|
Author: dach on April 01 2007 Viewed 3320 times. 27 people liked this blog. You can rate it below if you haven't already.
People who enjoyed reading this: Zanf, jarvis, ignatius, Otterfan, Roshi, energygiant, daswesen, EsromCole, RRine, SenorFrio, filarion, adcBicycle, em978, jogn, Adjective, crabster, jp, bodo, bla, delete, sAMsKi, celibacyclub, yaxu, bew, flies, mlbot, lysdexic
-->
So, this weekend I got some more of the hardware done. The color RGB leds are working properly now, and each panel on the sequencer lights up around the edges, and the colors slowly fade and mutate. Sadly my camera isnt picking up the shiny, wet crystal look of the plexiglass very well.
I spent some time getting the red and white step leds into place this weekend. The red leds are the clear plastic type, and are panel mounted, sitting just over the buttons. They poke up through the plexiglass and light up with a beautiful clear watery color.
I spent a lot of time on the white leds set into the button itself. These white leds are located inside the sequencer chassis itself, soldered onto a board in there. I drilled a 1mm hole in the top of each led and inserted a short fibre optic cable. This is threaded through a small hole in the button. A lot of care and frustration went into this design because the fibre optics are not particularly bendy, and they need to flex a little as the button goes up and down, hence the 90degree bend. You can see a close up of both the front and the inside of the step buttons in the attached pictures.
The white leds are very bright, and so some light leaked out from the inside through the screws and bots and around the red leds. I've since painted over the white leds with black acrylic to get rid of this.
The last picture shows the insides (upside down). I still have to build and install the motherboard, but I've got it pretty much designed and working on the breadboards already, so hopefully that will not take too long. The observant among you might notice that I've etched the boards manually by drawing out the schematics with a marker direct on the copper, and then mounting the components on the copper side. It's a lot easier to see whats going on. Jumpers are located on the other side of the boards.
| |
Comments
04/01/07
+
PM |
QUOTE |
PERMALINK |
REPORT
jarvis
lookin slick. nice work... you must have incredible patience.
04/01/07
+
PM |
QUOTE |
PERMALINK |
REPORT
sprouts
its looking quite fine
04/01/07
+
PM |
QUOTE |
PERMALINK |
REPORT
ignatius
nice. all these lights.. it's like a circus in a box! looks fun to play.
04/01/07
+
PM |
QUOTE |
PERMALINK |
REPORT
mynameismud
i want to have sex with that
04/01/07
+
PM |
QUOTE |
PERMALINK |
REPORT
EsromCole
I bet Dach does have sex with it. Looks awesome man, very much so.
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
em978
very nice. How's the the implementation of the knobs and buttons to the sequencer going?
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
Squeal
You are em411's nerd of the week! congrats!
But seriously, wow. Fiber-optics? Fucking great.
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
disturb
SEXY
Is it built on the specs of the midi seq @ ucapps ?
Recent blogs: dubstep = poo
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
dach
Nope, it's entirely of my own design. em978: I still havn't done much on the functionality side, im trying to get the hardware completed before I go forward on the programming side. It's too hard to test the code now without having the buttons working properly.
I dont have sex with it, but I do caress it, gently, lovingly. With my tongue.
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
crabster
Looks fabulous. What will you use for the sequencer implementation? Would a PIC suffice?
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
dach
There are four PICs in there to simplify the workload. A 16f670 scans the inputs (36 push switches, some acting as 'shift keys'). That sends keycodes back to the IO module (a pic 16f877 40pins) which also reads the analog pots, the 8 rotary encoders and powers the RGB leds. I've yet to discover if it can do all that at once, but the individual tasks do work  The IO module communicates over an SPI bus to the main brain chip (another 16f877) which transmits/receives MIDI, read/writes to the 32k ram and carries out the sequencing. It in turn sends commands to the output chip (a pic 18f2550 with future USB expansion possibilities) which drives the 3-digit display, the red/white step leds (32) and the other general purpose leds (16).
I think all the IO work could be done with a single pic chip if one had an elaborate serial to parallel conversion thing happening; the number of IO pins required is pretty high.
The main brain chip is totally stressed for working ram, and I'm completely out of IO pins with the address/data bus to the ram chip. It loads up each step a few microseconds in advance from the external ram chip and then does its processing (mostly sending midi). When buttons are pushed I calculate the address required in the external ram and write the appropriate data out. I'd like to redesign that part to support bigger memory, and perhaps have an eeprom in there somewhere too. Currently, saving is done via sysex msgs to a host PC/laptop, and that sucks major balls, but I cant see an easy solution that'll fit in the remaining space/number of IO pins. I'm dithering over whether to just bang it all together and get it working, or leave it on the breadboard for another few months while I work out the memory/saving issues. Difficult to decide since I dont exactly now the memory requirements yet...I could add a eeprom that records sysex data as an external save module later on, if needed... its very very tempting to just say 32k ram will be fine and solder it all together, after working on this for a bloody year I just want it finished!
Future blog entries on this project will be mostly about the software and that's very very unsexy, so enjoy the pretty bling in this post!
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
daswesen
This looksbeautiful. Regarding recording, you could use a i2c flash chip to store the sequencer data, that would be 3 pins to the uC. Or a flash interface which uses SPI as well (IIRC).
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
dach
I tried really hard to get I2C working, gave up after a while and resorted to SPI. I2C would be far better suited considering the fact that I have several chips, and also I have an eeprom thats I2C compatible. I may give it one more try. I dont have the option to add both I2C and SPI, so I need to overhaul the entire communications between chips if I get it working, ush. That'll take weeks. Do you know of an I2C NV-RAM that's at least 32k and ~30ns access time? 128 or up would be lovely.
if I2C works on a quick test, Ill consider working it all out again. You don't happen to knwo where I'd find a working schematic and piece of pic code?
04/02/07
+
PM |
QUOTE |
PERMALINK |
REPORT
sAMsKi
looking really cool man. congrats.
PAWEL is right, they are two different things, of which he can do neither...
07/22/07
+
PM |
QUOTE |
PERMALINK |
REPORT
ignatius
so what's the latest on this thing? you up and running w/it? did i miss the "i'm so happy w/my home built kit" blog?
Register / login
|
^
EM411 is Copyright 2001-2008 EM411.com
All rights reserved. | Contact | RSS
|
