bugfight

ok, several soldering errors (did i mention this is my first smd project?) and several software bugs later, we have all buttons/leds working in test program. had to kludge it a bit, but it's ok for testing. in the dark: Here i changed the duty cycle to 1/4 (the sm_simple program uses 1/8) which brightened things up considerably: and, as my adoring fans have come to expect, the dark shot: next i need to build out a second brd and change the software to handle both sides. also there is some amount of instability in these buttons. some buttons take a large amount of force to react, and there is a good deal of "bouncing." i will look into debounce and also cleaning and treating the pads on the board, it may be that (as SmashTV strongly suggests) ENIG is the only way to go...

V = I R aka ohms law.

that's the basic idea, not too difficult but not quite a piece of cake. i'm prolly not the guy to tell you how... but here's my take the only trick will be the timer. you will need a few vars to make this happen, at least -a counter (say dblclick_cnt reset on 1st click) -a var (say dblclick_chk) to indicate you are looking for a second click (because counter will loop) you will need a timer to inc the counter (may be a mios func you can reuse) on click, if dblclick_chk AND (counter < DBL_CLICK_DELAY) set dblclick_chk false call dblclickfunc else set dblclick_cnt = 0 set dblcick_chk = true call clickfunc (on double-click the click function will be called on the 1st click, of course) another switch function i've found useful is the long press. (if switch is held down for a while) suggest you post in assembly programming forum (search first though) just ask about double click, doesn't matter that it's a footswitch, that's just a switch like any other in code land.

hehe translation: yes. *whack*

way cool, can't wait to get them in my grubby hands! should be this week, hopefully...

akkk doesn't work at first, too many leds lit up, core won't boot etc etc ahhhhhhh ok calm down, remove transistors, try again akkk core boots, but buttons don't work. don't take torch to pcb! did I mention I tried to hack together a test program from seq code? too much in there... brain weakening... losing consiousness... ok screw it, use the sm-simple example pgm. It will work without the leds connected... ok now we get somewhere, found some solder bridges etc. did i mention this was my first smd project? the test rig: evidence of buttons working (or clever hacking *whack*) ok hacked sm-simple to handle one row of leds (was simpler that way) and connected one transistor: a bit dim for sunlight viewing but look at it in the dark: next let's see all them leds lit up. to be continued...

Next the pcbs, which I ordered from Gold Phoenix: One test subject on the slab: Did i mention this was my first SMD project?? 32 fargin diodes, 10 fargin resistors, 4 fargin caps, 4 fargin sr, 2 res packs etc later: 32 bi-color leds later ahhh through-hole, like a cool spring breeze... next the testing...

Next the meat of my concept, I use SMD components to make a 4x8 panel, with all shift registers on board. This allows the pcb to be tiled either right to left or top to bottom. This way with two boards I can make either the 4x16 MB-Seq duo matrix, or an 8x8 monome like control. With the interconnects being only serial, the usual spaghetti mess should be eliminated. But first thing is that the MB-Seq duo matrix design needs improvement. As TK points out in his comment "CIRCUIT NEEDS TO BE IMPROVED FOR HIGH CURRENT LEDS!" the DEFAULT_SRM_DOUT_CATHODES<n> shift registers are sinking way more current than they can handle. In the seq the extreme case is 9 leds at once and should be rare. But for general case max is 16 leds, and with these lighting through silicon buttons, each will require more current than TKs. According to my calcs, and tests run by SmashTV seem to concur, the source shift registers should be able to supply >18 mA per pin with all firing. So for first stab at improving the circuit I added current sinking transistors to the led pins on each DEFAULT_SRM_DOUT_CATHODES<n> shift register: NOTE: this requires inversion of those pins of the shift register in the seq software...

Ok, several have asked about my progress on this, and thanks to Ultra for the image hosting, I will post my current progress on making the duo matrix with these sexy SparkFun silicon buttons: http://www.sparkfun.com/commerce/product_info.php?products_id=7835 The duo matrix is currently used only in MB-SEQ (schematic here: http://www.ucapps.de/mbhp/button_duoled_matrix.pdf) but I see many more possibilities for it... First up the buttons themselves in their just opened at Christmas state: I also ordered some of the SparkFun PCBs thinking I could add SMD diodes but then decided to go another way... Here's some tests with bi-color led lit with 5 vdc and 150 ohm resistor: The hot spots are not quite as prominent as the photo's suggest. They do seem plenty bright, but when powered by scan matrix, they may not be. More to come...

Update: Albs should be shipping these today. They should be here next week...

hehe not quite the first... *whack*

like it says...

*whack*

haven't gigged in some time, but used to a lot. hehe, smashtv and i were talking the other day and discovered most of the deep ellum clubs i played are now gone...

woohoo! now the race to first fully completed MB-6582 begins...

yo, arlington here, welcome. i'm also working on seq now...

the "switches" would interfere with each other... you will still get interference if fretted strings are bent onto the same fret segment (i think, with the way i play, this would happen often)

it's mounted behind the cs pcb...

yet more awesomeness

this is now closed to new orders. only a few have not paid, reminders sent....

no, that's for "gear acquisition syndrome" duh!

coolness!

ok, since almost everyone has paid and i haven't received any new orders for a few days, i will close this order tomorrow, wed. september 26, 2007 those who haven't paid yet, get with it. as soon as all have paid, i will send the order. if you ordered and haven't received a paypal invoice, check your spam folders, let me know, etc. if you ordered and haven't received a confirmation email from me, let me know asap. if you haven't paid by 0ct. 1, i will send order without you... (plus everyone will be pissed cause they had to wait for your sorry ass)

akkk! forget object pascal for pic (or any object oriented language for that matter) in pic land, there's little benefit to objects and nowhere near enough program space... java?! you have to be joking! c is what you want. btw, delphi used to be my favorite for windows programming, but now i'll take visualstudio first. (hmmm look who's running the ms c# team...)

invoices are sent. i will be away for the weekend, so i'll have to handle any issues monday. btw, it's not too late to order, but it will be very soon.