1
0
mirror of git://projects.qi-hardware.com/antorcha.git synced 2024-11-01 13:15:55 +02:00
antorcha/tornado/doc
2012-11-23 12:34:52 -03:00
..
geom.fig tornado/doc/geom.fig: geometry as seen by the acceleration sensor 2012-11-23 12:34:52 -03:00
mech.fpd tornado/doc/mech.fpd: increase length of bottom part of stick from 5 cm to 12 cm 2012-11-23 11:07:19 -03:00
README tornado/doc/README: describe some technical parameters 2012-11-23 11:41:39 -03:00
sclk-1123.png tornado/doc/README: describe some technical parameters 2012-11-23 11:41:39 -03:00

Antorcha vs. Tornado feature comparison
---------------------------------------

Number of LEDs		32			64
Pixels			16			64
LED pitch		8 mm			5 mm
1 arc sec / pixel (1)	28 m			17 m
1 arc sec / text (2)	70 m			90 m
LED line length		25 cm			32 cm
Total stick length	63 cm			53 cm (TBD)
Outer radius		~ 90 cm			~ 80 cm (3)
Visible width		~ 100 cm		~ 110 cm (3)
Sweeps per second	~ 4, bidirectional	~ 3, unidirectional
Sweep speed (avg.)	~ 8 m/s			~ 15 m/s (3)

(1) http://en.wikipedia.org/wiki/Visual_acuity
(2) 2 text lines, 1/12 display height
(3) for chain length = 30 cm


Weight
------

Al stick: 52.5 g / 52.7 cm = ~ 1 g/cm
Chain:	44 g / 19 links = ~ 2.3 g/link
	305 mm / 21 links = 14.5 mm/link
	1.6 g/cm
Attachment (screw, washers, nut): 2.3 g
End cap: 1.54 g
Cable:	1.74 g / 382 mm
	group of 3: ~ 0.14 g/cm
LED PCB: 4.1 g
CPU PCB: 4.5 g

Not included in the weight calculation:
- glue
- adhesive tapes
- isolation tapes
- cable binders (if any)
- heat-shrink tubing
- cables/wires for inter-board connections


Update rate
-----------

For square pixels we need one update every 5 mm. With the edge moving
at 15 m/s, that's one update every 1/3000 s. If we assume that,
including overhead, each 64 bit shift and latch takes about 100 clock
cycles, the serial bus needs to run at no less than 300 kHz.

sclk-1123.png shows the ~250 ns clock pulse generated by the current
test firmware. CH2 (blue) is signal entering the first LED module,
CH1 (yellow) is the signal exiting the last module. The absence of
degradation in a 250 ns pulse suggests that we can go as fast as the
MCU is able to.