AT45DB041B
http://www.atmel.com/dyn/products/product_card.asp?part_id=2469
http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?Ref=106049&Row=304469&Site=US


i will build this h-bridge 4 times:
http://www.cs.uiowa.edu/~jones/step/circuits.html#practbi

maybe use my optos on the way to an atmel, or just the parallel port.

parport:
d0 power
d1 A enable
d2 A dir1
d3 A dir2
d4 B enable
d5 B dir1
d6 B dir2
d7 (reserved0


runs great at about 2ms delay between steps

blade-lowering solenoid is 75ohms.


X range is 3128 over the mat, which is 10.6cm
Y range is more than 2733, which is 9.25cm
cm/step in both directions is .00338
that's 0.00133 in/step


potrace -s -o out6.svg <(convert out6.png ppm:-)


priority path 10    path took 10.7 sec
priority path 7    path took 7.2 sec
priority path 9    path took 9.5 sec
priority path 12    path took 11.0 sec
priority path 2    path took 24.9 sec
priority path 0    path took 18.9 sec
priority path 8    path took 7.6 sec
priority path 16    path took 5.8 sec
priority path 17    path took 7.9 sec
priority path 15    path took 4.0 sec
priority path 25    path took 2.9 sec
priority path 24    path took 7.7 sec
priority path 23    path took 3.9 sec
priority path 22    path took 8.6 sec
priority path 21    path took 4.9 sec
priority path 20    path took 10.6 sec
priority path 38    path took 5.1 sec
priority path 37    path took 6.5 sec
priority path 31    path took 10.8 sec
priority path 33    path took 5.4 sec
priority path 35    path took 4.5 sec
priority path 34    path took 6.3 sec
priority path 32    path took 4.9 sec
priority path 19    path took 17.8 sec
remaining path 1    path took 7.0 sec
remaining path 3    path took 19.6 sec
remaining path 4    path took 8.0 sec
remaining path 5    path took 6.1 sec
remaining path 6    path took 2.9 sec
remaining path 11    path took 3.7 sec
remaining path 13    path took 9.0 sec
remaining path 14    path took 10.9 sec
remaining path 18    path took 8.2 sec
remaining path 26    path took 7.5 sec
remaining path 27    path took 7.9 sec
remaining path 28    path took 9.1 sec
remaining path 29    path took 14.6 sec
remaining path 30    path took 33.6 sec
remaining path 36    path took 7.6 sec
remaining path 39    path took 22.9 sec
remaining path 40    path took 8.6 sec
remaining path 41    path took 14.9 sec
remaining path 42    path took 24.9 sec
remaining path 43    path took 18.4 sec
remaining path 44    path took 5.4 sec
remaining path 45    path took 9.5 sec
remaining path 46    path took 6.0 sec
remaining path 47    path took 9.1 sec
cut all in 482.8 sec


blade 1 cut tests and outs 1..4
blade 2 in at 2006-07-04T23:00:00 

at #15 i started the log playback and relay pauses

after #24, i recentered one nudge to the right

#32, #33 are having slipping paper. add a light coat of glue after 33.

#46 didn't stick well at all (during the hot daytime). new mat.


To run more:

run nudge on dash

run motor_serve as root on bang. currently, it will playback log first
and then go home. to not do that, edit the code so it doesn't play
log. then use svgtrace to load an svg and run curves



notes:

It's surprising how a trashed cut doesn't sound very different from a
working one. I would have thought that I would notice an unusual sound
right away and catch errors quickly, but it often took a lot of
crumple before the different sound would get me to look at the cutter.

The main errors are paper coming off the sticky mat and the blade not
coming up after a cut. When the paper drifts around, I can sometimes
fix it if I catch it quickly, and if the current cut isn't a critical
one. Some cards have a smaller leaf somewhere because I had to restick
the paper during the leaf. Other times, the blade would stay against
the paper because of my lousy reed relay not releasing. If that made a
long cut between shapes, I'd have to scrap the paper.

A light coat of 'removable glue stic' from Avery works really well at
restoring the stick of the mat, and the residue on the next paper or
two is not very serious.

The closest computer to connect the cutter to happened to be my
mail/web server, so it is possible for the motor to stutter if I get a
large spam email, for example. But, the stutter is a timing thing
only-- the cut will be unaffected.

Whenever the repeated cuts started digging deep into the mat, I
staggered the position a little bit.

About 4 good cuts for every bad one, but the bad ones would usually
break in the first few cuts. I deliberately cut the tricky curves
first to avoid running for several minutes and then spoiling a cut.