your wideband readings, you go as lean as 11.77 at 4500rpm where your on peak boost - whilst your not knocking yet you will be subjecting the engine to higher EGTs than necessary. You logged the "fueling final base" (the ECU's target AFR) as lambda so you'll have to log again or convert that to AFR if you want it to be useful (wee excel job). It's probably gonna be quicker to update your logging params to make sure you've got the right scale/format for each thing though.
The fuel final base should be pretty close to the actual AFR recorded on the wideband. You can see even in the lambda reading the target moves around a little bit as the revs climb so again look at the engine load figures there and look at the fuel table in your map to see where you need to add a little fuel in (make the target lower/richer). If the fuel final base and the wideband readings are way out then you could have an air leak or the MAF scaling needs a little bit of adjustment. In reality so long as your wideband is reading a decent and consistent fuel reading then a little bit of error in the MAF can be lived with. It may be quicker/easier just to lower the primary open loop fuel target in a few other cells just until your getting a steady 11.0/11.2 AFR throughout your power runs. The rest of the fine tuning can come later.
That should be enough to keep you busy for a while, report back with how your getting on
I will reduce the WGDC when I overshoot and turn it up a bit where I undershoot the target boost.
The timing I agree needs work looking at the graph. So ideally the timing curve for a boosted car should be like a U shape, drop as boost climbs and raise back up as rpm rises.
My AFR I am happy with to be honest, I am targeting 11.2 but the wideband is placed after the cat so the reading is a bit lean. Looking at closed loop and idle numbers, it is about 0.3 - 0.5 lean. My learning view is only showing around 2% correction across all 4 ranges so happy there isn't a leak.
More tinkering to come!
Last edited by supersai101; 24 October 2016 at 07:53 PM.
the ltv corrections will only be relating to close loop fuelling, not open loop. The sensor is in the headers so when you get on boost the gasses pass too quickly for an accurate reading on the front o2, that's why the ECU ignores it.
Even if your target is 11.2 - if your fuel target is fairly flat then your wideband reading should be fairly flat as well, I'd do a few runs just to check out what's happening as you hit full boost and also reduce the AFR a little during spool up to get you going a bit sooner.
As for the timing, yes you want a bit of a U shape to it, there should be a dip as you spool up then slowly ramping the timing back up until you reach peak horsepower. A few flat areas you can live with but you definitely don't want the timing going down and back up again.
I thought the LTV corrections are learnt during closed loop and that the last range 40+ g/s will be used during open loop. That log was recorded not that long after reflashing the ECU which resets the LTV corrections. I will let the LTV settle after a bit of normal driving and try again.
I will reduce the AFR a bit during spool and see if that makes a difference
I have started using various timing editors now and can see the total timing easier, it makes a lot more sense now. I have found the issue and have smoothed it out big time. Just need to flash this change back and try it.
I found out why my AFR takes a bit of a dip around 3400 rpm to rich then climbs back up and then goes stable. I found a rogue cell in the fuel map that was set to much lower than everything else! Left over from the stock map. All sorted now I think
Looks like you are having fun. As an illustration of the differences between different engines, you could take some direct injection engines running the same boost and have it run lambda 1 (14.7 AFR if using petrol). The same setup might then richen as it measures or models that the EGT will be rising. It might also have a rising boost curve after the midrange to give a flatter torque curve, more predictable traction where traction limited in lower gears and mask the turbocharged nature of the engine, as well as take into account marketing or engineering limits for torque to protect engine/gearbox/driveline.
Tuning some engines is almost cheating (once you obtain or make the map definitions and buy or work around the security), you just tell it you want more torque and it does it. You might massage a fuel target map because you might decide the torque level you want should have a richer AFR in the midrange so that it stops the EGT getting so hot and then dumping fuel. There might be no other gain to be had from tweaking ignition or valve timing maps until you heavily modify.
Last edited by john banks; 25 October 2016 at 07:34 PM.
It is very fun! But also a bit annoying at times as I have found while trying to dial in my timing. The problem I have with my rom is I find the timing table resolution to be too low. A lot of times I find myself wanting to adjust timing at a range that sits between 4 cells lol. I have since rescaled the table, the low loads columns were actually identical so I got rid of them in favour for having more columns at the high load range. This is particularly important now I think as upping the boost has put me in a load range previously not reachable.
Getting there, my curves are looking a lot better now than when I first started, a lot smoother.
Last edited by supersai101; 27 October 2016 at 08:50 AM.
Smoothness is an important thing to develop, get comfortable with your own methods, whether that is using features in the editing software or otherwise. I have seen a lot of degradation in quality from larger timing tables because many end up making them bumpy. Occasionally there is a reason for a bump (perhaps more likely with funny exhaust manifolds or engines with fixed valve timing), but most of the time the mapper isn't thinking about smoothness and interpolation. Many modern engines and their management go to great lengths to achieve a flat torque curve from variable valve timing and lift, variable intake systems and models of everything that is not a linear contribution where all the variables have been analysed to shake out the inconsistencies. The complexity is moving away from calibrating individual engine specs towards accurately modelling everything once. The end game might be intended to use fewer lookup tables or maps, but that doesn't seem to be happening yet.
This may or may not help you, but some struggling with this seem to get the point if as an exercise they make 3 x 3 timing and AFR target tables instead of 20 x 20 or whatever it starts at. By having to choose RPM and load breakpoints you think about what the engine really wants and can throw away the noise of a big table.
Last edited by john banks; 27 October 2016 at 10:59 AM.
I have been working more on my map smoothing things out a lot more
So much better than before especially when comparing to the graph posted earlier.
Going to play around with getting the turbo to spool faster next.
I understand there are 2 school of thoughts, running leaner to get heat in the exhaust, or run richer and less timing so the fuel is still burning in the exhaust, again to do with adding heat.
Been playing with lean spool a bit but to be honest I find that not to work for me. I have found that dropping the timing a couple of degrees low down and adding more fuel feels better. But will continue to play around, although may have to take a short break as I have spent £100 in fuel this week playing around :/
Going to play around with getting the turbo to spool faster next.
I understand there are 2 school of thoughts, running leaner to get heat in the exhaust, or run richer and less timing so the fuel is still burning in the exhaust, again to do with adding heat.
Been playing with lean spool a bit but to be honest I find that not to work for me. I have found that dropping the timing a couple of degrees low down and adding more fuel feels better. But will continue to play around, although may have to take a short break as I have spent £100 in fuel this week playing around :/
On the quicker spool , i too have been dabbling with this and like yourself have found Richer AFR`s upto around 3200 and with timing ranging from low to mid 20`s before 3200 at low to mid load give better spool, when i say richer im talking 13`s , what AFRs are you using and timing at low to mid range ?
As an example at 2800 I am running around 23.5 degrees of timing with low - mid 12 AFR. So a bit richer than you. I did try running 13s but I found 12s better. I am at 0.6 bar of boost by this point. I booked myself on a dyno next week so interested to see how much power I have
if your planning to hit the dyno it may be worth preparing a few maps before hand with small adjustments to them and you may be allowed a quick flash between runs. Maybe try a few points leaner/richer with more/less timing and see what results you get.
also hopefully they can offer you knock sensor, manifold pressure and high speed AFR hooked up too to make the charts easier to reference later. Keep the laptop on for logging too if your allowed (can't see why not but some places may be funny).
Does no harm to have a little octane booster in the tank too - takes the edge of the higher intake temps you'd see on rollers vs real roads.
23 September 2016, 08:45 AM
Where to learn?
