Digital dash build log

[Extent]

Been having lots more problems with long traces on the proto board picking up interference and messing everything up  (fun fact, I once had the circuit completely disconnected from the bike, powered by a 9v battery, and sitting on the tail seat, and it was perfectly counting the RPM  :o )  So I started to build it up for real on a bit of breadboard



Lower left is the voltage regulation, with a 5v tap right at the output for testing and voltage sense resistors in the lower right.  Upper left is the Oled header w/ control bits.  The middle is the microprocessor, timing crystal, and various support capacitors, upper right is the super capacitor to let it gracefully shutdown when power is lost.  Lower right is the wheel sensor debounce circuitry and 2pin header for the sensor, and the hex inverter/Schmidt trigger.

With it built up this way it runs stable on it's own no problem, but issues resurface as soon as I plug in a lead to connect the tach signal.  Even a 6" wire not connected to the bike is enough to trigger spurious interrupts on all inputs (not just the tach input) eventually leading to a complete crash.  Even just attaching my test leads to the output pins I can pick up very clean 1.5v spikes in time w/ the tach pulse.

I'm rethinking my whole approach to the tach sense circuit right now.  The thought had occurred to opto isolate it, but I'm not sure how much that would gain me with a shared ground.  Instead I think I'll try to put the signal through an NPN transistor.  Since a transistor is a current controlled device it doesn't care about the voltage spikes, and it should give me a perfectly clean 5v signal without all the voltage dividers and diodes, and debouncing should be much simpler.

[Extent]

And the winning answer for all the "interference" problems?  Ground loop!  New tach sense circuitry is simpler and more tuneable.  Here's the direct output of the now cleaned up tach signal.


All the bouncing and spikes are gone, there's still a little bit of noise but not enough to cause false triggers (it could probably be filtered even less and still be good enough)

The now updated schematic


And the changes on the actual prototype.


The board hooked up to the bike for testing


And a video of it all running.  First part shows the tach reving up and down, with a couple of quick blasts up through the range.  I also show it running side by side with an inductive tach running off the rear cyl.  It's a little hard to read all the numbers through the flaring, but it's dead accurate.  At the very end it shows the speedo and odometer ticking up while I spin the front wheel by hand (which is why this part is extra shaky, but the big number is still readable :p )

Aside from a couple of very tiny simple tweaks (button network, ICSP header, Vin diode, wheel sensor ground, SD Card) this board is roadworthy.  Now just mounting it and software to write...

[The Prophet of Doom]

This is amazingly cool.
The OLED is fairly small - that would leave a lot of empty space in the dash.  Are there larger units available?

[The Prophet of Doom]

Another question - do you have any plans for the turn signal self cancellation? this is one of the great features of the XZ, but does need a pulse from the speedo to work

[QBS]

I watched the video.  Your bike sounds likes it's running very well.

[Extent]

It idles alright, but the YICS is a little wonky and the exhaust is in terrible shape flow-wise.  I don't mind the tone of the SV can, but I've still got a lot of work to do to get it to run right.

The display is very small, but it's the same size as the Sigma display that I've been using for years, so it seemed like a good balance of size/price to start off with.  Unfortunately it's not quite as daylight readable as I had hoped (not bad in daylight, unreadable in direct sunlight :/ )  so I've got to find a better display to upgrade to eventually.  There are larger ones from other manufacturers, I've just got to do more research and rewrite the code.

I hadn't been thinking about the turn signal cancel because I haven't had that working in years.  I was thinking about replacing the entire blinker circuit internally, but when I mount it up I'll see if I can find an old speedo around somewhere and duplicate it, shouldn't be that hard.

[Mid Life Crisis]

Nice!  I want One!

[Extent]

http://www.youtube.com/watch?v=0hyH8rG7l6s

video walk thru of the menuing system.  I can now useably reconfigure settings on the fly without having to reflash the entire software from the computer.

As a sign of how crappy my code is I'm already using 2/3 of the program space, and 3/4 of the RAM of the micro with just this.  Program space I don't think will be a problem since most of the large chunks of code are done, but with less than 500 bytes of RAM left that may come and bite me not too far in the future :/

Thankfully the only large bit of software that's still not in it is the SD card library, so I should be able to manage doing a bit of cleanup and getting everything to fit.  It does mean that for future hardware revisions I will probably have to bump up to the next grade of microcontroller (+10$, and no DIP package available), but we'll see how much we can squeeze out of this one first.

This weekend hopefully I'll start looking at mounting the thing up in the gauges and giving it a real live test.

[vadasz1]

Cool display.

What is the physical size of the OLED you are using?  Can you make it twice as big easily.

[Lucky]

I hereby nominate Extent for Visionary Of The Year!!
Most Awesome!!!!

[fret not]

For the next miracle there will be an improved TCI box! :o ;) :D (hint hint)

[The Prophet of Doom]

For the next miracle there will be an improved TCI box! :o ;) :D (hint hint)

Lots of the germans use and recommend these
www.ignitech.cz

[fret not]

Thanks for that.   :D

[Extent]

The screen is 1.5", there is a slightly large one available, but it's only a few pixels wider.  This manufacturer does have a 2.8" screens, but the price starts skyrocketing.  I'm also not convinced that this will be the best screen tech, so I'm not going to take it any further until I get some more testing and research done.

I have thought about the TCI before, and I do want to try fuel injection in my long term plans so there's always a possibility.  Those IgniTech units look pretty nice tho.  Maybe I'll just save the headache and buy one of them :p


Test ride today with mixed results.  Tach sense problems raised their ugly head again.  It seems to operate normally at idle for a period of time, but slowly starts to pick up fake wheel sensor triggers and eventually locks up completely.

Good news is with the tach input disconnected everything seems to run perfectly, so I've at least got a speedo and that's what I really need working to ride the bike :D

2nd bad news is that the bike itsself is running extremely poorly.  Gutless to the point of stalling off the line, bucking and surging at low speed, and with YICS capped an extremely slow return to idle (even thou it passes carb cleaner spray test)  She's really getting back at me for neglecting her for so long :/  So gotta go back to basics and attack the carbs from scratch (and hope that I haven't compromised the CDI somehow with some of my early test circuits)

Updated prototype, with the button resistor network in the upper right now.


Fitting the board.  It just barely works.  Thankfully I should be able to make it like half the size when I make a real PCB and change the whole design over to surface mount.  Also will gain tons of extra room by replacing the volt and temp meters eventually.


All wired up and getting ready to go.


All the components pushed in and the cover replaced.  Nothing fancy yet.


--edit--
20second video of the thing running while the bike was warming up.

Source code is finally posted here

[The Prophet of Doom]

Love your Volt meter - never seen it done like that before!

Microsquirters seem to have similar issues with noise generating false signals.  One common trick seems to be to wrap HT leads in a few layers of grounded aluminium foil, or to use shielded HT leads.

[Extent]

hmm, it's not just a false signal tho, it's effecting the entire micro and causing it to behave erratically and crash.  And it's nothing that I've been able to see with the scope hooked up to the input pins (when the input was floating previously and it was picking up tach signal out of the air I could observe ~1.5v spikes, the current circuit shows nothing)

The VOM I quite liked as well, unfortunately it was just a cheap Autozone meter and the needle is very light so it bounces around unreadable so basically when at anything other than a dead stop with the engine off :p  I still have to do electrical tests with my hand meter.

[Extent]

No big development to report so far, still doing a lot of debugging.  It seems that ignition interference is still the name of the game afterall.  Good(ish) news it it doesn't look like the board or processor is being effected too badly, the bad news is the interference is causing the screen to crash.  I'm going to have to start playing with shielded cables and enclosures.  It should improve a lot when I get the design printed up on a real board, but I still need to get a workable prototype in the meantime.

The shame is outside of this interference problem I think the soft/hard design is proving pretty solid.  I've just run two 24 hr tests back to back of the display code, and it just keeps on chugging, no problems.

Hooked up a 2nd screen for testing.  Both screens are equally effected by the interference (just not always at the same time)  The character LCD is also looking like a much much more likely contender for the "usable in the daylight" issue.  A transflexive char LCD just like this one is only 10-15 bucks.

[Hartless]

so hopefully im not jacking your build thread, but i had an interesting idea and i dont know how possible it is...... i was wondering if there would be some way you would be able to integrate this into a helmet to make a heads up display... then there would be no need to look down thus causing you to take your eyes of the road.....maybe just have a wire that attaches to your helmet or bluetooth that transfers it to a lens screen..... idk just sounds like a cool idea to me, but i dont know anything about that kind of stuff. maybe they already have one idk haha, i just figure you out of anyone would know if it is possible

[Extent]

There was one a few years ago, but it was expensive and looks like it's dead now.
http://www.motionresearch.com/sv.php
http://www.webbikeworld.com/heads-up-display/

HUDs are extremely difficult to do, I wouldn't count on that happening :p  The closest thing I've ever thought about would be like a single warning indicator (imagine the fringe of your visor glowing red if your radar detector triggered for instance)

[Hartless]

There was one a few years ago, but it was expensive and looks like it's dead now.
http://www.motionresearch.com/sv.php
http://www.webbikeworld.com/heads-up-display/

HUDs are extremely difficult to do, I wouldn't count on that happening :p  The closest thing I've ever thought about would be like a single warning indicator (imagine the fringe of your visor glowing red if your radar detector triggered for instance)

ahhh, i see. well, you sound pretty tech savy so i figured id ask :P