Project FreeEMS Powered WikiSpeed! (No 3g/56k)

[Fred]

November 12th

Yesterday evening I remembered that I had taken a bin log and added some special debug to the firmware in advance. I reached out of bed and pulled up the computer and opened the log, a glorious sight shone back at me. This morning I tidied up the ULV settings so it made sense and got this:






Which I eventually turned into some code after burning a lot of time on my good friend Sean and my good wife Isabel. Tonight I headed down to the shop and tried it, and it didn't work. I have now (probably) fixed it, and will try again in the morning after breakfast. Once I have sync and RPM I will try to wire up some coils and get some ignition happening. Then it's time to use a timing light to dial in the offset. Once that's nailed it's time to run the thing! First things first, though.

[Fred]

November 13th

Lots of good things hapoened today. I got the decoder working. I photographed an injector. I scoped an OEM injector pw and dwell (pw cold and warm). With a little help I set the decoder offset within 5 or 10 degrees of correct. I drove the wikispeed car about 30 feet total. 15 feet of that was with the wheels spinning.

And perhaps more?

Injector pw is about 4ms cold and 2ms warm, electrical/total.
Dwell is about 2.5ms idling.

I got screwed by EMStudio a lot today. Not too happy about that, but our man is a software guy, so he understands.

Here is a screen shot of a log of the engine cranking three times for extended periods. Can anyone tell me what is special about the way this decoder works? Hit me up in IRC with your answer :-)





Tonight I will research what the pulley markings mean and tomorrow I plan to wire up the Jag to the connector and try to run the engine. All sorts of things could and likely will go wrong. Wish me luck. :-p

Fred.

[Fred]

November 14th

Thursday consisted of dismantling the OEM ECU for its enclosure and modifying it to receive a new connector and a Jaguar board! Pics will tell the story best, so here they are:



Pressing it out took a significant amount of force:







But in the end I won:











PCB support removed to make room for the Jaguar:





Tabs constructed to hold the connector in tightly:











Novel use of the correct Jaguar case as a PCB holder within the OEM case:

[Fred]

November 15th

Today, like yesterday, I spent a bunch of time helping trace wires and find pins in the OEM ECU connector. The end result was a few wires soldered up ready for the Jaguar. We still need to find the 5V out pin(s), but that's about all. I think everything else is covered in terms of knowledge. I just need to cover the pins on solder, too. Oh, and grounds/powers and the splits between them. This isn't too key yet, but will become utterly key once the DBW stuff comes into play.

[Fred]

November 16th

Today we more or less completed the wiring tracing and pin testing to the point where I could start connecting the dots. Huggins, who some of you will meet soon, found and fixed the intermittent starting issue, go him! I hooked up power, ground, fuel pump relay output, cam and crank sensors, and plugged it in. It didn't work. I had the numbering confused, so had to move a single wire across by 4 (I got lucky on some others). Then power worked, but cam and crank didn't seem to. I couldn't plug in the USB cable, so I modified the case to allow that, then we went for dinner. Tomorrow I'll hook EMStudio to the box and try to log the cranking, if no luck, I'll diagnose that first, then move on to MAP, IAT, CHT, and eventually 4 coils and 4 injectors. At that point it should fire up. Video tomorrow, with some luck. Pics from today:



Unique modern style MAF + IAT:











Innards of a destroyed MAP sensor:







Configured for COP use:





Honda's timing mark scheme. The white mark is TDC and the red mark is 8 BTDC with the two each side of it +/- 2 degrees. On the engine there is a single mark to align with each of these, depending upon your purpose:











Plugged into the loom in the car:







Spot the weirdness:





Spotted? Count from 20 to 25 :-p I was thinking 9, 10, 20, 30, 40 were the right most pins. Wrong. Woops.

Notched for USB connection:

[Fred]

November 17th

viewtopic.php?f=55&t=2347

She runs! Badly! On 3 lol. (1,2,4) Proof: http://youtu.be/UXa16bqa_kM (uploading, try again in an hour)

Injector guess was close, used Ben's default tables, used a leak in the throttle to raise RPM, had timing pretty wrong, flooded it, cleared it, advanced it, had it splutter a bunch, finally fired up. TODO list for a general clean up of existing work done:

1. Get number 3 running too!
2. Get timing dialed in properly!
3. Get it warmed up and idling smoothly at a reasonable guessed AFR based on electrical PW from scope of Honda ECU
4. Get it tuned up to whatever the leak provides
5. Get wideband setup and working, somehow?
6. Use an injector to open throttle wide for rev limiter action?
7. Backport all live changes to code base
8. Change decoder event order to bring required offset down low and change order of cylinders back to 1342 instead of 4213 as it is now.
9. Make it start easily by tuning the priming pulse with zero VE until it fires then dies, then add VE back in
10. Ensure warm up curve is reasonable for this engine
11. Debug EMStudio crashes and help mike fix it (this could come earlier than 11th)
12. ECU keeps itself powered up, somehow. Need to resolve this as turning off the key did not kill the engine. Hmmmm.
13. Add missing two grounds to FET setup and CPU setup that I couldn't find earlier. Unessential, but would make it a bit more proper.
14. Add fan control on LSD1 once Andy tells me how it should be connected up (board markings conflict)

Next step:

Add DBW support including APP (Accelerator Pedal Position) reading and PID throttle control. TODO:

• PID code
• PWM code
• Dual TPS interpretation code
• Dual AAP interpretation code
• Fail safe behaviours (mirrored variables, correlation mismatches, etc)
• Soft limits to movement (don't slam throttle with full current into closed or open positions)
• Non-linear P? Voltage dependent base value? Voltage dependent PID values?

Bonus step:

Get tip-in enrich patch and apply on top of published base tune and decoder stuff for crisp drivability.

Final step:

Figure out the automatic gearbox and control that enough to drive it around.

Bonus post-final step:

Brief road trip for "testing" :-)

Fred.

[Fred]

November 18th

Tried to fire up the engine again first thing this morning and it went wild. The board went wild. Took me a while to figure out what was going on, but it was a barely broken regulator lead being intermittent in a high frequency way. Fixed.

Replacing the FET didn't help. Then I measured 270k between CPU and FET input. Then I changed resistor which seemed to fall apart. Still no go. Then I eventually found a fraction of a millimeter of trace missing, bridged that and we were all go.

Fired up the engine on four cylinders and shot this video: http://www.youtube.com/watch?v=FTE6DZ9oBeg

And got this log up to about 3800 RPM, peak:





Runs really nicely when it's in a good spot, and stumbles and barely lives and spits and coughs when not. I have a strong feeling that it's ULTRA lean and will get a wideband on it tomorrow, IF I can get EMStudio behaving, and do some basic tuning till it runs OK. Then I'll dial in the timing exactly and do some iterations from there.

Meant to do some work on the data and source-coding of it tonight, but didn't get a chance. Too busy waiting for a glacial browser. Ugh. Speaking of which, if anyone had a PM conversation with me on IRC with the following nicks, please email me the content of the chat:

BusyTidyingFred, SatisfiedFred, BloatedFred, R18A11Fred0, Round2WithR18, R18WithOffset

Fred.

[Fred]

November 19th

Today was another interesting and chaotic experience! I got the car running OK with some by-ear tuning. Starts nice, revs high, feels smooth, etc. Then I hooked up the test lead to put the timing light on and verify the base timing... and the CPU lockups began! Non-resistor plug + COP unit + short lead outside the engine = massive RF. Runs fine with the COP in place, but not with that test setup in place, not at all. So tonight I'll try to add COP (computer operating properly) monitoring to the firmware so that if it does lock up, it resets and we don't have what happened today happen again: flooded engine. One of the lockups the injectors all stuck hard on. Not good.

On the brighter side, the engine is sounding pretty healthy now, as evidenced here: http://youtu.be/sdm3yVVEXw4

And let me see if I can dig up a log... damn, 24 log files... ULV needs a new feature: batch open.

Nothing too interesting. The highest one I saw was 5k RPM with a ~10mm air leak, plus throttle bypass.

[Fred]

November 20th

Today was an epic day, and still is being!

It started just before I was due for a shower: Skype message: "I'm at the workshop" - reply: "You're here???" reply: "yes" reply: "come to the hotel, will meet you outside." And Peter shows up in a dodge ram, not the AMC I was looking for. First meeting of one of my long time un-met friends! Awesome!

Isabel and Peter and I got in the car and headed down to SeaTac airport to pick up Mike and Andy and drop off Isabel to meet a friend. Headed back to the workshop with Mike and Peter and Andy and got stuck into the car.

Managed to fix some issues with EMStudio, kudos to Mike. Managed to get a wideband installed, kudos to Peter, Andy and myself. Did a little bit of work on getting the DBW stuff setup, but nothing concrete, no solder or code.

Headed down to airport AGAIN with Peter and picked up Jeff, went back via the workshop to see the car etc. Tried to pick up wife, failed. Picked up wine instead, drank two bottles in hotel room between 5 people. Poor effort. Thank you Peter for the nice Kiwi Syrahs! <3

A good day!

Tomorrow, calibration for wideband, some tuning of three load sites and things in between, then more DBW work! Good times.

Fred.

[Fred]

November 21st

Another epic day!

Peter and Andy worked epicly hard for most of it diagnosing and fixing the fan wiring/relays/fuses/etc. The car now has fans! Great, we needed that.

Mike hacked hard on EMStudio all day fixing numerous things! Great work, what a star. Of course, without Jeff bitching at him constantly, there wouldn't have been anything to fix! Go Jeff!

I worked with Aitor to get the table converter utility in a good state for mostly hands-free conversions! Wonderful utility, Aitor! <3

I hooked up some extension wires with obvious colours to a pigtail for the throttle and played with that a bit. This enabled us to both open and close the throttle while the engine was running with manual-Fred-PWM-Idle-and-DBW-control. Look forward to the video, such good fun.

I also generally coordinated and helped everyone, much like I do the rest of the time, just in person instead.

Amazing team work! Totally stoked! :-)

No pics/vids now, but some time soon, hopefully!

Fred.