Ecotrons Engine Management

[husaberg]

E85 fuel, fuel injected Kawasaki 350 BigHorn, I think there will be more info on the Foundry thread. They also fuel injected a late model YZ250 Yamaha. Both are running the bigger Ecotrons type fuel pumps. E85 and Methanol are much easier to get right with EFI than petrol because they can be run way over rich well clear of the lean danger zone.

Follow the post below to read about the BigHorn and YZ250 EFI projects



That is the Ecotrons fuel pump you can see in the middle right of the picture.

http://www.kiwibiker.co.nz/forums/sh...aha-YZ-250-EFI

[speedpro]

Unless you have an earthing issue shouldnt the ECU be using the same AFR as shown on your gauge? Can you display the AFR from the ECU on a laptop and compare it to the gauge? That said tuning to 13.xx may not be that far off the mark even for an economy setting. I have just put a wide band in my Megasquirted Range Rover it it likes to idle at 13:1.

I might have this wrong but are you saying that the ECU is auto tuning a load(MAP)-rpm VE table by using a Speed-TPS table and associated AFR table? If that is the case wont that lead to all sorts of problems when you throw the turbo in the mix? The system seems a bit strange in that it sounds like it wont let you edit the auto tune table and or run from that table with auto tune turned off.

The ALM supplies a voltage to the ECU which varies with the mixture . If the ECU thinks that 1v from the ALM is Lambda=1, and it receives 1v, then as far as it is concerned the fueling is at lambda=1 and autotuning is complete. The problem seems to be that 1v is NOT Lambda=1. I can see 1000mV on Procal but the ALM gauge says 13.?:1. I'm sure I've seen a table where these values can be adjusted but can't remember where.

Autotuning has 2 separate maps in the ECU that are the same configuration as the regular manually tuned maps. 1 - Volumetric Efficiency(VE) map. This map uses manifold pressure versus rpm and has VE values entered and 2 - a load map which uses TPS and RPM and has load values entered.

I have configured my system to use the VE map up to 98% throttle. As I have a turbocharger fitted the throttle position is not a good indicator of load. Manifold pressure however is a good indicator and varies with throttle position and turbo operation.

Not allowing the user to edit the autotune tables makes perfect sense. There are tables which can be manually edited if the users wishes and you can switch between them as desired.

I suggest that the way to use the autotune function is to let it run and update the values in the autotune tables. In my case it is only altering the VE table. This information can be exported to a "csv" file where the table data is in the same format - rows, columns, & values, as the manually edited table. I then import those values to the manually tuned VE table. The manually tuned and auto tuned tables are now identical. I can then make minor adjustments to say the low speed/load values to make the engine run sweet if needed. I would have to stop autotuning and switch to manual tuned date for the manually tuned VE table to be effective. For a road bike this would be sweet. You can have one map which is perfect for economy and emissions and with a flick of the Eco/Power switch ramp the fuel up to the "Driver Desired Lambda" which would be set to .8? for best power.

In autotune by default with the Eco/Power switch set to Eco it tunes for Lambda=1. If you switch to Power it tunes to the "Driver Desired Lambda". Apart from taking time to retune the maps for the different Lambda it defeats the whole purpose of autotuning. You could of course leave the switch in Power whilst autotuning and just have it auto tune to your desired Lambda value. A better option is to do what I have suggested above.

In manually tuned mode the ECU assumes that the VE & TPS tables are tuned for Lambda=1 or the ideal mixture. In Eco mode it uses the tables to determine the fuel to inject. When you switch to Power it simply adds a percentage of fuel as determined by the difference between Lambda=1 and the value of "Driver Desired Lambda". If you have the Narrow Band(NB) Lambda sensor and were using it for feedback in Eco mode, it is no longer used in Power mode as feedback is disabled. Power mode is open loop and relies on the VE & TPS tables being correct to then add a % to in order to obtain the desired lambda. NB lambda sensors only work at stoichiometric and their output only switches if the mixture is richer or leaner. It doesn't matter how much richer or leaner so they cannot be used to monitor how rich a mixture is, only that it is richer.

In my case autotune is creating an autotune table that has the engine running a bit rich. I have exported that autotune data and then imported the VE values into a manual tune map. In excel I can simply multiply each cell by about .9 to reduce the fuel calculated which should bring the Lambda back to 1 whilst maintaining the overall profile. Loaded and run, in manual tune mode and using the manual tuned maps, I should obtain Lambda=1 when running the engine. I have actually done this and it seems to work, though I have only run the bike in the garage without load so only monitored the very low or no load areas.

I have also determined that my ECU by default has a set of autotune VE & TPS tables in the ECU, probably for a GPz250. If I enable autotuning and select "reset to default data" it loads my manual tuned VE & TPS maps. If you have spent any time sorting out your own maps you should make sure your maps, and not some default maps, are loaded before starting the autotune process. There is something else going on in the ECU around these maps. Supposedly you can "flash" the ECU and the autotune maps are unaltered but mine has reverted back to the default maps a couple of times but I haven't sorted when that happens. Another good reason to export the data and then import it into manually tuned maps.

Regarding which maps can be used when autotune is off, you can select which maps the ECU uses, either the auto or manual tuned maps.

[Cheesy]

There is something up with your wideband by the sound of it then, 1V should be somewhere around 11:1 and stoich around 2.5-3 volts depending on the controller. Just as a note I wouldnt be tuning your motor to anywhere near 14.7, it will not like it and may even do damage under load. I take it that the auto tune adjusts the whole table to a single lambda? this is the target AFR table I started for my car a while ago, it kind of negates the need to switch between a power and economy table

[speedpro]

There is something up with your wideband by the sound of it then, 1V should be somewhere around 11:1 and stoich around 2.5-3 volts depending on the controller. Just as a note I wouldnt be tuning your motor to anywhere near 14.7, it will not like it and may even do damage under load. I take it that the auto tune adjusts the whole table to a single lambda? this is the target AFR table I started for my car a while ago, it kind of negates the need to switch between a power and economy table

You are absolutely right. My controller output is Lambda=2 - output=5v. Lambda=1 - 1.667v, Lambda=.9 - 1.33v, Lambda=.8 - 1.0v, Lambda=.7 - 0.667v, Lambda=.6 - 0.333v. In Eco mode the target lambda is 1.

Like I've suggested above though we can then export the autotune data, divide all values by, or only a certain few, by say .8, and save as the active VE map. OR, tune for Lambda=1 using autotune then again copy all the data to the active manual map. Then switch the Eco/Power switch to Power whereupon extra fuel is calculated on top of what is required for Lambda=1. Extra fuel is added to adjust the Lambda to whatever values are in the Drivers Desired Lambda table - see attached. This feature is open loop and depends on the VE or TPS maps being accurately tuned for Lambda=1 as it simply calculates the fuel to add to that already calculated.

If the Eco/Power switch is set to Power during autotune the target Lambda is whatever is in the Drivers Desired Lambda.

Sorry about the pdf but hopefully you can get the idea.

[2005bully]

http://www.kiwibiker.co.nz/forums/sh...aha-YZ-250-EFI

WOW!! Several months back I thought that I saw something on KB about an EFI YZ250 running methanol. But was unable to find it later. Thanks a lot guy's....This is great!!!!
Kermit Buller

[TZ350]

WOW!! Several months back I thought that I saw something on KB about an EFI YZ250 running methanol. But was unable to find it later.

This YZ250 and the Kawasaki Bighorn are running 100% E85, not methanol. I just checked and the Ecotrons pumps are not supposed to be used with Methanol or Gasoline plus Water.

A list of their components and data sheets can be found here:- http://www.ecotrons.com/support/

[husaberg]

WOW!! Several months back I thought that I saw something on KB about an EFI YZ250 running methanol. But was unable to find it later. Thanks a lot guy's....This is great!!!!
Kermit Buller

Finding stuff on KB
http://www.kiwibiker.co.nz/forums/al...hmentid=314958
subscribing to threads
http://www.kiwibiker.co.nz/forums/al...hmentid=309283
spell check
http://www.kiwibiker.co.nz/forums/al...hmentid=317054
Add attachments via your computer
http://www.kiwibiker.co.nz/forums/al...hmentid=309050
Image via net URL
http://www.kiwibiker.co.nz/forums/al...hmentid=309051
Deleting a post
http://www.kiwibiker.co.nz/forums/al...hmentid=309285

[speedpro]

Now that I have my rolling road working I have been running the bike and using the "autotune" feature. It's taken a bit to get my head around how it works but I'm on top of it now. All the manuals say that with the Eco/Power switch set to 0 or Eco that it should autotune to Lambda=1. However it has been tuning to a lower value meaning richer. It seems to aim for about 13.1:1, on the ALM gauge, rather than 14.7:1 which is Lambda=1. The ECU log file was indicating around Lambda=1. I was checking through the ALM manuals tonight and found one that gave slightly different values of Lambda for 0v and 5v.

My ALM was configured for 0v@Lambda=0.5 and 5v@Lambda=2. In test mode this gave an output of 1.667v at Lambda=1 and 1.33v at Lambda=.9
I reconfigured the ALM for 0v@Lambda=0.7 and 5v@Lambda=1.95. In test mode this gave an output of 1.2v at Lambda=1 and 0.8v at Lambda=.9.

So originally my ALM was configured to output a higher voltage at any particular Lambda. The ECU tuned the engine richer to obtain the required voltage it expected for Lambda=1, possibly 1.2v, which has resulted in my autotune map running the engine a bit rich. 1.2v would be Lambda=.8? which is what I was seeing on the ALM gauge.

I'm not worried about it as I want to run a bit rich with the turbo anyway and the idea is to map the fuel requirements. If I decide to run a bit richer it is simply a matter of multiplying each value by some amount.

What I have been doing with the autotune data, only the VE table in my case, is exporting the data and then importing it to the regular manual tuned VE table. This is one method of retaining your autotune date in the event of some sort of problem with the ECU where your autotune date is lost. If there is a problem and you lose autotune data, if the VE table has been updated, you can reset the autotune data to default which imports the current VE & TPS tables which it then tunes automatically. This is the only way to manually alter the autotune data - resetting to default which imports the current manual tables.

I ran the bike last on John Connor's inertia dyno to check ignition timing requirements by actually measuring power. Found that it likes a bit of advance. The problem was that running a bit of boost and revving through to 16,000rpm it was leaning off quite a bit, nore than I liked that's for sure. This was expected as the autotune had only been run up to 1bar and about 14,000rpm. It had filled in values to the end of the tables but it was a flat plane in the 3D view rather than the expected hump. What I did was export the autotune data, imported the data into a VE table and manually edited a hump into the values out to 1.6bar and 16,000rpm. I can run that VE table in normal mode and/or import the default data to autotune and the updated values will be where the autotune will then start from.

[speedpro]

I've had it back on the rolling road. 1/2 confident of the ignition timing, fuel pressure is good, injector pulse width is topping out at about 5mS or about 70% of the time available.

It was in autotune with the ALM output as it came so the ECU was tuning for what was actually a bit rich, but what it thought was Lambda=1. The engine ran flawless at a few throttle settings under 100% and you could see it hunt over and under Lambda=1 with a diminishing swing each time until it settled more or less on 1. Finally ran it at 100% which because of the setup it switched to the TPS table, set for anything over 98% throttle. The TPS table had what was probably values for a Kawasaki Gpz250 which the kit is from and the 3D chart showed a pretty typical fuel map. I ran it at 100% and at 10, 12, 14, and 16,000rpm. Very interesting effect on the TPS table. Even though the throttle was only ever at 100% when the table was tuned EVERY value in the table was altered. Now the 3D chart is a flat plane sloping down from the 100% values. Supposedly it reached 140% load which is probably about right with the turbo.

What I have decided is to alter the ALM output to the correct values as per the latest version of the ALM manual. In the self test this has the effect of causing 1.2v to be output at a test Lambda of 1. The ECU indicates Lambda=1 as well so now they match. Next time I have it on the rolling road I will set the Eco/Power switch to 1 which will make the autotune software tune to the Drivers Desired Lambda values rather than 1. It should only make a small change as the current autotune table values in the engine are a bit rich already.

Tuning mostly done I spent a bit of time just pissing with the throttle and I can't fool it. One thing I did check was if the waste gate was working but it's only managing about 1.2+bar so no activity. Funnily it manages 1.2bar from about 10,000rpm but doesn't seem to increase with revs, could be a tuning thing.

Once finished I exported the autotune date to a CSV file, cycled the power, checked all was still good and then bought it home, aaaaand, the ECU didn't start the fuel pump when power was turned on. It took 5-6 tries before the pump started. When I checked this time the autotune tables were blank. I've done the process of loading the autotune date into the regular VE and TPS tables and then reset the autotune data to default which imports those tables to the autotunetables so nothing lost, thankfully.

One other thing I noticed was that the throttle had to be opened a small amount before the TPS indicator registered movement. As previously I "flashed" the ECU and now it is back to working properly. Doing this blanked the autotune tables again, which supposedly it shouldn't.

Tonight when I powered it up the fuel pump again did not start on the first try. Hopefully it isn't the ECU on the way out. I have a spare pump from Rob so will swap it out if the problem persists.

If the weather is OK I'm looking to have a ride around on Saturday at Mt Wgtn.

[TZ350]

Certainly nearly driven me crazy learning what to do ......

But tackling new chalanges is the Bucket way.

Attachment 319293Attachment 319294Attachment 319292

EFI 2T, three injectors, Throttle body with TPS and dyno chart with runs at different throttle openings.

I have found the trick is to use split injection for high and low power and the smallest low power injector possible because that gives the widest range of tunability for closed or trailing throttle under 30% TP (throttle position).

The low power runs on the graph show that I have not got the nearly closed throttle fuelling anywhere good enough to be useful on the track yet.

Attachment 319292

100% TP = 27hp
50% TP = 24hp
40% TP = 20hp
30% TP = 12hp
(TP = throttle position)

From 60% to a 100% TP typically (for any engine) there is not a great change in power, 25 - 27hp, a 2 hp change and consequent fuel demand in this case. The tricky fueling problems are all in the area between 0 to 40% TP, which is a 20hp change or ten times the variation in fuel demand as 60 to 100% TP on my EFI engine.

So you need a primary injector that can be turned right down to next to nothing and ramped right up to 50% of the total fuel demand. The secondary big injector, actually the two side injectors fired together as a pair in my case need to be larger but their on time does not vary nearly as much as the single smaller injector needs too.

Hopefully get the new bike finished soon so I can get back onto developing the EFI thing.

So there we have it, a 2T EFI Bucket wannabe.

Now that I have a good F5 bike of my own I am very much looking forward to getting back to this EFI project.

[speedpro]

I have the rolling road now if you want to run it at constant speed and load.

[TZ350]

The two stroke EFI story so far with pictures if you follow the links back to the original posts.

Posted because someone asked where to find it all, so a recap on the basics of the Beasts Engine.

TeeZee tested his 24 against a 30 in a back to back test. The 30 didn't make anymore power than the 24.

Ok, making a bit of progress with setting up the EFI systems throttle body.

Det Sensor. Next week I hope to get all the chopping and changing of the wiring done then I can start playing with setting up the Ecotrons EFI program for a trial run...

The mid chamber bleed re visited as it might be useful for some low end boost......

One of the most useful things I have added to the fuel injection system is a switch that turns the fuel pump off.

Tim gave me a hand tonight, we started to make progress when we moved the injection end point. I think we originally had the injector squirting into closed ports.

Flettner makes the point that it is important, in fact the whole key to successfully fuel injecting a two stroke, is to have the injection period timed to end at the transfer port closing.

First cut at verifying the timing of the injection end point.
Lashed a regular trigger coil to a 12V solenoid with a bolt through it. Set the Ignitec map to zero, now it will fire immediately it sees a trigger pulse.

Spent sometime tonight trying to figure out why there was no signal at the injector. Then realized that at TPS = zero the injection map is = to zero. So bumped the first line up to 100 and bingo the injectors were firing.

The next move is to try the lash up Flettner suggested, and yes I still had sparks and by switching the leads plugged into the injector socket I could easily figure out which was the injector energizing and shut off points. And now I can setup the injector close point and with the timing light I can clearly see where the injection events are happening on the flywheel.

Ok ..... I dusted off my old valve powered signal generator and oscilloscope. And brought some science to the problem, and it looks like it is not me.

Later realised it was me, I did not have the steps in the map as even as they needed to be in real life and when the ECU tried changing between injectors with the engine under load it was tripping over a step and becoming confused.

Connected up a signal generator and oscilloscope to the Beast and had a look at how the EFI injector pulses behaved at a simulated 2,000 to 12,500+ rpm. And then with the motor running. Five things I learned about my EFI setup today.

Ok finally got the O2 and exhaust temperature sensors installed. No idea if they are in the optimum place but they will do for a start. Now that it all seems to be working I am really looking forward to getting a bit of time on the dyno to play with this thing.

I am now starting to make some headway with tuning the Beast and I am beginning to enjoy using the Ecotrons small engine EFI kit for two strokes and its 2T tuning software. Recorded some RPM and TPS data along with the Lambda O2 sensor.

Ok, making progress, that's if you can call getting it wrong progress. Ecotrons has a very clever self tuning feature where you tell the map what Lambda you want at what RPM and the ECU will read the ALM O2 sensor and adjust the mixture for you. Being able to make planned mistakes is a step better than just making ignorant mistakes and not knowing why. So .... its progress.

Posted because I was recently asked about the det sensor I used. I earthed one of the two wires. Probably does not matter which.

The detector/amplifier I got of Ebay comes with lights and I currently use an optional extra output to signal the ignition to retard when there is detonation.

Started making a new injector manifold that points the small Ecotrons injectors across the transfer port windows, previously the outside pair had just fired straight into the B transfer ports and impinged on the back of the cylinder liner.

Making some progress with the new fuel injection manifold. The injectors are angled so they squirt more horizontally across the transfer port. The small (bottom left) injector from the Ecotrons Small Engine EFI kit for two strokes has had the clips on its top cut back so the fuel barb can be swiveled in different directions.

Unlike finicky carburetors, with fuel injection there is not the over enriching problems you get when the inlet stream bounces back and forth several times in the inlet tract. Passing the carbs needle jet several times and picking up fuel on each pass. With EFI I don't have to be so conservative with the rotary valve timing and can experiment a little.

New injector manifold to direct the fuel streams across the transfer ducts. There will be three 126 g/min injectors, with the middle one firing first then the outer ones come in when the engine gets up on the pipe. The center injector is angled up so as to cool the underside of the piston crown by squirting through a slot in the piston skirt.

The plenum re visited ... The problem was, with the carb on the outside we got excessive fuel dropout in the plenum, EFI should cure that.

Started work on the fueling map, things are starting to come together, but probably making adjustments in to big a steps. More patience is required. When I shut the throttle after a 12k pull the motor sounded like a hive of angry hornets, to lean on over run.

The big learn of the night was the truth of the quote below.

"" One thing you have to always remember with alpha-N is that you don't actually know where the effective WOT is any more (i.e., when you have enough throttle that opening it further doesn't affect the amount of air being ingested). At low RPM WOT could be only 20% throttle. ""

Ok this is where my thinking is at tonight, possibly wrong but there you are. So I am none to sure about how useful the BSFC would be as it looks more rpm and engine efficiency than load dependent.

But what interests me is the Delivery Ratio as I think this is the total volume of air that is passing through the motor and as we all know fuel and air needs to be mixed at a set ratio, even if some of it gets wasted out the exhaust.

I knew the VE table was all about cylinder filling and I thought the cell values on an Alpha-N table were all about the volume of fuel to squirt but I have had to re-think that. From the Ecotrons manual. LOAD based Alpha-N system. For two-stroke engines, the default fuel mapping is Alpha-N method.

For my bike, at 12k ish the delivery ratio is about 120% of the cylinder volume and the amount of air (charging ratio) in the cylinder at BDC is only about 80% of the delivered air and less than that actually gets trapped (trapping efficiency) at exhaust port closing, so you can see that a fair swag of air is lost out the exhaust port. And my guess is that the bulk of the air that is lost is the first air entering the cylinder as the transfers start to open and if there was extra fuel already in the transfers that would be lost with it.

Now the opening time can start before the transfer opens and the close time after so that leaves me a clear 1.5 to 1.7ms of full on injection time. The start point can be early but it looks as if the timing of the injection end point is going to become critical if we are going to get the bulk of the injection happening towards the end of the transfer event.

That is pretty much what I have been trying to achieve with the upwards directed injection stream from the central injector.

If a carburettor has one advantage over fuel injection, this is it. When the quantity of fuel is controlled via the open/shut time of an injector, it is practically impossible to create a homogeneous mixture at the injector. And I'm not even talking about vaporized fuel here; I'm just talking about every part of air getting the same number of fuel droplets. Fortunately for us much of the necessary homogenisation takes place in a two-stroke's crankcase, with the con rod stirring things up and transferring its heat to the yet-unvaporized fuel droplets. But with direct fuel injection you lose this advantage, and the con rod bearing looses most of its cooling.

Hi Speedpro, I agree with everything you say but "weak mixture - no", well not in this case because this engine runs well with a carb but not with transfer port injection EFI so the problem must be with the EFI or at least with the way I am trying to use it.

Remember with a carb all inducted air has fuel mixed with it but with EFI this is not necessarily true. The problem with EFI and transfer port injection is that as the rpm gets up, a lot of air can get through the port but there is not necessarily enough time to inject sufficient fuel into the air stream.

Making progress, touching 12,000 rpm and nearly 25hp 3 to go.

26 hp and repeatable, but so fiddly to get and still detos like mad, however much fuel I throw at it. All the extra fuel does is make more exhaust smoke.
I think there really is just not enough time, 1.8ms for the injection cycle through the transfers at 12krpm. Ok time to change over to injecting down the inlet tract.

Ok, rearranged the injectors, injector one is still in the rear of the cylinder and injector two is firing down the carb. With the injection cycle timed to end at inlet closing. Red line is both injectors in the rear of the cylinder. Blue line is with the second injector in the inlet tract. The lines diverge at the point the inlet injector takes over. Its interesting because it goes to show that carbs can be pretty good.

The Blue line is my best with transfer port injection the Green line is where I have got to today with the twin injectors aimed down the inlet throat. It was the same map as the transfer port injection, and better than yesterday's single injector effort.

Well tonight's effort was a total bust, the two injectors send out such a cloud that it wets out on the bellmouth and dribbles down into the plenum which totally negates the reason for having injectors in the first place. Ie to stop fuel pooling in the plenum.

I can see the advantage of having the injectors behind the throttle slide like they have on the BRC engine. Gave up for the night after the EFI filled the crankcase with raw fuel.

Alpha-N maps need to be smooth without any peaks or severe troughs between cell steps. Ecotrons has a function where you can export a map to an Excel spread sheet and use the charting function to help smooth the map data. Raw data and the chart showing how rough the data is and the engine ran just as roughly. Smoothed data and chart.

Making progress towards the magic 28, at least now I know this EFI thing is going to work. Just a lot of careful step by step adjustment of the maps is required. I expect time and patience will see it running very nicely. I also thought it had a terminal deto problem. Deto turned out to be the piston hitting the head.

27.6hp - getting close with EFI to the best ever achieved with a carb on this cylinder. Blue line is where we started, Red line is where we are now. The big issue is getting back on the gas and running to rich below 8,000 rpm. Anyway it is coming right bit by bit.

The EFI is touching 28, but its still very reluctant to come back on the throttle after rolling off.

It will be a pig to ride so I guess I will take my laptop to Tammers and the weekend for me will be all about starting off the back of the grid and adjusting the EFI map to try and get the Beast track ride-able.

I am looking forward to it as I might have more luck getting it right by riding it than trying to tune its low to mid range on the dyno. Anyway the weekend away camping with every one else will be fun and the local RSA does a real great Saturday night dinner ..... Yum ...

Tammers was a bust, Bike was a real pig to ride.

Ok after months in the wilderness I finally figured it out. I was right about needing smaller injectors. So it looks like, to maximize tune-ability you need to minimize the injector size......... in hindsight its so obvious really .... Not perfect yet but very very promising.

With a smaller primary injector things start to come right.

Injectors turned up from Ecotrons today, four days from date of order and two of them were weekend days, pretty good service in my books. I fitted a smaller injector and did a bunch of runs at different throttle settings. Interesting that the 60, 50 and 40% runs carry on out to close to 13,000 rpm. Blue line is the TPS. 100-90-80-70-60-50-40% throttle settings. It is certainly throttling and coming back onto the power much better. I will have to analyse the data to see if I can use an even smaller injector.

The EFI lesson I learnt last month was, that to get the greatest range of tune ability I needed to use the smallest injectors possible that will still do the job. I know, pretty obvious, but there you go.

I read the Ecotrons EFI manual and saw that the map must have as many even steps as possible and I thought that meant dividing the rpm range up as evenly as possible, seemed logical, but wrong again.

The bike bucked around swapping between the staged injectors as the EFI struggled to jump from one cell to another on the steeper parts of the fueling curve.

I guess what they really meant was that the changes between cells need to be as even as possible.

So I don't need even spacing between the rpm break points and can have big or small gaps between them so long as there are no big steps between the values of adjacent cells.

I can now throttle on/off pretty well and do a series of part throttle runs, the last one is at 30% TPS, it looks a bit lean.

Its not perfect but is running a lot better, so for what its worth here is the current Alpha-N map.

Carburetors are great because of their analog ability to change the fuel delivery with changes in air flow at the same throttle setting. EFI, well EFI is not forgiving at all.

The Ecotrons Alpha-N map has 16 rows to work with and I had concentrated them in the area where the power is, as you do. Well that turns out to be a mistake because there is not much difference in fuel demand per cycle between 100% throttle and 90 or 80 or even 70 at max power and virtual no difference at all above 40% in the 3 to 6000 rpm range.

The big jumps were between the lower throttle settings where there was not much excitement happening, well the mistake is that the CPU has a hard time bridging the big gaps below 40% especially when it is also trying to swap between low and high injectors at the same time.

As it turns out, it was a mistake to concentrate my map in the power area, but I guess I will forgive myself.....

29hp

Different problem, but I feel your pain Dave. Been working my arse off on this EFI thing. 90% throttle consistently makes more power than a 100%, 80's pretty close to and 70 is not far behind.

Attachment 309768

I keep making more power but can't get it to come back onto the throttle properly.

I just want it to throttle properly so I can ride it , Bah humbug, and stamp my feet.

Finally getting some place with the EFI after strapping a small laptop onto the tank and riding the bike around looking for the bad spots in the map. I can now wind it out, throttle off and have it come back onto the throttle again. Its now ride-able enough to make it worth taking to a practice day for further refinement. Dyno runs recorded at 100-90-80-70-60-50-40 and 30% throttle.

I have made more power in the past but the point of these dyno runs is that it is running reasonably well at different throttle openings and it is also something of a guide for refining the Alpha-N map some more.

Its a two stage injector system with three injectors, the middle 60 g/min unit is for starting, slow running and over run after shutting off. A pair of 80 g/min units provide fuel as the engine comes on the pipe then the middle injector chimes in again so all three are firing as the rev's get around 10,000 rpm.

I checked the data logger and the middle injector is chiming back in at 9,500 rpm on WOT so I have achieved one of may aims, squirting raw fuel at the underside of the piston crown to cool it when the engine is making real power, I am very happy with that.

The biggest issue was selecting the right sized injectors, I had chosen ones that were to big, to big and there is not enough range of tuning adjustment and it took a while to work my way down to quite small injectors.

The quest for even steps on the Alpha-N map.

Now that my new bike is nearly ready I need to develop a new Alpha-N fuel injection map. The axis's of an Alpha-N map is throttle position TPS and RPM. The steps between cells need to be in even steps. That is not even RPM or TPS steps but even changes in engine load.

"Torque" is largely related to the motors Trapping Efficiency of the air passing through it.

"Delivery Ratio" is everything when it comes to quantifying airflow through a two stroke motor and it is the air flow that needs fuel added to it in the correct proportions. Trapping Efficiency is the measure of how much of this air/fuel is trapped to be com-busted.

Initially I thought changes in Torque could be used to represent changes in engine load and could be used for developing a fuel injection map. Well that might be Ok for a four stroke but now that I have tried that approach I now think that Delivery Ratio is the more suitable measure for developing a two stroke map.

Air flow (Delivery Ratio) through a throttle is not linear. The greatest changes happen between 20 - 40% throttle opening.

Attachment 320573

The first requirement is to find the throttle openings that corresponds to even Delivery Ratio (air flow) steps.

Attachment 320572

Then I modeled these throttle openings in EngMod2T and got a reasonable correlation and reasonably even layer of Transfer Efficiency steps in the power region of the simulation.

This information at each TPS position and RPM point should make a handy start for developing the new map.

The rest of the simulation data (all 12 of them) are posted on the Ecotrons thread. Because the Transfer Efficiency graphs are probably similar to any performance 2T with a few adjustments to the rpm points the numbers may make a useful start to any 2T Alpha-N map project.



Now to map it all out.



Attachment 320596

Attachment 320597

Job done, now to see if it actually works.

[TZ350]

Page 1 ....

There is very little information that I can find on the net about what a two stroke electronic fuel injection map should look like or how to develop one.



Once I started to understand how EFI worked it was relativly easy to get full power runs but it was quite hard to develop a map that is good for part throttle and transition from full throttle to closed and back to open again. This has been the biggest issue holding me back from having a track ready EFI bike.

I was stuck for a long time until I realised I need to use the smallest injector possible to give myself the widest tuning range. So the lesson here is; use the smallest injector possible that you can get away with, not the largest.



Once I realised that I was able to develop a map that more or less works down to 30% throttle but under that it is still rat shit.

So I figured it is time to bring some science to the party. And found this air flow vis throttle position curve. Increases in airflow do not follow throttle position. The biggest changes in airflow are between 20-40% throttle.



Airflow on the left and throttle position along the bottom. Airflow vis TPS summarized on the right.



I took the throttle numbers from the airflow vis throttle position curve and ran them through the EngMod2T model for my air cooled engine.

Throttle position numbers used:- 3 - 5 - 7 - 13 - 20 - 27 - 33 - 40 - 47 - 52 - 57 - 61 - 67 - 78 - 87 - 100%

The objective was to get a series of evenly spaced "Transfer Ratio" curves. Transfer ratio represents the amount of air transferred to the cylinder and this amount of air needs fuel added to it in the correct amount.

I originally tried to work with simulated torque curves but torque is a product of transfer ratio and trapping efficiency. The torque curves did not take account of air that was transferred but not trapped. So basically with crankcase or transfer port injection torque underestimates the amount of fuel needed for correct mixture strength of the trapped air as some air/fuel is unavoidably lost but needs accounting for. Direct cylinder injection would be different and cleaner but beyond what I can manage engineering wise at the moment.



At 4,000 rpm the transfer efficiency at 40 and 61% throttle is greater than at 100%, this needs to be reflected in the map.

[TZ350]

Page 2 ....



I found that I needed a value of 65 in the EFI map cell that represented full throttle at 9,000rpm and slightly less at 12,000 rpm. This 65 is just one cell number and depends on my injector size, and the amount of air ingested at that point. But from this known starting point and the EngMod simulations we can develop a map.

Because pretty much all performance two strokes will have a curve with a steep transition that looks something like mine. I am posting all the simulations with their values so anyone else can use them as a rough guide. All they have to do is find their starting point and adjust my RPM points to suit their power curve.

All I have to do now to develop my map is some maths using my "65" starting point and the simulated delivery ratio numbers to find values for each cell on the map. The Ecotrons Alpha-N map is 16 by 12, 16 TPS values and 12 RPM points. Other maps may be different.

3,500 rpm

4,000 rpm

[TZ350]

Page 3 ....

5,000rpm

6,500rpm

7,000rpm