Page 16 of 25 FirstFirst ... 61415161718 ... LastLast
Results 226 to 240 of 362

Thread: Speeduino 2T EFI Project

  1. #226
    Join Date
    13th February 2016 - 09:21
    Bike
    None
    Location
    Holland
    Posts
    44
    Quote Originally Posted by TZ350 View Post

    Attachment 341951

    This is an actual scope reading at about 11,500 rpm of a MAP sensor reading the crankcase pressure of my GP/NSR110. The yellow spikes are the ignition trigger pulses.

    The scope shows a cycle time of nearly 6ms, just over 10000 rpm. It looks like the scope's calculation slipped over the chopped blocks of the sensor.

    With some fantasy you can see a pseudo cycle of 5 ms (and the disadvantages of the slow sample cycle).

    You cant choose when it takes a sample, so you could make a voting system over a longer period. Or walk away from automotive parts in general.

  2. #227
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    Quote Originally Posted by Carel H View Post
    The scope shows a cycle time of nearly 6ms, just over 10000 rpm.
    The trigger was on the yellow channel and set to 4,75V rising edge of the initial trigger pulse and I was looking at the f=193 Hz displayed top right. But, yes you are right, time scale is 2ms a division so 6ms a cycle or 167 Hz. ?????? Why would the scope show two different frequencies??? more shit I don't understand ......

    Quote Originally Posted by Carel H View Post
    You cant choose when it takes a sample, so you could make a voting system over a longer period.
    At the moment my pseudo MAP system takes the best high and low reading of each cycle and rolling averages them over ten cycles. I would welcome ideas around how I could do this better.

  3. #228
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    .
    Quote Originally Posted by Carel H View Post
    .... walk away from automotive parts in general.
    I was hoping not to have to and at the price of a KTM spare sensor I don't think KTM are using anything too fancy themselves.

    Automotive sensors can be picked up cheap on the Net.

    The Speeduino EFI CPU's firmware code is free and open source so you can modify it to suit your needs.... https://speeduino.com/wiki/index.php..._-_Easy_Method

    TunerStudio for adjusting the Speeduino's mapping, the basic version is free.... http://www.tunerstudio.com/index.php/tuner-studio

    The Speeduino Shield DIY naked board for the Arduino Mega is cheap and they can also be brought fully made up.... https://speeduino.com/forum/app.php/page/buy

    The Arduino Mega itself is relativly inexpensive.... https://www.amazon.com/ARDUINO-MEGA-.../dp/B0046AMGW0

    The Arduino Nano is cheap .... https://www.amazon.com/OSOYOO-Arduin.../dp/B00UACD13Q

    So Inexpensive, Cheap or Free. As a DIY project I hope others might try, I am keen to keep this affordable.

  4. #229
    Join Date
    13th February 2016 - 09:21
    Bike
    None
    Location
    Holland
    Posts
    44
    Quote Originally Posted by TZ350 View Post
    The trigger was on the yellow channel and set to 4,75V rising edge of the initial trigger pulse and I was looking at the f=193 Hz displayed top right. But, yes you are right, time scale is 2ms a division so 6ms a cycle or 167 Hz. ?????? Why would the scope show two different frequencies??? more shit I don't understand ......



    At the moment my pseudo MAP system takes the best high and low reading of each cycle and rolling averages them over ten cycles. I would welcome ideas around how I could do this better.
    I saw on the bottom left screen "CH1" and I assumed the blue line was CH1.

    Doesn't matter, for it applies to both: a scope will very happily give you a nice frequency reading on clean wave signals, but with complex signals the going gets tough.

    Looking at the signal, I would synchronize with the ingnition signal, take a few samples during 1.5 - 2 ms, take the lowest and put it in the moving average filter.

  5. #230
    Join Date
    13th February 2016 - 09:21
    Bike
    None
    Location
    Holland
    Posts
    44
    Quote Originally Posted by TZ350 View Post
    .


    I was hoping not to have to and at the price of a KTM spare sensor I don't think KTM are using anything too fancy themselves.

    Automotive sensors can be picked up cheap on the Net.

    The Speeduino EFI CPU's firmware code is free and open source so you can modify it to suit your needs.... https://speeduino.com/wiki/index.php..._-_Easy_Method

    TunerStudio for adjusting the Speeduino's mapping, the basic version is free.... http://www.tunerstudio.com/index.php/tuner-studio

    The Speeduino Shield DIY naked board for the Arduino Mega is cheap and they can also be brought fully made up.... https://speeduino.com/forum/app.php/page/buy

    The Arduino Mega itself is relativly inexpensive.... https://www.amazon.com/ARDUINO-MEGA-.../dp/B0046AMGW0

    The Arduino Nano is cheap .... https://www.amazon.com/OSOYOO-Arduin.../dp/B00UACD13Q

    So Inexpensive, Cheap or Free. As a DIY project I hope others might try, I am keen to keep this affordable.

    The sensor you use now is meant for a multi cilinder 4 stroke engine and will give for that purpose a nice signal. It is also calibrated and will therefore give no tuning demands on the assembly line or replacement.

    No need for fast peak to peak measurements. The signal you want is on the chip, not on the thermal compensated and calibrated output.

    That's automotive.

    As for cheap and fast, not calibrated, not compensated (check the box "Freescale"):

    https://www.reichelt.nl/Sensors-SMD-...97&LANGUAGE=EN

    I had a quick look at the Speeduino and stopped at the marketing talk of Tunerstudio. Same as the KTM site: hardly at specification level, but all about the experience.

    Also intended for multicilinder four strokes, which makes parametric tuning harder and will eventually make it necessary to change the software.

    For your purpose an Arduino plus two Mosfet injector drivers would be a nice (cheap) start.

    I looked at EFI systems before. I find it a narrow bridge between programmers and mechanics. The programmers don't understand an engine, the mechanics don't understand electronics and proud of it on both sides.

    Resulting in these stacked rough database contraptions, far removed from engine management.

    Preferably in a heavily finned, colour anodised, billet machined housing.

    End of rant.

  6. #231
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    Quote Originally Posted by Carel H View Post
    For your purpose an Arduino plus two Mosfet injector drivers would be a nice (cheap) start.
    That could easily be the future.

    Quote Originally Posted by Carel H View Post
    As for cheap and fast, not calibrated, not compensated (check the box "Freescale"):

    https://www.reichelt.nl/Sensors-SMD-...97&LANGUAGE=EN
    I am grateful for your input. This sensor looks like the unit that came on my Speeduino board.

    Why would it have so many wires? Do you think it would be worth getting a spare and trying it?

    Click image for larger version. 

Name:	Screenshot (157).png 
Views:	50 
Size:	501.1 KB 
ID:	341966

  7. #232
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    .
    Quote Originally Posted by Carel H View Post
    Looking at the signal, I would synchronize with the ignition signal, take a few samples during 1.5 - 2 ms, take the lowest and put it in the moving average filter.
    Pretty much what I am doing:-

    The ignition signal initiates an interrupt.
    From that the time for a complete cycle is determined.
    Then the time in micro seconds are calculated to the six sample points. Three low around TDC and three high around BDC.
    At the appropriate time the analog reads are done.
    The best of the lowest and highest are chosen.
    Then they go into their respective FILO 10 cell liner arrays and then the 10 current entry's are averaged.
    The rolling average values are used in the Delta pressure calculation. Ie; High - Low = Delta
    The Delta is rolling averaged too then multiplied by 3 to give a pseudo MAP value.
    The pseudo MAP value is continuously output and updated each cycle.

    Click image for larger version. 

Name:	Air Cooled Dyno Runs.jpg 
Views:	40 
Size:	658.2 KB 
ID:	341971Air Cooled and Fuel Injected Suzuki GP125

    A series of EFI dyno runs taken at different throttle positions on my old air cooled 125. I had spent a lot of time mapping the Alpha-N fuel injection table. Alpha-N only looks at TPS and RPM and for each cell on the table you map an engine "Load" value that is used by the CPU to determine how much fuel to inject at that particular point, every time, nothing changes. This worked well for anything above 45% throttle position but it is not a flexible fueling system. More suited to drag racing than road racing.

    I got the Alpha-N mapping fairly good and even rode the bike like that a few times and towards the end, sort of successfully but throttle control coming out of corners was not so sweet.

    The current plan is to use Alpha-N above 45% TP and VE, volumetric efficiency below 45%. VE looks at engine volumetric efficiency vis manifold absolute pressure (MAP). That way fueling can be variable in response to the airflow changes induced by the pipe.

    This way I can have the best of both worlds. Alpha-N which has proved reliable and very tune-able when the motor is on song and responsive variable fueling at lower throttle openings.

    Combining the two mapping typologies, Alpha-N and VE should be relativly easy with the Arduino Nano pseudo MAP system. Below 45% I make the output variable and above I make it fixed at atmospheric.

  8. #233
    Join Date
    13th February 2016 - 09:21
    Bike
    None
    Location
    Holland
    Posts
    44
    Quote Originally Posted by TZ350 View Post
    That could easily be the future.



    I am grateful for your input. This sensor looks like the unit that came on my Speeduino board.

    Why would it have so many wires? Do you think it would be worth getting a spare and trying it?

    Click image for larger version. 

Name:	Screenshot (157).png 
Views:	50 
Size:	501.1 KB 
ID:	341966
    It's the "DeLuxe" version with build-in amplifier to make life easy, three pins are No Connect:


    https://www.nxp.com/docs/en/data-sheet/MPX4250A.pdf

    Globally speaking, you need a differential one, because you want to measure vacuum and pressure, and specify a pressure range.
    But for experiments you can start with the one you have and go from there. If your board is assembled, all you need to do is connect a tube and the scope.

    To add: the datasheet specifies a 10-90% swing in 1ms. Meaning a 30-70% swing in 0.5ms. So a higher pressure sensor is faster with a lower resolution. So one step at a time.

    Did some further research, because the inbuild amplifier inhibits positive/negative readings, just to answer my own question. I did pos/neg before and now it was gone.

    On this page you will see sensors with -x/x specifications:

    https://www.nxp.com/products/sensors...M_-50-115KPA#/

    Select one at the top:

    https://www.nxp.com/products/sensors...sor:MPXV7007DP

    Go to Buy and do a sample request over several pressure ranges and as much as they will hand out per pressure spec. Works normally well and it's as cheap as you can get, ie free. The law of permanence of problems here is that zero pressure is at 2.5 volt. So while you're busy, request also a couple of vacuum sensors.


    I think they will do it, considering it is for New Zealand.

  9. #234
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    .
    Successfully fuel injecting a single cylinder high performance two stroke, i.e., something making more than 10 bar BMEP and greater the 10,000 RPM has been very difficult until now. Because it was impossible to apply four stroke air flow measurement techniques like MAF, manifold mass air flow and MAP, manifold absolute pressure to a two stroke.

    A two strokes inlet manifold and crankcase have entirely different pressure profiles to a four stroke. Whereas a four strokes manifold pressure decreases when the throttle is closed and is at its highest when at WOT, wide open throttle with a range of 0.2 bar to 1 bar on a normally aspirated engine. A two stroke is the other way around. Maximum manifold and average crankcase pressure at minimum throttle and reduced pressure at WOT, wide open throttle and the range is 1 bar at no throttle to about 0.7 bar at WOT, wide open throttle.

    The reason for the two strokes pressure profile is because at WOT and in the power band, the pipe is sucking hard on the crankcase and so reducing the pressure there. At closed throttle there is little or no pipe suction and atmospheric pressure floods back down the stinger and up the expansion chamber flooding the crankcase to atmospheric pressure with exhaust smog.

    A two stroke can idle with a crankcase full of smog because a small amount of fresh mixture randomly finds its way through this soup to the cylinder. At idle a two strokes cylinder is filled with random proportions of smog and fresh mixture each cycle. So you have to run a bit rich here to give a fuel molecule a better chance of finding a O2 to combine with.

    Click image for larger version. 

Name:	Air Cooled Dyno Runs.jpg 
Views:	32 
Size:	658.2 KB 
ID:	342089

    It is very easy to setup a fuel injected two stroke for maximum power on WOT wide open throttle as air flow there is consistent and the Alpha-N fueling topology works well there. Alpha-N is TPS throttle position vis RPM and relies on consistent airflow.

    But accurate fueling becomes increasingly difficult at lower throttle settings because the suction action of the pipe becomes increasingly variable and changes in air flow less consistent or predictable at lower throttle settings. At lower throttle settings the VE, volumetric efficiency topology is more appropriate. VE, volumetric efficiency is MAP, manifold absolute pressure vis RPM and is suited to variable airflow.

    In a two stroke variations in air flow through through the motor can be seen by watching the Delta crankcase pressure. The difference between the lowest crankcase pressure near TDC and the highest near BDC. This delta is a reliable indication of changes in air flow. The greater the delta the greater the air flow.

    The theoretically correct crank angle for the highest pressure is around 160 ATD and lowest about 20 BTDC but the quality and response of your pressure sensor may change things.

    Click image for larger version. 

Name:	11580 RPM Delta P Trace.jpg 
Views:	33 
Size:	711.4 KB 
ID:	342090

    The first step is to see how your sensor responds. Maybe run the engine up with a carburetor and view the sensors output and timing on a scope.

    Click image for larger version. 

Name:	AN vis VE.png 
Views:	62 
Size:	345.6 KB 
ID:	342088

    This is how I would setup the EFI fueling using the delta crankcase pressure concept.

    The Red area below 45% TPS and in the RPM area where the pipe is working I would setup the fueling for VE volumetric efficiency. MAP vis RPM.

    Every where else in the Green area I would setup fueling as Alpha-N. TPS vis RPM. Airflow is consistently predictable here.

    And in the Purple area below the RPM where the pipe works and TPS = 0% the MAP value needs to be set at 0.2 - 0.4 bar. This is necessary because when the pipe is not sucking strongly the crankcase pressure rises to someplace close to 0.8 - 0.9 bar of dirty air. This happens because the pipe is not drawing airflow through the motor and atmospheric pressure and stale exhaust gasses flow back into the crankcase raising the pressure there and in the inlet tract.

    The motor can idle with a crankcase full of dirty air because some fresh air/fuel makes its way through the smog and randomly finds its way into the cylinder. At idle the cylinder has some fresh air and a lot of smog.

    Click image for larger version. 

Name:	PseudoMAP.pdf 
Views:	36 
Size:	136.5 KB 
ID:	342091

    A copy of my PesudoMAP software in PDF format for an Arduino Nano.

    Basically this determines the Delta between high and low crankcase pressures and outputs it as a pseudo MAP value that can be used by any EFI system that can use MAP as an indication of airflow.

    With their latest 150cc TPI offering KTM have shown that high performance two stroke EFI is possible. They use a crankcase pressure sensor but aren't about to reveal how seeing changes in air flow is done. But with the Delta crankcase pressure reading concept I have revealed the secret of how to see the all important changes in airflow through a two stroke motor.



  10. #235
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    Click image for larger version. 

Name:	3 x delta P = 4S MAP.jpg 
Views:	41 
Size:	429.5 KB 
ID:	342099

    Click image for larger version. 

Name:	PseudoMAP.pdf 
Views:	31 
Size:	136.5 KB 
ID:	342100

    This is the software I developed for finding the difference between High and Low crankcase pressure using an Arduino Nano.

    1) It has one interrupt triggered by the ignition pickup.
    2) From the time for the last cycle it works out the time in micro seconds to each analog read point. Six in total.
    3) It takes three analog high pressure readings and three analog low pressure readings.
    4) It selects the best of each three because the exact crank angle moves with rpm so you have to straddle the optimum area.
    5) The best of each gets added to a moving average.
    6) The Delta difference is averaged again.
    7) The Delta is multiplied to get its magnitude to look something like a real MAP value.
    8) The result is continuously output via a DAC card as a 0-5V MAP value until the end of the next cycle.

    Click image for larger version. 

Name:	Arduino Nano and MCP4725 DAC.jpg 
Views:	42 
Size:	779.8 KB 
ID:	342102 The Speeduino EFI CPU is in the box. The Arduino Nano and MCP4725 DAC, MAP co-processor are shown beside it.

    Currently the software looks at the TPS and if its over 45% then the output is standard atmospheric as read by the map sensor at startup for Alpha-N fuel mapping.
    If it is less than 45% then the Delta pressure is used for the MAP value and VE fuel mapping.
    And if the TPS is 0% then the MAP value is forced to 0.2 bar otherwise the Delta pressure would be to high due to exhaust gasses flooding the crankcase.

    Click image for larger version. 

Name:	AN vis VE.png 
Views:	38 
Size:	345.6 KB 
ID:	342101

    Clearly the high vs low pressure comparison works, but I can't get my head around only using 3 interrupts for taking the readings? For my engine, which uses crankcase reed valve induction and using the same cylinders as yours. The transfer port timing is symmetrical regardless of the revs, opening at 118 degrees ATDC (62 BBDC) and closing at 242 ATDC (62 ABDC or 118 BTDC). So shouldn't there be more interrupts around the opening/closing times?
    The low crankcase pressure is just before TDC where the piston is still moving up and the incoming air has not had time to completely fill the crankcase. The high pressure is just before BDC when the piston catches up with air that has not yet been sucked up the transfers by the pipe. Gas inertia is everything here, port timing has only a small part in it.

    The quality and speed of the pressure (MAP) sensor changes the timing of things so best to scope the sensors output at different RPM and loads then adjust the software's timing to suit. The three readings are so you can straddle the optimum crank angles for the best pressure readings.

  11. #236
    Join Date
    3rd May 2017 - 04:03
    Bike
    1997 Yamaha rd 350
    Location
    Russia
    Posts
    171

    Student 8 cylinder ECU

    May be interesting - uni project based on high speed ARM processor:
    8 cylinder ECU with parts for 64USD
    https://skemman.is/bitstream/1946/29...laRikardur.pdf

  12. #237
    Join Date
    24th May 2018 - 00:52
    Bike
    2010 YZ250
    Location
    Auckland
    Posts
    4

    Misfire Detector

    Saw this posted on thumpertalk.

    Could this sensor be used to detect miss-fire and therefore know the engine is not on the pipe and hence dial back the fuel?

    https://www.cycleworld.com/2015/07/2...ng-ionization/

  13. #238
    Join Date
    1st June 2011 - 14:39
    Bike
    Honda NC50
    Location
    Straya
    Posts
    145
    I’ve had a lot of success with alpha-n on my YZ125... it runs much better than a carb at all throttle positions, especially low throttle and low/mid revs, and revs out to over 13,000. Makes 40 at the wheel so it’s in a moderate state of tune, not too high or too low. ItÂ’s one of the best bikes I’ve ever ridden off-road.

    To be honest it’s very confusing because everything you’re saying makes perfect sense about using VE, and before I started building it a didn’t think alpha-n would work properly on a 2 Stroke however this bike has responded extremely well to TPS/RPM only since the very beginning. It’s been running for almost 3 years now and I’ve come full circle back to a throttle body with 2 injectors in it. Previously tried injectors in the cylinder and cases but TBI works fine with much less hassle.

    IÂ’m still unsure as to why it’s worked so well though. Maybe there’s something about the engine/pipe/riding style (off-road) thatÂ’s more forgiving for EFI?? Maybe temps are more stable and therefore have less impact?? I’ve got no idea!


    As an aside I’ve been reflashing the KTM TPI ECUs... figured out a way to access and modify the hex data using a BDM interface. Took about 200 hours on the dyno to figure out where the maps were and how to change them. Getting huge gains in performance though, especially at part throttle. Based on the maps I’ve identified I’m pretty sure the crank pressure sensor is only referenced up to 4,000rpm and less than about 20% throttle. Could be wrong though as I’m still figuring it all out and it’s not at all straightforward. The stock KTM ignition maps are ridiculous and are the cause of most of the spluttering that the TPIs suffer from at part throttle.

  14. #239
    Join Date
    18th May 2007 - 20:23
    Bike
    RG50 and 76 Suzuki GP125 Buckets
    Location
    Auckland
    Posts
    10,473
    .
    That is very interesting, very encouraging, thanks.

    If you would like to post more about your project and your adventures with two stroke EFI, that would be great.

  15. #240
    Join Date
    3rd May 2017 - 04:03
    Bike
    1997 Yamaha rd 350
    Location
    Russia
    Posts
    171

    Crazy Idea

    This is automatically tuning system FAI EFI for 4-stroke.
    https://www.ebay.com/itm/FAI-motorcy...-/273178456487
    Seems should be simple and elegant way use it on 2-strokes?

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •