|
TL, DR: I'm building an EFI from scratch to install on a 1976 GL-1000. I bought a 1976 Honda GL-1000 about a year ago. The Previous Owner was mechanically inclined, but not so much electrically. The carburetors on this motorcycle are notoriously difficult to set up, so at some point in the last decade he got rid of them and put in a single Weber (knockoff) Progressive carburetor. The throttle linkage is completely messed up, the geometry is crazy, and it takes an awful lot of twist to get the second barrel to open up. Also, it's drat near impossible to jet the thing right. Choke? Solenoid wires have been cut flush and then the choke area inside the carb was filled with silicone. In short: This carb has never worked right, and it's time to go. EFI needs several things: Computer, Throttle Bodies, fuel pump, Sensors, Timing/igntion. For tthe Computer, I have decided to use Speeduino, and specifically the "No-overhang two-channel" or NO2C. This is a daughter board for (in my case) an Arduino Mega 2560 which plugs into the top in standard shield fashion and buffers all the I/O. I'm buying a box of pieces from the manufacturer. For throttle bodies, cursedshitbox was kind enough to throw some junk my way, and I've got two racks of throttle bodies, one from an '03 R6, and one from an '06 R6. I'm not sure which to use. They both have MAP and TPS built in. The '06 is drive-by-wire and has two TPS and the motor that actually drives the butterflies. I can probably make all of this work and not have to worry about IAC separately and get cruise control for free. The '03 is just standard throttle body stuff, and should be fine. If anyone wants to know why I think I can use a pair of 600CC throttle bodies to feed a 4-cyl 1000CC engine, I'll make a post for that. Fuel pump: no clue. The standard goldwing pump runs off the right side and is 1-2psi. I'm hoping to be able to put a mechanical fuel pump in its place, but I'm worried about the strength of the drive components coming out of the head. Sensors: You need three sensors minimum to run EFI: TPS, crank, MAP. TPS and MAP are taken care of on the throttle bodies. See below for crank sensor. I've got a spot for the coolant sensor and an intake air temp sensor. Battery voltage is free on this board, and I've got a current shunt coming so I can get alternator/battery current, too. I'm undecided about whether or not I'm going to run O2 sensors. I've got a wideband logger on loan; I'll see if some cheap wideband sensors show up in my mailbox for no reason, or pull some out of whatever car comes with them stock. Timing/ignition: I'm going to be using the "distributor" output. It hangs off the left side of the head, in the spot corresponding to the fuel pump. There's currently a dyna-s electronic unit there, but that uses the stock mechanical flyweight spark advancer. I'm going to get/cut/make a custom gear and put a missing-tooth style wheel in that spot and use the dyna-s to generate a (probably) 30-1 crank signal. Once I have the timing signal, the ECU will output spark signal. I'm currently using a Dodge Neon coil, as the dyna will drive it fine, and it uses wasted spark. I'm probably going to use a Miata coil in the final version as it is logic-level and deals with dwell and all that nonsense for me. Connecting all this mess together is tricky. For wiring, I'm gonna go down to the pick-a-part and just start cutting wire ends off of everything, since sourcing connector part numbers is a pain in the rear end. Fuel plumbing is a no-brainer. The only really hard part is going to be making an airbox and manifold. I've got a 3d printer that's going to be heavily leveraged in this project. Once all the models are finalized and running in plastic, I have access to an industrial-quality TIG machine which I will use to turn aluminum tube into scrap.
|
#
?
Apr 1, 2018 16:16
|
|
|
|
| # ? Apr 27, 2024 22:40 |
|
|
How to tune from scratch. MAP only. Alpha/n is for losers and MAF is expensive and annoying. They all basically use the same stuff, though. This sucks really bad, but must be done sometimes. The zeroth step is to input all your engine settings. Cylinders, displacement, injector characteristics (number, flow, firing order), number of squirts per cycle. This lets the ECU generate a "required fuel" number which is basically how long your injector stays open. You can use this number to sanity check your choice of injectors. Then you set number of sparks per cycle, TPS sensor type, oxygen sensor type, MAP type, temp sensor types and calibration values, that kind of crap. Then it's on to base tables. The first thing you want to generate is a VE table. VE stands for "volumetric efficiency" and represents how good your engine is pumping air. At max horsepower, this value is given to be 100. Basically, the ECU takes the max amount of fuel your engine can burn at max power and scales that value to every other possible RPM and intake pressure. You generate the table by finding someone else with your exact make, model, and modification motorcycle and copy their table exactly. If you are (for some stupid reason) the only human on the planet to ever try this idea, you build a table from scratch. So I'm building a table from scratch. To do this, I found a dyno run of a stock GL1000 motor. I then said "the top of the HP curve is 100" and then scaled stuff off of there. If the engine has peaks and valleys, you want to set your RPM points closer there, as the ECU interpolates between values. Fortunately, Honda built this motor like a tractor and its torque curve is basically 100% flat, so the RPM curve is just linearly increasing until torque falls off as the motor runs out of air. OK Great, VE table done. Now to generate the spark table. I just mimicked the stock mechanical advance unit, then threw in some vacuum advance. Mechancial linear rpm-based advance is fine and very conservative, but you lose a lot of MPG without vacuum advance. Next is the AFR table. This is what lets the O2 sensor work its magic. The stock table from everywhere is fine for now. In short, you want cruise to be lean of peak, idle to cruise to be rich of peak, and most of WOT to be rich of peak. Peak being, obviously 14.7. Now, curiously, this is a table you don't touch for initial tuning. Initial tuning is all done open-loop. This table is tuned by riding experience, and is mainly a tweak for long-term steady-state cruising. Ok bud, great. How do you tune? Ok, there are five main zones in a VE table. The top couple of rows (maximum MAP) are WOT. These are pretty easy to tune for on the dyno. The bottom couple of rows are closed butterflies. This is the decel zone. This is also pretty easy to tune for, especially if you just want to shut fuel off as your're decelerating. The left couple of columns are idle. Towards the bottom specifically. I'm using a drive-by-wire, so my TPS will go to zero, but the ECU will send an idle control signal to the throttle unit to keep the throttle cracked at low RPM. This is instead of an idle stop screw like you see in earlier EFI setups or on carbs. The middle of the table is basically cruise. Your power and gearing determine where in the RPM and load range you like, but it's likely in the left half (below half RPM) and around the middle (about half throttle). This bike doesn't have a 6th gear, so I spend a lot of time at 5kRPM, which is 75-80mph. The rest of the table is whatever. I just leave interpolated data in there. It's possible to tune for these on a dyno using part-power pulls. In general, you turn on data logging, stick a wideband O2 in your exhaust and then try to get a few data points at every cell in the table. Ideally, you'd like to get 5 seconds continuous at each cell to make sure the wideband stabilizes. You then adjust your VE to get the right AFR. Repeat ad nauseum. There's also auto-tuning once your map is pretty close; this is a paid portion of TunerStudio, and I have no idea how it works, if at all. If I find money lying around, I may use it. Every time you make a table-wide change to VE, you should go back and make sure your ignition timing is still OK. This is typically done with plug chops and listening to the sound of your engine exploding (detonation). I'm trying to figure out how to tell TunerStudioMS to use a knock sensor, and how to wire a knock sensor into the speeduino. babyeatingpsychopath fucked around with this message at 02:48 on Apr 3, 2018 |
|
#
?
Apr 1, 2018 16:17
|
|
|
 I notice you're using a MAP instead of alpha-n; I take it that's just cause it's easier that way? Also early r6's had pseudo-CV TB's that had vacuum plungers, presumably to reduce snatchiness but I'm guessing you aren't using those.
|
|
#
?
Apr 1, 2018 20:23
|
|
|
babyeatingpsychopath posted:You need three sensors minimum to run EFI: TPS, crank, MAP.
|
|
#
?
Apr 1, 2018 21:35
|
|
|
Rev. Dr. Moses P. Lester posted:First generation BMW K bikes had no TPS, just an air flow sensor.  literally a car engine.
|
|
#
?
Apr 1, 2018 21:42
|
|
|
Slavvy posted:
I've got a rack of those, too. I'm gonna throw my hand in with these drive-by-wire ones because why not. So to clarify, there are three main ways to run EFI: MAP, MAF, and alpha/n. These use manifold pressure (the vacuum your engine is pulling or how much boost you've got going in), actual weight of the air going in, or by flat-out guessing (tps + current RPM). In "early" anything EFI, the systems are garbage, and we should ignore them. They're basically just as good as a carburetor, and no better. I'm only really going to get into systems that make sense to implement today: closed-loop EFI using off-the-shelf sensors. Open-loop anything is garbage. It is like a carb: you set it up for specific stuff at specific RPMs, and it delivers that fuel then. It doesn't know or care what the engine is actually doing. That said, there can be more inputs, so it acts (at every rpm) like a carb precisely tuned to that rpm at those temperature conditions. This is Quite Good, as the people at Harley have found out, but isn't the best. The best is closed-loop. Adding an oxygen sensor into your EFI system allows your system to respond to what your actual engine is actually doing, in close to real time. If you got a bad tank of gas, it can fix that, and keep your bike at peak power/efficiency. If you wander off the end of your tuning chart because of temperature or humidity, it can correct for that. The computer makes a really good guess at how much fuel to spray in, and then the oxygen sensor lets it know if that was too much/too little/just right. This feedback is called closed-loop mode and is where the true efficiencies of EFI exist.
|
|
#
?
Apr 1, 2018 22:53
|
|
|
I'm interested to hear how that arduino thing works out. I recently put together an ignition computer for my 2t race bike using a Teensy. It seems to work ok ish but my code isn't great and I think it'll have issues with very high RPM and running for longer than 20 minutes.
|
|
#
?
Apr 2, 2018 00:23
|
|
|
This is so fuckin cool. I'm really glad someone out there has done most of the heavy lifting on the arduino side because it's something I've been thinking about off and on. I'm definitely going to be following this thread because EFI is on the long-term list of upgrades to my moped, and I had no idea about the differences in open and closed loop. I would be interested in a post about throttle body situation, honestly. Having run into mis-matched carbs on stuff before I'm all ears as to the physics of air/fuel flow.
|
|
#
?
Apr 2, 2018 00:45
|
|
|
So today I dug a Mega2560 out of the parts bin and grabbed the R6 rack. It's got two dual TPSs on it. One is the throttle cable TPS, and is what you, the meatsack, set your "desired" throttle to be. The computer then, in its infinite wisdom, drives a servo which drives the butterflies, which will make the second TPS match the input, or at least what the computer thinks you should be doing. This (on the R6) prevents you from twisting the throttle open like a retard and having all the horsepower a high-strung I4 can throw at you come on all at once. This is drive-by-wire. It's also got the MAP onboard. I grabbed an Uno out of the parts bin and plugged it in to act as an RPM signal generator. I grabbed some protoboard and a fistful of resistors. I flashed the Speeduino Dec2017 firmware into the 2560. Loaded up tunerStudio, made a new project, pointed it at the .ini file from the firmware, and had live gauges! Threw some resistors on the board, dumped some fake values in for intake air temp and coolant temp so that TS would think it's 70F outside and the bike is at 180F; all warmed up. Wired up the MAP and TPS, and calibrated them. I then wrote a tiny arduino sketch to simulate a 30-1 missing tooth input wheel mounted to the camshaft. (note, it's RPM/4 square wave). Set the on time to be 10us, flashed it, and now I can control the most important gauges: TPS, MAP, and RPM. Plugged some reasonable numbers into tunerstudio for expected fuel and a VE table, and when i whack the throttle open, duty cycle goes way up. when MAP goes down, duty cycle goes up. when RPM goes up, duty cycle goes up. This is all expected and correct. So I have the ECU responding in a logical manner to inputs wired in a logical way. This was what I thought would be one of the harder parts, and I got it done in an evening.
|
|
#
?
Apr 2, 2018 03:06
|
|
|
If you want to go full ham, and DBW, I have a set of FZ10 throttle bodies, injectors, tps, and throttle servos that could work nicely with a custom intake.
|
|
#
?
Apr 2, 2018 06:11
|
|
|
Ehhhh. Second post edited. It's wild and rambling at best. In short, today I built tuning tables from scratch using math and pictures. Who knows if it will start and idle the bike. Stock mechanical advance units are garbage. I implemented that originally, then threw in a vacuum advance, because it should help with idle. I updated to the new controller firmware (Speeduino Feb 18), and got some new lights. What does "reset lock" do? I can't find it in the code, so  .
|
|
#
?
Apr 3, 2018 02:55
|
|
|
Were you trying to use the mechanical advance as a crank sensor for the whole ECU?
|
|
#
?
Apr 6, 2018 01:27
|
|
|
Rev. Dr. Moses P. Lester posted:Were you trying to use the mechanical advance as a crank sensor for the whole ECU? No, the mechanical unit is in there now; I'm getting a new missing-tooth gear made for this ECU. What I did originally in my spark map is duplicate what the stock mechanical advance unit does. That makes a spark table that is the opposite of smooth and efficient, and it also doesn't incorporate vacuum advance, so the thing runs super retarded in cruise. By my calculation, I should be able to run 40-42 degrees of advance at quarter throttle at 5200RPM, but the stock unit is just "37 degrees, everywhere" above 2700RPM. Actually the stock unit is 5deg from 0-1400 rpm, then it ramps up to 37deg at 2700RPM, and stays there. I also found out that these camshafts like to wear and flex so that the timing isn't the same between the 1-2 and 3-4 firings. This leads to crappy timing on the 1-2 at idle, which is something I've noticed. EFI and electronic spark can't come quickly enough.
|
|
#
?
Apr 6, 2018 03:37
|
|
|
This what I'm doing to my Gen 1 Vmax http://roadstercycle.com/yamaha_vmax_fuel_injector_cv_car.htm Microsquirt, batch-fired that uses carbs like throttle bodies.There's a fellow in Houston that's done this already, his bike runs great. Some might frown at the 90* deflectors but his bike ran smooth, no hiccups. I'd like to take this design approach with a ZRX 1200. R-Type fucked around with this message at 21:13 on Apr 6, 2018 |
|
#
?
Apr 6, 2018 21:11
|
|
|
R-Type posted:This what I'm doing to my Gen 1 Vmax That's clever. That's basically what the '03 R6 rack is. It's CV carbs with injectors instead of all the jets. One thing the page you linked brought to my attention that I'd forgotten was a tipover switch. I actually have a spare directional gyro and attitude indicator. I can probably get that wired in to the ECU 
|
|
#
?
Apr 7, 2018 05:25
|
|
|
Excellent thread so far. Can't wait to see the guts of it.
|
|
#
?
Apr 18, 2018 21:59
|
|
|
Quick question: if I wanted to go from alpha-n to a MAP, would I have to construct new tables etc from scratch? My TB's have the appropriate fittings and I can get any kind of sensor my heart desires from my wrecker, I'm just trying to work out what the effort:effect ratio is so I know if it's worth bothering with or not. Keep in mind I have no plenum or air box, it's just TB's with foam filters on the ends (yes I know this is bad but I can't afford velocity funnels ATM).
|
|
#
?
Apr 19, 2018 06:35
|
|
|
Slavvy posted:Quick question: if I wanted to go from alpha-n to a MAP, would I have to construct new tables etc from scratch? My TB's have the appropriate fittings and I can get any kind of sensor my heart desires from my wrecker, I'm just trying to work out what the effort:effect ratio is so I know if it's worth bothering with or not. Keep in mind I have no plenum or air box, it's just TB's with foam filters on the ends (yes I know this is bad but I can't afford velocity funnels ATM). You'll have to find a pressure sensor that works. ITBs with foam on them produce very very little vacuum, so it's hard for a MAP to sense anything. Find "a sensor from the wrecker" and plug it in and see what kind of readings you get. There are whole families of pressure sensors for <$30. You can probably just buy one that works for your range. After you find a working pressure sensor, then yeah, you're pretty much gonna have to build tables from scratch. Your alpha-n table is a good starting point, but it's probably only accurate in a fairly straight line corresponding to "closed throttle/idle" up to "WOT@maxRPM."
|
|
#
?
Apr 20, 2018 00:37
|
|
|
Quick update: wtmtronics.com's credit card processor got messed up so my order failed to process without letting the front end know, so my board order didn't go through. Got through to the owner/developer on slack and he reprocessed the order by hand last night. Couple of days for parts now. edit: I hope it comes in time to get everything running by May. The registration is due then, and I want to get the custom plate "76 EFI"
|
|
#
?
Apr 28, 2018 15:46
|
|
|
I'm currently rewriting my ignition timing code for the fifth time. It involves math, and I loving hate math.
|
|
#
?
Apr 28, 2018 23:43
|
|
|
And we're back! I had to go out of town for two weeks the day after I got the board. Soldered everything up today, got the code flashed, and now it's time to wire up connectors and things. The only parts I didn't get in the kit that would be nice are the daughter board double-length pins and two 8-pin IC sockets. I have those things in the junk drawer, though. edit: I connected the bluetooth module and that works, too. I've now got a dashboard on my phone. 
babyeatingpsychopath fucked around with this message at 00:02 on May 21, 2018 |
|
#
?
May 20, 2018 23:24
|
|
|
Needless to say, life and stuff got in the way. I'm not dead. I'm still mocking parts up and tuning the 3d printer to try to iterate through manifold designs. The tune I have will probably start the thing. I do need to go steal a wideband O2 out of something, though; that's the last sensor I need.
|
|
#
?
Jun 15, 2018 02:30
|
|
|
A small snag occurred. The water pump seal and bearing decided to fail, pressurizing my coolant system with hot engine oil! The water pump impeller has about a quarter inch of endplay now. So a water pump and gasket kit is on its way. While the engine is apart, I'm probably going to put all the EFI bits on. Wish me luck. edit: that's the drained coolant. The oil was clean, there just wasn't as much of it as I would have wished.
|
|
#
?
Jun 23, 2018 21:50
|
|
|
Yay working so much I never have any time to do anything! Finally got the engine front cover off today. I was going to do it last week then got called out of town for four days. The front cover is held on by 9 M6 JIS screws. Someone at one point had attempted to removed these with a phillips screwdriver and turned them all into golf tees. I used helical screw extractors and left-handed drill bits and got all 9 out. 7 with the extractors, 1 with anger and a hammer and chisels and things, and 1 with vice grips. New screws on order, then water pump in and ready to ride again. I'm thinking about an alternative crank angle sensor coming off the drive gear for the cam belts. The only problem is it's a normal gear, so there needs to be some way to get the TDC signal. More studying is in order. Note: my screw-extractor fu is quite strong. 
babyeatingpsychopath fucked around with this message at 23:55 on Jul 11, 2018 |
|
#
?
Jul 11, 2018 23:31
|
|
|
It's not a proper AI thread if a screw extractor doesn't make an appearance, though successfully using one is definitely an anomaly
|
|
#
?
Jul 18, 2018 01:19
|
|
|
Breakfast Feud posted:It's not a proper AI thread if a screw extractor doesn't make an appearance, though successfully using one is definitely an anomaly I'm a screw extractor master. That's seven extracted screws with one, and one with a left-hand drill bit.
|
|
#
?
Jul 19, 2018 00:18
|
|
|
This is a cool project. The 8-bit AVR in the arduino mega is enough to keep track of all the sensors and timing?
|
|
#
?
Jul 20, 2018 18:26
|
|
|
babyeatingpsychopath posted:I'm a screw extractor master. That's seven extracted screws with one, and one with a left-hand drill bit. But... how?
|
|
#
?
Jul 20, 2018 18:54
|
|
|
How do I make sure it's set up to use oil temperature instead of coolant? I can't find an option for it anywhere in tunerstudio and I can't imagine the two are interchangeable considering how hot oil gets compared to coolant. I'm pretty sure this is my problem, it runs well initially because of cold start enrichment but as soon as that goes away it runs like poo poo again. This has been a very frustrating experience for me because I can't even get it to loving idle. I just want it to idle and then I can play with the VE numbers gradually and methodically, like jetting a carb, but I can't do that if the loving thing wont idle!  E: I've found 'air cooled expanded CLT range' which according to the manual is what it should be set to. I still don't understand how that can work though. The MS uses CLT to calculate how much fuel to deliver, I don't understand how it can treat coolant temp and oil temp as the same thing and have no problems. I just don't understand in general.  I figured it out. Turns out my slight reduction in fuel was nowhere near enough so it was just killing sparkplugs. Two sets of plugs later and I have it idling reasonably clean and stable. Anyone know why sometimes if you're running pig rich the plugs foul and then never work again? As in I was able to wire brush them perfectly clean and still got no spark, this bike has cost me approximately $80 in spark plugs thus far and is not the first bike I've seen this phenomenon on.Next question: How do I reduce the amount of data logged by tunerstudio? It logs a billion variables and I guess because my laptop sucks this means the log fills up after twenty seconds. I only need like seven variables. Slavvy fucked around with this message at 05:54 on Jul 25, 2018 |
|
#
?
Jul 25, 2018 03:21
|
|
|
Slavvy posted:I'm now at the point where I'm questioning everything on this thing and the thing that jumps out at me the most is the fact that it's set up to run a coolant temp sensor, but the bike is air cooled and the temp sensor lives in the engine oil. If a plug is fouled enough, it shorts and burns out the electrode internally. Wire brushing won't fix that. Bead/soda/glass blasting MIGHT recover it, maybe. As far as datalogging: Find your project directory\projectCfg\mainController.ini It's full of good stuff. There's a datalog section in there that you can trim if you wish.
|
|
#
?
Jul 28, 2018 04:06
|
|
|
babyeatingpsychopath posted:You can make your tables ignore coolant temp. That's fine. Look at the warmup enrichment table. Set it to 100% at anything above room temperature. In fact, for an oil-cooled bike, I'd set two points. 150% WUE at 40F and 100%WUE at 90F. Thanks to this post and a shitload of frowning, I now understand what I'm starting from and where I need to go and pretty much everything. Merrily I went for a logging ride, which is a lot funner when the bike isn't gargling fuel. Feels like it already has a bigger midrange than factory carbs. Then I found out megalog viewer only lets you look at 400 records unless you fork out for the full version. 400 records is like 20 seconds. I'm forking out for the full version, aren't I? E: I forked out for the full version. Now to wait for dry weather. Slavvy fucked around with this message at 08:06 on Jul 28, 2018 |
|
#
?
Jul 28, 2018 07:40
|
|
|
Slavvy posted:Thanks to this post and a shitload of frowning, I now understand what I'm starting from and where I need to go and pretty much everything. I cannot stress this enough: tunerstudio defaults are designed for American Big Iron. If you don't touch a table or don't know what something does, assume it's to make a big-block V8 run well without tinkering from the box using a megasquirt. The further your motor deviates from a V-8, the more stuff you have to touch directly and know what it does to make the tunes work for you.
|
|
#
?
Jul 28, 2018 13:07
|
|
|
Thing is, all this work has already been done by someone who appeared to have known what they're doing, I'm just trying to make it work with a different exhaust. So it isn't as arduous as doing it from scratch.
|
|
#
?
Jul 28, 2018 21:14
|
|
|
My logging setup, complete with FTP camo: Yes officer this is my motorcycle with a headlight. Things I've learned: Happy running is around 13.5-14 AFR pretty much across the board, but idle has to be like 12.5 or it's too choppy and lean from I guess stalling airspeed around the TB plates. It's harder to get good static throttle running than good clean acceleration, I can see why the factories struggle so much what with emissions regs, noise et al; lack of velocity funnels is really holding the bike back in a rideability sense. I'm hopefully rectifying that soon. Surprisingly, off-idle response is really crisp and clean most of the time. My cut up katana 1100 muffler has an interesting effect on the volumetric efficiency of the motor, judging by the changes VE analyse makes and comparing that to the original tune it came with. The midrange at large throttle openings has been gutted, yet some parts of the top and bottom of the range are actually pretty decent. I guess that's where the cam is at it's best and the pipe really strangles it, up top it just bulls through but isn't breathing that hard anyway. Because I need gixxer cams. Oil temperature is a lot better and holding steadier now that I've changed the oil. Bandit oil capacity: 3.5L filled by me. Drained oil volume: 5L. It came out looking like 0w20, it's been running so rich 1.5L of petrol has made it into the sump over the course of like five hours' running. Please god let my 80's tank engine hold together. Oil heat soak I think will continue to be a problem, proves alpha-n really is no better than carbs as it gets thrown out a bit when it's been running for ages and the oil temp is soaring. A wideband of my own is on the shopping list, do I need something super fancy or will a cheap shitter do me? It's an air cooled I4 at the end of the day. In terms of performance and what it feels like, I have a mildly tickled GSX1200 on hand to compare to. Roll-on and general response is pretty savage, the midrange is loving biblical and really encourages me to get it road-legal so I can ditch that pipe and take full advantage of the motor! Slavvy fucked around with this message at 08:56 on Aug 6, 2018 |
|
#
?
Aug 6, 2018 08:51
|
|
|
I've been so busy with my job that I haven't had any time at all to work on this until today. EXCEPT FOR SOFTWARE! I'm pretty sure I have a new cruise-control mode figured out. Because I can directly control fuelling and there's a knock sensor, I can lean the engine as much as I want. Also, because it's got throttle-by-wire, I can close the throttle plate if stuff starts going south. This means I can have a cruise control mode that won't go below X mpg. I'm not sure what X is right now, because I don't know how efficient the engine is going to be. Basically it works like this: Set cruise mode to "efficiency." Use the display to set the max, min, and target speed and the MPG target. There will eventually be a button for common settings. The computer will try to hold the target speed setting by injecting AT MOST, the amount of fuel required to get the required MPG. If the engine can provide the power to keep that setting, then the speed goes up. There's a PID loop (that's going to get real-engine tuning) to adjust the throttle position to keep the speed. The IMU is currently sensitive enough to detect a pretty small downhill grade, and that lets the computer know to allow the speed to get up to max. Likewise, if on an uphill grade, the speed will be allowed to drift down to min. This all means that in reasonably flat territory, I can just tell the bike "get me 30mpg" and it'll do it. If it ends up that all this whizbang timing stuff lets me actually get decent power out of this lump, then I could theoretically set it to 40 or 50mpg at reasonable speeds and just cruise like that. Of course, actual road speed will drift around a bit. The other, more boring cruise modes are "throttle lock" where it just keeps the current throttle opening. The MPG seems to change based on engine load, which makes sense because of differing vacuum means differing injection, and that's OK; it's what a carb does with a locked throttle, too. There's also the very-boring-indeed "constant speed" which is what normal people think cruise control should be.
|
|
#
?
Sep 3, 2018 15:18
|
|
|
OK I feel like I'm losing my mind here. I've been merrily playing around with logging and VE analyze for the past few weeks and everything has been working fine. The last time I connected to the bike about a week ago, everything worked ok. Now tunerstudio can't connect at all. Things I have tested: USB to serial cable is ok, I can do a successful feedback loop through it on my laptop. The wiring from the comms jack to the MS connector tests fine. Port settings on tunerstudio are all exactly the same as they always were, using the detect port function just leads to it saying it can't find anything. I have touched nothing on the bike since last time, the battery is fully charged and the bike starts and runs fine. At this point I have no choice but to assume some sort of freak butterfly effect has caused the MS to poo poo itself in some very specific way. Is there some way of testing the microsquirt directly or resetting it from scratch somehow? All the MS documentation is a baffling maze, I can't seem to find anything explaining what to do in this situation.
|
|
#
?
Oct 14, 2018 02:32
|
|
|
Slavvy posted:OK I feel like I'm losing my mind here. I've been merrily playing around with logging and VE analyze for the past few weeks and everything has been working fine. The last time I connected to the bike about a week ago, everything worked ok. Now tunerstudio can't connect at all. When you say "port settings" in tunerstudio is that serial port settings, like com port number and stuff? You could check device manager in windows to see what com port number the equipment is getting, it might change on different USB ports on your computer
|
|
#
?
Oct 14, 2018 02:39
|
|
|
I'm on Linux and yeah I mean the com port and baud etc. Like I said everything is exactly the same as last time I plugged into the bike, the specific USB port I use has never mattered in the past but I tend to use the same one every time anyway. I am able to use minicom to test the port and cable all at once in Linux and everything checks out ok, I'm 99% certain the problem isn't laptop related.
|
|
#
?
Oct 14, 2018 02:53
|
|
|
Ah yeah Linux doesn't have com port numbers they're /dev/ttyS0 or whatever right? It still kinda sounds like a problem in tunerstudio to me, does it give the same error message if the cable is disconnected entirely, "can't detect port"?
|
|
#
?
Oct 14, 2018 03:02
|
|
|
|
| # ? Apr 27, 2024 22:40 |
|
|
Sorry I think I was being too ambiguous in my frustration. There's no error message, it just behaves like it isn't plugged in When I go into communications settings/detect device, it says 'no controller found' and to check your connections, drivers etc. Exactly how it does if it isn't plugged in or the bike/ecu aren't powered up. The result is the same if I try it with the bike unplugged, it just gets there instantly whereas when it's plugged in the progress bar takes a couple of seconds longer. My com port is /dev/ttyUSB0 yeah, I only ever had trouble with that when I first had to get everything working as I'm pretty new to linux. I've used minicom to loop test the entire connection chain right to the ampseal plug at the MS, so I'm convinced linux and my cable are not the problem. I can believe it's TS but it hasn't had an update and I haven't changed any settings so IDK why that would be, but I also don't see how MS could have a problem with comms from a surge or whatever but still run the bike hunky dory. Evidently it's possible to communicate directly with the MS in linux using minicom and the MS in boot loader mode, I'm desperately afraid of breaking something or losing some setting that isn't saved in the tune files so treating that with caution for now. Slavvy fucked around with this message at 19:57 on Oct 14, 2018 |
|
#
?
Oct 14, 2018 19:54
|
|