Fred's firmware development diary

[Fred]

48 hours after the fix and two more cars are running:

Number 8: viewtopic.php?f=55&t=1507
Number 9: viewtopic.php?f=55&t=1508

Epic success, 3 days into the new year and 2 more cars running, how many by 1st Jan 2013? Maybe I should start a poll?

viewtopic.php?f=41&t=1509 Done :-)

Fred.

[Fred]

Dev is stalled for a few days, starting 2 days ago, and ending in a day or two, due to moving house. Hopefully I can knock out most of the unpacking and arranging and cleaning and so on tomorrow, we'll see. I'd like to complete the work that I've half done. I'll be mostly offline for the next week or two as well due to lack of organised-ness to get a DSL connection setup here. Fail. Anyway, I can go to a library to sort out stuff on the forum and push commits and continue working on things offline here. I'm more productive overall that way most of the time anyway. There shouldn't be too much that I need to google, hopefully. Hopefully not too much that I need from the forum to do this little bit of work, also. Might check that I have what I need open now... :-)

[Fred]

Generic coarse bit bang code works great first try! Coded without test hardware or internet, another win while offline :-)

Will sort out some config for em_knaps and sim and get it pushed.

Fred.

[Fred]

Video proof from NZ of the new GP BB code:

http://www.youtube.com/watch?v=j6QgtQygP_0

Pretty cool :-)

Fred.

[Fred]

The missing tooth + lacerated pempheridae + stim hack has shown up something I've known to be bad for a few years:

http://stuff.fredcooke.com/12minus1.0-50k.sweep.png

Behaviour is good to 13500, and at that point one output stays "stuck" on, except that it's not stuck, it's just the first one to get scheduled, so it is always output right after sync before losing sync again and therefore not scheduling the other events. There are a few fixes required to correct this behaviour. Number one is when losing sync after initial starting, to wait to confirm resync by at least one decoder cycle, configurably more. Number two I can't remember right now, somewhat tired, 3am. They're all already known and in the tracker, though.

The two lines at 13500 and 36000 or so are where it starts losing sync and where the scheduling for that output is no longer on the first tooth, respectively. The other artifacts are the same thing happening again with different schedule values and different sync points, some falsely.

Source file if you'd like to examine it yourself:

http://stuff.fredcooke.com/12minus1.0-5 ... .logicdata

Will test various other combos tomorrow :-)

Fred.

[Fred]

Tomorrow, along with putting away the washing, flattening a bunch of cardboard boxes, doing the dishes, shopping for a knife and/or chopping board, etc, I'm going to attempt to trace the issue with the serial stopping and get that fixed. Wish me luck. Got 4 hours sleep last night and drove a couple of hundred km today, so pretty knackered.

[Fred]

^ that didn't happen. Instead while I was writing a post the lovely wife accidentally took my wifi connection (and her phone) to work! So I said "fuck it" and went back to sleep. When i re-awoke I suddenly had an idea and spent an hour writing 20 lines of code to prevent bad things from happening on noisy crappy RPM sensor setups. It's not complete/correct yet, but it's pretty much there. I need to make some changes to the decoders (or it'll never run anything again) and setup some variables and config for it. It's now 4:40am and I'm writing this because I can't sleep. Tomorrow is going to be a slow day, I think. DSL is supposed to be connected, various chores need to be done, the above code should get finished, the further above debugging and bug fixing should get done. And numerous other things on the "must do today" list that I wrote a week ago :-/

Back to a steady routine with the internet back and good speed/bandwidth to play with. Hopefully I can make more rapid progress. After that bug fix and some missing tooth improvements and the new code and a few other things, I'm finally going to add some rev limiter code. Well overdue, but should be fun to watch em_knaps test on his frenchy at 7k in boost :-)

Fred.

[Fred]

I've got net! Lots to do today, and house stuff over the weekend, but I'll definitely get some good work done too, I promise.

[Fred]

Lots of Ravage review, Puma HowTo review, Jaguar (new) review, OLV testing, tool testing, and general mentoring has gone on in the last few days.

Tonight I've ordered a bunch of parts and bought a new domain name, to be revealed soonish, with some luck. I'll continue ordering parts tomorrow. Hopefully I can place some bigger orders in the next few weeks and get something assembled and available to buy soon afterward. I'm working on something very DIY in nature. That is, after all, the spirit of this site, and always will be.

Fred.

[Fred]

Comms bug is understood now, just need to fix it, should be fairly straight forward.

http://stuff.fredcooke.com/CommsISRGoneWild.png
http://stuff.fredcooke.com/WhereTheShitHitTheFan.png





Fred.